May 3, 2012 - Enterprise Architecture Programs . ...... particular, on federal agencies' EA programs and their assimilation. ...... University of California Press.
A MULTI-LEVEL INVESTIGATION INTO THE ANTECEDENTS OF ENTERPRISE ARCHITECTURE (EA) ASSIMILATION IN THE U.S. FEDERAL GOVERNMENT: A LONGITUDINAL MIXED METHODS RESEARCH STUDY
by GEORGE K. MAKIYA Submitted in partial fulfillment of the requirements For the degree of Doctor of Philosophy
Dissertation Committee: Kalle Lyytinen, Ph.D., Case Western Reserve University (chair) Richard Boland, Ph.D., Case Western Reserve University Bo Carlsson, Ph.D., Case Western Reserve University Jeanne Ross, Ph.D., Massachusetts Institute of Technology
Weatherhead School of Management CASE WESTERN RESERVE UNIVERSITY
August, 2012
CASE WESTERN RESERVE UNIVERSITY SCHOOL OF GRADUATE STUDIES
We hereby approve the thesis/dissertation of George K. Makiya candidate for the
(signed)
Doctor of Philosophy
degree*.
Kalle Lyytinen (chair of the committee) Bo Carlsson Richard Boland Jeanne Ross
(date)
May 3, 2012
*We also certify that written approval has been obtained for any proprietary material contained therein.
ii
© 2012 George K. Makiya All Rights Reserved
DEDICATION This dissertation is dedicated to my loving daughter Taylor, who is my life and legacy.
iv
TABLE OF CONTENTS Dedication .......................................................................................................................... iv List of Tables .................................................................................................................... xii List of Figures .................................................................................................................. xiv Acknowledgements ............................................................................................................xv Abstract ........................................................................................................................... xvii CHAPTER I: INTRODUCTION........................................................................................1 Purpose of the Study ....................................................................................................... 4 Significance of the Subject .............................................................................................. 6 Dissertation Setting ......................................................................................................... 8 A Review of the Literature ............................................................................................ 12 EA Overview ............................................................................................................. 12 Enterprise Architecture Programs ............................................................................. 15 EA Assimilation within Adopter Units ..................................................................... 18 The EA Assimilation Framework (EAMMF) ........................................................... 19 EA Adoption and Assimilation.................................................................................. 22 Institutional Theory and Innovation Assimilation ........................................................ 25 Coercive Pressure ...................................................................................................... 26 Three Levels of Enterprise Architecture Assimilation .................................................. 27 What We Know and Do Not Know of IT Assimilation ................................................ 30 What We Know ......................................................................................................... 30 What We Do Not Know ............................................................................................ 33 Research Motivation ..................................................................................................... 35 Research Scope ............................................................................................................. 35 Section Multilevel Research Study ............................................................................... 35 Choice of a Research Design ..................................................................................... 35 Structure of the Research Design .............................................................................. 39 Data Sampling ........................................................................................................... 40 v
Research Highlights ...................................................................................................... 42 Study Dependent Variables ....................................................................................... 43 Major Findings .............................................................................................................. 45 Study 1: The Antecedents of EA Program Assimilation............................................... 45 Linking Study 1 Findings to Study 2 & 3 Constructs ................................................... 47 Study 1 findings .................................................................................................................47 Corresponding Study 2 & 3 construct ................................................................................47 Definition of corresponding study 2 & 3 construct ...........................................................47 Study 2: The Antecedents of EA Assimilation in EA Adopter Units ........................... 48 Study 3: Antecedents of EA Assimilation Stages and Phases ...................................... 48 Limitations .................................................................................................................... 49 Summary of Key Issues................................................................................................. 50 Future Research ............................................................................................................. 51 General Summary.......................................................................................................... 51 CHAPTER II: INVESTIGATING THE FACTORS THAT INFLUENCE EA PROGRAM ASSIMILATION – QUALITATIVE STUDY of U.S. FEDERAL AGENCY EA PROGRAMS ...............................................................................................................53 Introduction ................................................................................................................... 53 Literature Review .......................................................................................................... 54 Organizational Characteristics ...................................................................................... 55 Organization Culture ................................................................................................. 55 Organizational Structure ............................................................................................ 56 Organization Complexity .......................................................................................... 57 Methods ......................................................................................................................... 58 Methodology.............................................................................................................. 58 Sample ....................................................................................................................... 59 Data Collection .......................................................................................................... 60 Data Analysis ............................................................................................................. 61 Findings ......................................................................................................................... 62 vi
Finding 1: Agency culture, size, structure and complexity are inhibiters of E.A. assimilation progress ................................................................................................. 62 Finding 2: Framing and labeling of EA as an IT function instead of a strategic business tool reduces its perceived value to the organization, inhibiting program assimilation ................................................................................................................ 63 Finding 3: A high resistance to change and low EA assimilation progress was reported by program leaders in agencies with strong political constituencies, while program leaders in agencies with less politically strong constituents reported less resistance to change ................................................................................................... 65 Finding 4: Program leader collaborative style and use of creative and aggressive tactics, business knowledge and marketing communications can moderate the effects of agency/department size, complexity and culture on EA program assimilation ................................................................................................................ 66 Discussion ..................................................................................................................... 69 Conclusion..................................................................................................................... 76 Implications for Practice and Future Research ............................................................. 77 Limitations of the First Study ....................................................................................... 78 Chapter IV: INVESTIGATING THE Antecedents of Enterprise Architecture assimilation AT ADOPTER UNIT LEVEL: CROSS-SECTIONAL QUANTITATIVE STUDY OF U.S. FEDERAL AGENCIES ........................................................................79 Theoretical Framework ................................................................................................. 83 Enterprise Architecture ................................................................................................. 83 Enterprise Architecture in Organizations .................................................................. 83 EA Assimilation ........................................................................................................ 85 MIT Assimilation Model ........................................................................................... 86 The GAO Assimilation Framework .......................................................................... 88 Comparing MIT and GAO Frameworks ................................................................... 92 Institutional Theory and Innovation Assimilation ..................................................... 95 Research Model and Hypotheses .................................................................................. 96 Overview of the Research Model .............................................................................. 96 Access to Resources .................................................................................................. 98 Parochialism and Cultural Resistance ..................................................................... 101 Top Management Value Recognition ...................................................................... 102 vii
Organization Complexity ........................................................................................ 103 Organization Scope.................................................................................................. 105 Moderating Effect of Access to Resources.............................................................. 108 Moderating Effects of Coercive Pressure ................................................................ 110 Research Design and Data Collection ......................................................................... 114 Data Sample and Scales ........................................................................................... 114 Construct Operationalization and Scales ................................................................. 115 Construct Definition and Operationalization ........................................................... 118 Dependent Variable ................................................................................................. 118 Independent Variables ............................................................................................. 119 Controls ................................................................................................................... 123 Statistical Analysis Method ..................................................................................... 123 Measurement Model ................................................................................................ 125 Findings ....................................................................................................................... 130 Direct Effects (H1a-H1e)......................................................................................... 131 Results H1a – H1e ................................................................................................... 131 Results for Moderation Hypotheses H2a – H2d ...................................................... 131 Test of Hypotheses H2a – H2d ................................................................................ 133 Results for Moderation Hypotheses H3a – H3e ...................................................... 134 PostHoc Analysis ..................................................................................................... 135 Discussion and Conclusion ......................................................................................... 136 Implications for Practitioners and Managers .............................................................. 140 Limitations .................................................................................................................. 141 Future Research ........................................................................................................... 142 CHAPTER IV: INVESTIGATING THE DETERMINANTS OF EA ASSIMILATION ACROSS ASSIMILATION LEVELS AND PHASES: A QUANTITATIVE LONGITUDINAL STUDY .............................................................................................143 Introduction ................................................................................................................. 143 Theoretical Foundation ............................................................................................... 148 Configuration Analysis Typology ........................................................................... 149 viii
Theoretical Perspectives on IS Assimilation ........................................................... 150 Assimilation Phases ................................................................................................. 151 EA Assimilation Stages ........................................................................................... 155 Theoretical Perspectives on Environmental Jolts .................................................... 157 Research Model and Hypotheses ................................................................................ 158 RQ I: The Change of Determinants of EA during Distinct Assimilation Phases ... 159 Access to Resources ................................................................................................ 162 Parochialism and Cultural Resistance ..................................................................... 163 Top Management Value Recognition ...................................................................... 164 Organizational Complexity...................................................................................... 165 Organization Scope.................................................................................................. 166 RQ 2: The Impact of Coercive Pressure .................................................................. 166 RQ 3: What are the Determinants of Movement Between Assimilation Stages? ... 170 Overview of the Research Model ............................................................................ 171 Hypotheses H3a-3e .................................................................................................. 172 Research Design and Data Collection ......................................................................... 175 Research Context: Federal EA Programs ................................................................ 175 Research Methodology ............................................................................................ 176 RQ 1: The Determinants of Assimilation Phases .................................................... 176 RQ 2: The Impact of Coercive Pressure on Determinants of EA Assimilation ...... 177 RQ 3: The Determinants of EA Assimilation Stages .............................................. 177 Data Sample and Scales ........................................................................................... 177 Operationalization ................................................................................................... 177 Dependent Variable: EA Assimilation ..................................................................... 177 Statistical Analysis .................................................................................................. 179 Findings ....................................................................................................................... 180 Results for hypotheses H1a – H1e ........................................................................... 181 Results for Hypotheses H3a – H3e .......................................................................... 184 PostHoc Analysis ..................................................................................................... 185 Discussion and Conclusion ......................................................................................... 185 ix
Limitations .................................................................................................................. 189 Future Research ........................................................................................................... 189 CHAPTER V: DISSERTATION CONTRIBUTIONS ...................................................191 Study 1 Contribution ................................................................................................... 191 Study 2 Contribution ................................................................................................... 192 Study 3 Contribution ................................................................................................... 193 Implications from the Three Studies for Managers and Practitioners ......................... 195 Contributions to EA Assimilation Theory .................................................................. 196 Future Research ........................................................................................................... 198 Appendix A: Interview Protocol .....................................................................................199 Appendix B: Comparative Analysis of Tempered Radical Characteristics ....................200 Appendix C: Comparative Analysis of Tempered Radical Tactics ................................201 Appendix D: Construct Definitions ................................................................................202 Appendix E: GAO Assimilation Model Core Elements .................................................204 Appendix F: OMB and GAO Evaluation Criteria ..........................................................205 Appendix G: OMB Assessment Ranking .......................................................................206 Appendix H: Covariance Table ......................................................................................207 Appendix I: Correlation Matrix ......................................................................................208 Appendix J: Hypothesis Testing Results ........................................................................209 Appendix K: PLS Models and Interactions ....................................................................210 Appendix L: Multicollinearity, Skewness and Kurtosis .................................................220 Appendix M: GAO Survey instrument ...........................................................................221 Appendix N: Sample GAO EAMMF Scoring Charts.....................................................222 Appendix O: OMB EAAF Service domains ...................................................................224 Appendix P: List of Agency EA Programs .....................................................................225 Appendix Q: OLS Regression Results............................................................................228 x
Appendix R: Model Fit Summary...................................................................................233 Appendix S: Multinomial Logistic Regressions .............................................................234 Appendix T: Factor Loadings .........................................................................................240 Appendix U: Results for Hypothesis # 4: EA Assimilation Enhances Strategic Business Outcomes .........................................................................................................................245 Appendix V: OLS Correlation Matrix ............................................................................248 Appendix W: Moderation Testing Results .....................................................................250 Appendix X: Interactions ................................................................................................255 References ........................................................................................................................257
xi
LIST OF TABLES TABLE 1: Federal EA Sequence of Events ....................................................................... 8 TABLE 2: Attributes of EA as a Management Program ................................................. 17 TABLE 3: What We Know.............................................................................................. 32 TABLE 4: Research Methods & Techniques .................................................................. 39 TABLE 5: Sampling Techniques Used In Study ............................................................. 41 TABLE 6: High Level Description of Studies ................................................................. 43 TABLE 7: Dependent Variable Definitions and Operationalization ............................... 43 TABLE 8: Summary of Key Findings ............................................................................. 45 TABLE 9: Link between Study 1 Findings and Studies 2 & 3 ........................................ 47 TABLE 10: Summary of Key Issues ............................................................................... 50 TABLE 11: Learning Requirements of EA Levels.......................................................... 87 TABLE 12: Comparative Analysis of MIT Assimilation Framework and the GAO EAMMF ............................................................................................................................ 93 TABLE 13: The Study Constructs ................................................................................... 98 TABLE 14: Data Details and Sources ........................................................................... 114 TABLE 15: Scales and Measures Overview ................................................................. 116 TABLE 16: Measurement Model .................................................................................. 127 TABLE 17: Hypothesis H1a-e Results ......................................................................... 131 TABLE 18: Moderation testing results for hypotheses H2a-H2d ................................. 133 TABLE 19: Moderation testing results H3a-H3e .......................................................... 134 TABLE 20: EA Assimilation Stages ............................................................................. 156 TABLE 21: Study Constructs and Definitions .............................................................. 162 TABLE 22: Construct Operationalization ..................................................................... 178 TABLE 23: Model Fit Statistics .................................................................................... 180 xii
TABLE 24: Hypotheses H1a-e Results ......................................................................... 181 TABLE 25: Results for Hypotheses H2a-c .................................................................... 182 TABLE 26: Study 3 Multinomial logistic regression results for hypotheses H3a-H3c 183
xiii
LIST OF FIGURES FIGURE 1: Dissertation Triangulation Visual Design Model ......................................... 36 FIGURE 2: Agency Characteristics ................................................................................. 63 FIGURE 3: EA Framing and Labeling ............................................................................ 64 FIGURE 4: Agency Culture and Structure ...................................................................... 65 FIGURE 5: Reward Schemes .......................................................................................... 67 FIGURE 6: Examples of Guerilla Tactics ....................................................................... 68 FIGURE 7: Language Use ............................................................................................... 69 FIGURE 8: Marketing and Information Dissemination .................................................. 69 FIGURE 9: EA Leader Tempered Radical Strategy ........................................................ 73 FIGURE 10: Revised Conceptual Model ........................................................................ 76 FIGURE 11: GAO Assimilation Framework Elements and Attributes ........................... 91 FIGURE 12: Research Conceptual Model ...................................................................... 97 FIGURE 13: Regression Results with Access to Resources Moderation ...................... 130 FIGURE 14: Regression Results with Coercive Pressure Moderation .......................... 133 FIGURE 15: Research Conceptual Model 1 .................................................................. 161 FIGURE 16: Research Model 2 ..................................................................................... 167 FIGURE 17: Research Conceptual Model 3 .................................................................. 172
xiv
ACKNOWLEDGEMENTS My profound gratitude goes out to my principal supervisor, Prof. Kalle Lyytinen for his invaluable support and guidance throughout my academic journey at Weatherhead. Kalle not only served as role model for the highest standards of scholarship, his untiring efforts sharpened my focus and helped position my research. I consider myself quite fortunate to have had a committee that comprised of academic luminaries and mentors including Jeanne Ross, Dick Boland and Bo Carlsson. This thesis is the culmination of an incredible support network of the Weatherhead School of Management’s academic and administrative staff. I would like to thank in particular Sue Nartker, Marilyn Chorman and Karen Oye, who provided invaluable material and moral support as I labored through the maze that is doctoral level academic bureaucracy. While this dissertation serves as an important quest for personal and professional achievement, it is equally the product of support and guidance from several people I’d like to acknowledge. Words can hardly express my gratitude to Dave McCue, the VP and CIO of Computer Sciences Corporation, my employer, whose support facilitated my travel to the Cleveland Ohio campus for classes every third week for the four years that I pursued the PhD, in the midst of high demanding job assignments. I would also like to thank the Federal EA community who embraced this research and provided vital support. Notably I’d like to thank Dr. Scott Bernard, the U.S. Federal Chief Architect, who not only taught me at Syracuse University but maintained an important professional presence that helped shape and guide my understanding of federal Enterprise Architecture. I’d like to express sincere gratitude to Richard Burk, former U.S. Federal Chief Architect, Randolph Hite, the former Director for IT at the U.S. Government Accountability Office xv
(GAO); Peter Orszag, the former Director of the U.S. Office of Management & Budget (OMB), Vivek Kundra, the former Federal CIO; Michael Tiemann of Federal Enterprise Architecture Institute (FEAC); Michael Holland of the GAO; Martha Johnson, Director of the U.S. General Services Administration (GSA), Robert Carey, CIO, U.S. Navy, as well as several other federal EA community members that preferred to remain anonymous. I would also like to thank former Senator/Secretary of Defense William Cohen, Sen. Joseph Lieberman as well as Congressmen Tom Davis and Dan Burton whose tremendous logistical support helped secure vital material for the dissertation research from the U.S. National Archives. I’m also grateful to all my classmates and cohort partners to whom I owe tremendous debt for sharing their insight, knowledge and friendship that made this academic journey bearable. My sincere gratitude goes to Ronald Eastburn, Amol Kharabe, Ruth Bernstein, Moraima De Hoyos, Renson Muchiri, Glen Weaver, Kathleen Buse, Mimi Lord and Alice Walker. Finally I’d like to thank family members and friends who provided much appreciated moral and emotional support including Desha, Nelly and Florence, Abasi, and Argwings. My sincere apologies to my daughter Taylor, who had to endure divided attention her entire childhood as I pursued this doctorate
xvi
A Multi-Level Investigation into the Antecedents of Enterprise Architecture (EA) Assimilation in the U.S. Federal Government: A Longitudinal Mixed Methods Research Study
Abstract by
GEORGE K. MAKIYA
This dissertation reports on a multi-dimensional longitudinal investigation of the factors that influence Enterprise Architecture (EA) diffusion and assimilation within the U.S. federal government. The study uses publicly available datasets of 123 U.S. federal departments and agencies, as well as interview data among CIOs and EA managers within select Federal Government agencies to conduct three multi-method research studies: 1) a qualitative study to investigate organizational and institutional factors that enhance or impede EA assimilation at program level; 2) a quantitative study to examine the antecedents of EA assimilation at adopter unit level and 3) a longitudinal quantitative study to examine: 1) the antecedents of EA assimilation within adopter populations as marked by prominence within each of the EA assimilation phases 2) the influence of sudden changes in environmental (institutional) context on the EA assimilation process; and 3) the determinants for each EA assimilation stage. I use time-lagged partial least square, ordinary least square and multinominal logistic regression to analyze these xvii
effects. The study shows that an innovative leadership style is the key to advancing EA program assimilation within adopter units. Framing and labeling of an EA program as an administrative driven innovation or reform as opposed to a business essential strategic tool greatly influences its value perception, adoption and assimilation. Institutional coercive pressure is not a long term sustainable strategy in driving EA assimilation, though it has a “jolt” like short term effect in accelerating assimilation. EA assimilation has distinct micro and macro level antecedents. Factors also have "differentlydirectioned effects," that is factors that promote EA progress at certain assimilation phases and stages inhibit progress at other phases and stages. Changes in the temporal environmental context have “factor elasticity” effect on the explanatory power of the antecedents. That is, antecedents lose and regain their explanatory power commensurate with changes in the environment over time. Overall, the study’s findings have several major implications for policymakers: 1) complex administrative innovations such as EA require strategic frameworks as opposed to blueprints to 1) overcome dynamic complexity and 2) drive multi-level assimilation; 2) EA assimilation at each of the levels have different sets of definitions and antecedents 3) each of the levels have different properties and characteristics and require different approaches and strategies, 4) institutional coercive pressure is only effective when applied as a temporal strategy, 5) individual EA program and adopter unit assimilation are interdependent. That is, successful assimilation of EA within the organization is highly dependent upon the degree of embracement and legitimization of individual EA programs.
xviii
Key words: enterprise architecture; diffusion of innovation; assimilation of innovation; coercive pressure qualitative research; quantitative research
xix
CHAPTER I: INTRODUCTION Over the past twenty five years, organizations have invested enormous resources in the acquisition and implementation of information technology (IT) innovations in pursuit of competitive advantage and business value (Swanson & Ramiller, 2004). Over the same period of time, the idea of developing general plans for the deployment of IT resources in line with the strategic direction of the firm - often expressed by the moniker Enterprise Architecture (EA) - has emerged as one mechanism to use IT innovatively for business value. As a result EA has witnessed exponential global growth in popularity as a highly complex, albeit value enhancing administrative innovation. While some companies have achieved significant benefits through the deployment EA, others have failed miserably in their implementations, or experienced serious budget overruns coupled with disappointing performance results (Kappelman, 2009). Available evidence suggests that the potential value of such complex innovations can only be unlocked through persistent and successful assimilation within the organization (Armstrong & Sambamurthy, 1999). Assimilation can be defined as the extent to which the new innovation is utilized and routinized or becomes part of the operational fabric of the organizational processes (Purvis, Sambamurthy, & Zmud, 2001). This thesis incorporates both this definition and the one by Ash (1997) as the “depth of integration and its pervasiveness” within the organization to study EA assimilation levels and phases – changes of assimilation environment over time - as well as their antecedents and effects. EA assimilation is viewed in this dissertation as a complex and a costly undertaking, regardless of its motivation of adoption – i.e. whether voluntarily, or mandated. For instance, the U.S. government promulgated EA via a 1996 1
congressional act, and has since invested – with dismal results, approximately $550b in the implementation and assimilation of EA within the various agencies (Burk, 2005. Several taxpayer watchdogs argue that the dollar amounts are actually substantially higher. Failures associated with assimilation of IT innovations within the private sector have also been highly publicized and many have had financially ruinous consequences (Bajwa, Garcia, & Mooney, 2004; Miller, 2000; Xue, Liang, Boulton, & Snyder, 2005). Over the past two decades, Information System (IS) practitioners and researchers have shown growing interest in organizational and institutional factors that influence IT innovation assimilation within organizations. Our collective understanding of the antecedents of succesful EA assimilation still remains poor, despite having identified, for well over a decade that progressive assimilation is a key critical success factor for IT effectiveness. Many fundamental questions regarding the factors that inhibit or enhance EA assimilation remain unanswered. Relentless pursuit of answers to these questions would substantially increase the cumulative understanding of antecedents and effects of complex IT innovations. It would also immensely benefit the IT profession, IS scholars and the IT industry at large. As an indication of the growing attention that EA continues to draw globally, an extensive search by the author through databases of authoritative global research firms such as Gartner, Forrester and IDC reveals that for the period ranging from 2000-2011, there were over 410 conferences, 194 workshops and seminars, 169 top journal published research papers and 19,201 active blogs focused on EA globally1
1
Lexis Nexis searches (2000-2011) on EA conferences; SIMEAWG; Gartner Research reports 2000-2011; Forrester; IDC (2000-2011); FEAC; ACT-IAC statistics
2
In spite of the growing interest within the practitioner and scholar communities, existing intellectual perspectives on IT innovation assimilation remain inadequate to explain the dynamics of complex administrative innovations such as EA. This is because EA is not only a highly complex and dynamic phenomenon, but it is also multidimensional and multi-level - inviting multidisciplinary research. Extant research shows that adoptions of complex and disruptive administrative innovations are typically foreshadowed by institutional intervention (Bhalla & James, 1988; Kaul, 1987; Kraemer et al., 1992). However, most of the IT innovation research adopts psychological, sociological or economic perspectives to account for success in IT innovation diffusion (e.g., Fichman & Kemerer, 1999; Lyytinen & Robey, 1999; Zhu et al., 2006). Therefore, in this thesis I will investigate the antecedents of EA assimilation by adopting a multi-disciplinary as well as multi-level approach. Accordingly, I frame the dissertation questions around three units of analysis: 1) a program, 2) an adopter unit and 3) an adopter population. These are all examined simultaneously to surface the effects of powerful institutional measures that are expected to influence and shape EA assimilation. To this end this thesis seeks to converge several intellectual traditions from the sociological studies of IS innovation assimilation and institutional theory (e.g., DiMaggio & Powell, 1983; Fichman & Kemerer, 1999; Tolbert & Zucker, 1983; Zhu et al., 2006). By doing so it pursues a synthesis of perspectives in order to provide fresh insights into the antecedents of EA assimilation. This dissertation is organized as follows. This opening chapter provides a detailed background to the study; motivates the research and defines EA both as an administrative innovation and a multi-faceted management program. The next section conducts an in3
depth literature review of EA as well as the theoretical foundations of the study. The following section discusses and rationalizes the research methodology and the contributions that this study makes to the intellectual discourses around complex IT innovation. I also note the study’s limitations. The next chapters provide the crux of the thesis by reporting three studies, one qualitative and two quantitative ones. The thesis concludes with a discussion of the importance of the study in terms of its potential contribution to our understanding of EA assimilation drivers and effects, and also more broadly to IS assimilation literature. I finish the thesis with recommendations for additional research in this area. Purpose of the Study The purpose of this study is to provide new empirical and intellectual insights into the antecedents of EA assimilation. Both IS research and EA practice alike would greatly benefit from an empirical investigation of the antecedents of EA assimilation – especially within the study’s settings for two reasons. One is that the U.S. government EA implementation remains the largest in the world (Bernard, 2005). It is also much touted as the most complex due to its advanced governance structure and huge federal budget (Burk, 2005). The second is that there are no empirical studies within either the private or public sector on the antecedents of EA assimilation. Much of the existing literature on EA assimilation comprises case studies that provide anecdotal evidence (e.g. Ross, Weill, & Robertson, 2006; Schekkerman, 2006). The dearth of empirical studies in IS assimilation is fueled by several factors. One is that reliable assimilation studies require longitudinal data, which are neither easily available nor accessible. The second is that most IS implementations are conducted in singular and often dissimilar organizations 4
rendering it difficult for researchers to obtain multi-organizational or industry-wide data for comparative analysis. The third is restrictions on information sharing for legal (such as HIPPA, FOIA), commercial or intellectual property protection reasons. There are two major theoretical shortfalls in research that this dissertation hopes to address. The first is that current IS research has often failed to separate the literature streams to reflect program level, adopter unit level and adopter population level, leaving a void in our understanding of the factors that influence each level (Lyytinen & Damsgaard, 2011). The second is the prevalent (and erroneous) reified notion that innovation assimilation within government is not only linear, but also ably orchestrated by the omnipotence of institutional pressure. This study defines an institution as any standing, social entity that exerts influence and regulation over other social entities as a persistent feature of social life, outlasting the social entities it influences and regulates, and surviving upheaval in the social order (Hughes, 1939). The primary research questions addressed by this thesis are: 1. What are the antecedents of EA program assimilation? 2. What are the antecedents of EA assimilation within individual adopter units? 3. What are the antecedents of EA assimilation within adopter populations? 4. What is the effect of changes in the environmental context on the antecedents of EA assimilation? To answer these questions, the research devised three separate, but related studies. The aims, methods and results of each study will be discussed in detail in a later section of this introduction.
5
Significance of the Subject Every year, analogous to an annual ritual, the Government Accountability Office (GAO) and other consumer watchdog reports spark U.S. taxpayer outrage by detailing massive losses - in billions of dollars from failed Federal Agency IT projects. For instance, the 2008 GAO reports showed that 585 of the 810 major IT projects (about 72 percent) faced imminent failure (Grasso, 2009). Failure in this case means missed deadlines, massive budget overruns and high project scope variance. The GAO and several federal agencies in 2009 identified 413 out of 585 (70%) projects totaling $25.2b (or 36% of the annual Federal IT budget) as being failures (GAO-08-105IT). The report categorized 352 projects totaling over $23.4b as critical failures and the remaining 87 projects totaling $4.8b as high risk. The 2013 U.S. Federal IT budget is $79 billion, and has steadily increased at a rate of 7% per year since the early 2000s (FederalComputerweek, 2012). Some taxpayer watchdogs (e.g. taxpayers for commonsense) estimate that since 1996, the U.S. Government has spent over $900 billion on IT projects. EA assimilation as a major IT initiative is a highly complex undertaking and organizations can expect to encounter numerous difficulties in building requisite IT capabilities. Weill & Ross, (2004) write that the process of developing enterprise architecture is not orderly as often assumed. For example, close to two decades since promulgation, less than 50% of the federal agencies have attained high EA assimilation levels. Numerous Government Accountability Office (GAO) EA audits provide ample evidence that federal agencies have, since the passage of the act, made only woeful progress in assimilating EA. For instance, in the 2006 GAO audit, conducted ten years 6
later after EA promulgation, found that 48% of federal agencies were still at early levels of EA assimilation, 36% at medium level, 11% at higher levels, while only 5% were at the highest assimilation level (Rico, 2006). The 2001 audit, involving 93 federal agencies showed that 100% of the agencies were below the highest assimilation levels, with 83% languishing at the lowest assimilation levels. The 2003 audit, involving 96 agencies showed that 99% were below the highest assimilation levels, and that 90% were still at the lowest assimilation levels (Beamer, Henning, & Cullen, 2004). The intense public attention fueled in part by the large amounts of taxpayer money involved, the high profile policy failures and numerous government interventions and executive actions make this study very significant. In addition, a large number of countries around the world are not only adopting and assimilating EA, they are equally promulgating highly complex IS innovations such as electronic governance and electronic healthcare delivery (e.g. Bolgherini, 2007; Wong, 2001; Moon & Bretschneider, 2002; Prahalad, 2010). Judging by the intense curiosity this study has generated within the wider EA community and the enthusiastic and generous support from the U.S. government officials, I have strong conviction that this study’s findings are of great interest and importance to the IS practitioner and research community. It is also a clear testimony of the glaring intellectual vacuum into the factors that influence EA assimilation.
7
Dissertation Setting The United States government as of now has the single largest and best known wide-scale implementation of EA in the world (Bernard, 2005). For this reason I have chosen to situate my research study within the federal government domain, focusing in particular, on federal agencies’ EA programs and their assimilation. The remainder of this subsection provides background about those programs and associated policies. In 1996, the U.S. Congress passed the Clinger-Cohen Act, which requires all federal agencies to establish an EA program, develop and use EA for IT investment plans and decision making (Bellman & Rausch, 2004) The Table 1 below shows the sequence of events leading up to and beyond passage of the Clinger-Cohen Act. TABLE 1: Federal EA Sequence of Events Date October 12, 1994
July 25, 1995
February 1996
September 1999
April 2000 2001
Description Investigative report of Sen. William S. Cohen of Maine presented to Congressional subcommittee on oversight of government management and senate governmental affairs committee Title: Computer Chaos: Billions wasted buying federal computer systems Report catalogs wasteful spending and calls for reining in of computer systems procurement GAO Report to Congress by Gene L. Dodaro Title: Legislation Would Strengthen Federal Management of Information and Technology. Recommends legislation to reform Government IT This report calls for Congressional intervention in the management of Government IT. Lays groundwork for an Act of Congress to mandate reform of Government IT The "Information Technology Management Reform Act' is passed by Congress It is later renamed "Clinger-Cohen Act" after its co-sponsors, Rep. William Clinger, R-Pa., and Senator William Cohen, R-Maine. The Act: 1) mandates EA as an IT capital planning and investment management tool and 2) establishes office of CIO in every department and agency within the federal government. The Chief Information Officers Council releases “Federal Enterprise Architecture Framework Version 1.1.” aimed at providing guidance on implementation of Enterprise Architecture. It also becomes the official maturity assessment tool for EA programs The Office of Management and Budget distributes the "OMB Circular A-130” which finally enforces the Clinger-Cohen Act mandate The Federal CIO Council releases the “Practical Guide to Federal Enterprise Architecture” to provide guidance on how to create, manage, and use enterprise architectures
8
2001
February 19, 2002
March 21, 2002
December 17, 2002
2004 August 14, 2006
Federal Enterprise Architecture Program Management Office (FEAPMO) is formed in the OMB to replace ITA (IT Architecture) and provides a “businesslike blueprint” to transform government into a “results-oriented and market-based approach to IT resource management. Creates “Reference models” to standardize and streamline EA within federal agencies GAO releasers report: GAO-02-6 – “Enterprise Architecture Use Across the Federal Government Can Be Improved.” This report shows dismal EA maturity results following a government-wide audit GAO releases report: GAO-02-389T – “OMB Leadership Critical to Making Needed Enterprise Architecture and E-government Progress” This report opens the way for OMB to get directly involved with agency EA implementation Congress passes the “E-Government Act of 2002”, to “to improve the management and promotion of electronic government services and processes”, creating office of Federal Chief Information Officer and Federal Chief Enterprise Architect. It also introduces “agency program performance goals” and “transparency and accountability” for agency executives, paving way for punitive measures from OMB. Act gives OMB powers to act on non-compliant agencies OMB rolls out the EA Assessment Framework (EAAF) which micro-manages federal programs, including EA programs GAO report: GAO-06-831- Leadership Remains Key to Establishing and Leveraging Architectures for Organizational Transformation” and to “transform agency operations by utilizing, where appropriate, best practices from public and private sector organizations”
In 1999, the Federal CIO council published the Federal Enterprise Architecture Framework (FEAF) to establish guidelines and formulate approaches to implementing EA and capital planning processes in federal agencies (Bernard, 2005). The framework was business-based and intended for government-wide improvement. It was equally envisioned to facilitate efforts in transforming the U.S. Federal government to one that is citizen-centered, results-oriented, and market-based (CIO Council, 1999). The primary objective of the FEAF was to organize federal information on a government-wide scale, promote information sharing among federal organizations, facilitate federal organizations develop their respective architectures, facilitate federal organizations develop their IT investment processes, and serve customer needs better, faster and cost effectively (Saha, 2004.
9
In July 2002 the U.S. Congress passed the E-Government Act (H.R. 2458) commonly referred to as “e-Gov”, that promulgated a comprehensive framework to expand the use of the Internet and computer resources in order to deliver Government services for a citizen-centered, results-oriented, and market-based Government (Bellman & Rausch, 2004). The Act also extended and strengthened the 1996 Clinger-Cohen Act by 1) inclusion of sanctions for non-compliant agencies including budget cuts and other punitive measures and 2) granting oversight authority to the Office of Management and Budget (OMB), which belongs to the executive branch of Government as opposed to the GAO, which belongs to the legislative branch of government. Unlike the GAO, which had putative powers, the OMB wielded coercive and punitive powers. For instance the OMB budget process required that agencies indicate compliance to the mandate on the yearly budget request form (Exhibit 300B) in order to receive their appropriations. Colorcoded OMB assessment results including an intimidating “red-rating” for non-compliant agencies were posted online for public viewing. Non- compliant agency heads faced the omnipotent threat of public Congressional summonses and hearings (often televised), an unpleasant prospect for political appointees. Even though the U.S. Congress failed to extend the e-Government Act in 2006, some of its provisions such as the Program Appraisal Rating Tool (PART) remained in force. PART, which was launched by executive order to streamline government program portfolio oversight, still required agencies to report on a quarterly basis, performance of their respective programs portfolio - including EA programs to the OMB. My dataset covers the period from 1999-2007, which spans both the pre-e-Gov period as well as the post e-Gov period. The GAO has conducted, at the direction of Congress, four government-wide EA program assessments 10
– in 1999, 2001, 2003 and 2006. These assessments constitute the reference points used by this dissertation as proxies for 1) the assimilation time lags and 2) the environmental changes. The measures and scales remained the same throughout the four audits. A sample of the EA program audit instrument used is shown in Appendix M. In this dissertation, I used the GAO EA program maturity framework ranking as a proxy for the agency (or adopter unit) EA assimilation level. The GAO EA Framework scale is computed as an index from several indicators reflecting the extent to which different aspects of EA principles have been integrated into the agency’s operations. The GAO assimilation framework, which is mapped to this scale, is a self-explanatory scorecard used for the assessment of the EA assimilation process. The scorecard uses a combination of two complementary methods used to determine an organization’s assimilation level. The first method is focused on the weighted mean assimilation level, whereas the second method is focused on the percentage achieved at each assimilation level for the thirty one EA core elements. This measure was derived from the EA program audit data as contained in the GAO assessment reports (GAO-04-40; GAO-04798T; GAO-06-831). The GAO EA assimilation measure uses a five point scale captured in a Guttmann scale (Suchman & Presser, 1981) and it shows the consecutive levels at which an EA program has been formalized and integrated into the business operations of an agency (Bernard, 2005). As noted the framework has 5 assimilation levels, 4 critical success attributes and 31 core elements that help link the attributes to the assimilation levels. Appendix E enlists the core elements that were used for the scoring and determination of the assimilation level. The next section conducts an in-depth review of the EA literature and particularly how the concept has evolved over the last two decades. 11
A Review of the Literature In this section, I conduct an in-depth literature review of EA. I first begin by providing an overview of the concept and motivation for EA, followed by its definition and discussion of various ways to structure EA innovation programs. I then introduce and discuss EA program assimilation frameworks- ways of determining the level of assimilation. I also review in detail, the GAO EA framework used for assessment of EA programs within the U.S. federal government EA Overview As a fairly young business discipline, EA continues to attract varied definitions which have continued to evolve over the years. In this paper I do not engage in the theoretical debate over the proper definition of EA as it is beyond the scope of this study. Indeed, the world has yet to settle for example on precise definitions of architecture or architecture description and how these terms relate to enterprises, systems, or software (Schekkerman, 2006). Likewise, the term enterprise remains poorly defined. In this thesis I will contend myself with the following definition: an enterprise is an “organization performing functions of some scope, complication or risk” i.e. they are highly complex, dynamic and in a state of constant change (Hobbs, 2006). In this context architecture forms “a structure or the practice of designing structures”. The absence of a universal definition has inadvertently resulted in divergent streams of research and views on the concept and role of EA in managing and coordinating organization wide IT activities. To wit, EA has acquired over the years four divergent but related definitions: 1) as a management program 2) as a documentation methodology (Bernard, 2005); 3) as a process modeling tool (Schekkerman, 2006) and 4) 12
as a tool for business strategy execution (Ross et al., 2006). These views have naturally led to divergent streams of research, with some focused on the technology aspects while others have focused on the strategic business role that EA could play within organizations (e.g. Bernard, 2005; Kappelman et al., 2008; Ross et al., 2006; Saha, 2006). The idea of EA dates back to the mid1980’s. John Zachman, then working at IBM, noted the need to use the idea of a logical construction blueprint (i.e. architecture) as a paragon to define and control the integration of organizational IT systems and their components (Bernard, 2005). In general, EA has consequently become an overarching term for enterprise level strategic and operational alignment activity for IT i.e. it covers most of the work that is needed in an organization to align its IT assets with its business goals and needs (Nunn, 2007). Strassmann (2005) argues also that complexity and costs will rise with IT usage, absent of the structure and order that EA potentially brings. At the strategic level, EA defines how an organization’s core processes, core data and core technologies work together to drive business value in accordance with the operating model of the business (Ross et al., 2006). EA ensures congruency between organizational strategies, process, and IT requirements forming an inclusive IT strategy (Webb & Young, 2007). A workable operating model guides the desired level of business process integration and business process standardization for delivering goods and services (Ross et al., 2006). The several divergent views of EA have led to varied conceptual definitions of EA programs depending on the main objective for its adoption and implementation. EA in its simplest form is viewed as a systematic application of architectural fundamentals to manage the complexity of enterprises (Kappelman, 2010). EA is therefore a 13
“management program” to do the alignment and a documentation method to describe the IT assets and their relationships. Viewed as a management program, EA provides overall governance and processes that determine resource alignments, develop standardized policies, enhance decision support and oversee development activities (Bernard, 2005: 35). EA can thereby help in identification of performance gaps in business activities and IT support capabilities. As a resource alignment tool, EA supports strategic planning and operational resource processes that can maximize efficiency and effectiveness of the resources, enhancing the organization’s competitive posture. EA also helps in the implementation of standardized management policy for the development and utilization of IT and other organization resources (Bernard, 2005). Viewed as a modeling methodology, EA models five domains across the enterprise: 1) business architecture which is used for modeling and defining how a business unit or significant business processes operate; 2) information/data architecture which is used for guiding the use of data resources; 3) applications architecture which is used for the architecture of individual applications; 4) integration architecture which is used for the design and implementation of methods and solutions to achieve application interoperability and 5) infrastructure architecture which is used to guide the selection and configuration of hardware, software and communications components for specific domains (Sessions, 2007). Viewed as a documentation methodology, EA is about the “analysis and documentation of an enterprise in its current and future states from an integrated strategy, business, and technology perspective” (Bernard, 2005: 31). Enterprise architects use various EA metadata and process orchestration tools to understand and document the 14
structure of the enterprise. In doing so, they produce artifacts and blueprints which are often broken down into four areas: business, applications, information and technology (Saha, 2004). EA is thus viewed as “illuminating the business and IT domains, as well as the connections and relationships between them, by seeking to make explicit all the knowledge about the entire enterprise” (Kappelman, 2008: 14). In what follows I will be mostly using the idea of EA as a complex strategic innovation program that seeks to align the mission of the organization with the deployment of its IT assets. EA is also viewed as a transformation tool - it helps transform IS planning and management into an integrated, enterprise-wide, strategic activity, rather than what is all too often a stove-piped, disintegrated, series of tactical planning exercises centering on specific and separate IT solutions (Kappelman, 2008). EA is also known to be disruptive. Schekkerman (2006) argues that an enterprise architectural approach is also a huge change in culture of an organization with the people aspect as well as the process. I now define and discuss EA programs. Enterprise Architecture Programs An EA program as a complex strategic initiative forms an overarching IT governance structure that includes strategic planning, enterprise architecture, program management, capital planning, security, and workforce planning (Bernard, 2005). An EA program therefore concerns itself with the integration of other enterprise-level management processes, a process and capability that can be assessed against assimilation frameworks (Buchanan & Soley, 2002). The mission of EA Programs within the federal government is therefore to provide stakeholders the guidance, tools, training, and
15
assistance required to define enterprise architecture and facilitate governance of IT (Schekkerman, 2006). The 1996 Clinger-Cohen Act recognized EA programs as the vehicles for prioritization and alignment of IT investments within federal agencies (Burk, 2005; Hite, 2005). In conjunction with the stakeholders, the EA program team: 1) develops policies, procedures, standards, and guidelines to manage and execute the EA Program 2) prepares responses to reporting requirements to oversight organizations 3) implements and maintains an architecture repository and tool 3) develops and maintains an organization’s business reference model as well as a transition strategy and sequencing plan. The EA program institutes a collaborative, shared planning process within an organization. An EA program is supported by frameworks, methods, models and techniques, which help coordinate many critical facets that make up an enterprise (Schekkerman, 2006). In large organizations such as conglomerates and U.S. federal agencies, the EA program is really a collection of EA programs at various levels throughout the organization. As a management program, EA provides the following outputs as described in Table 2.
16
TABLE 2: Attributes of EA as a Management Program (Sources: Bernard, 2005; Luo et al., 2006) Function Resource alignment
Standardized policy Decision support
Resource oversight
Satisfaction of Legal and Regulatory Requirements
Description Resource planning and standards determination. Reduction in the amount of redundancy and overlap of business processes. Enables and facilitates efficient and optimal use of human and IT resources. Resource governance and implementation. Setting and enforcing EA policies, standards and designs Financial control and configuration management. Being the first source of guidance for all cross-cutting technology investments. Positioning EA as an integral part of technology investment planning. Lifecycle approach to development/management. Reduces redundant data while enhancing data sharing, and reduction in the overall effort for data management. Identification of opportunities (i.e. in planning, training, maintenance, etc.) where economies of scale can be leveraged through common software and enterprise licensing Satisfying the requirements of the Clinger-Cohen Act, the E-Gov Act, the oversight requirements of GAO and OMB, and Budget Circular A- 11 & A130 that impose legal and regulatory requirements for agencies to develop and maintain enterprise architecture.
The EA program not only makes important strategic contribution to the organization, but also drives alignment of business components to the enterprises' strategic direction by increasing integration of business components across the enterprise. In large and complex organizations, the EA program enhances understanding of the complexity of the enterprises' business elements and their relationships to aid in planning, building and operating the enterprise. Perhaps the EA program’s most valuable contribution is orchestration of change management. The EA program has the ability to identify significant transformation opportunities and innovations. The next section defines how EA programs evolve as expressed by the concept of EA program assimilation. It also discusses EA assimilation assessment frameworks.
17
EA Assimilation within Adopter Units In this section of the dissertation the terms “federal agency” and “adopter unit” are used interchangeably. EA assimilation within adopter units refers to a set of internal distinct and often designated and benchmarked capacities possessed by the adopting unit over its course of assimilating the innovation. These capacities are path dependent and can therefore be rank ordered as expressed in widely adopted EA assimilation scales. In addition, each stage assumes a distinct and unique set of skills and competency requirements that need to be fulfilled prior to progression to the next stage. Additionally agencies must satisfy all elements required for one stage in order to advance to the next stage. For instance, an agency that satisfies all elements required for stage 4 ranking, but has yet to satisfy a particular element required at assimilation stage 1 is still considered to be at stage 1. An assimilation framework can be viewed as a set of structured measures organized at multiple levels that describe how well the behaviors, practices and processes of an organization can reliably and sustainably produce expected outcomes (Watts, 1988). An assimilation framework can therefore be used as a benchmark for comparative assessment of across organizations when there is common body of practices and structures that can be used as a basis for comparison. EA assimilation therefore addresses accordingly in a holistic way the elements of strategy, frameworks, the overall EA process, methods and techniques, standards and tools (Schekkerman, 2006). Adopter unit EA programs can thereby be assessed through assimilation ‘rankings’ based on scales that focus on the level of integration of the EA program portfolio with management capabilities and their alignment to the organization’s business objectives (Schekkerman, 18
2006). Ross et al. (2006) argue similarly that such assessments are critical to establishing the organization’s capacity to implement new technologies and promote innovation. The assimilation level is typically expressed in a scale of 1-5 (a.k.a. Likert scale) with assimilation level 5 EA demonstrating that the EA program portfolio is consistently driving and accomplishing the adopter unit’s business goals and objectives (Luo et al., 2006). The scale taps into what extent the EA principles are interwoven into all appropriate business processes and procedures, and affords a set of IT services that reach a ‘harmonious operating environment.’ Next, I review the de facto EA assimilation framework deployed within the US Federal government. The framework (Enterprise Architecture Management Maturity Framework-EAMMF) is the Agency-level (or adopter unit level) maturity assessment framework. This framework has been used by the Government Accountability Office (GAO) to assess the progress of EA programs within the U.S. government and provided the main data points for the quantitative studies in this dissertation. The EA Assimilation Framework (EAMMF) The EAMMF is the U.S. Government Accountability Office (GAO)’s assimilation assessment framework. (Please see Appendix N for details). The framework assesses EA at two levels: 1) the foundational management capability development and 2) the execution framework which includes the application of EA principles within the agency operations. The framework requires that a strong EA foundation is laid at every stage prior to embarking on the next level of elements. At the foundational management level, the framework follows a clearly defined capability development pattern. For instance the stage 2 elements are about setting up of an 19
implementation committee as a prerequisite to the setting up of the EA program office. This is then followed by the appointment of a Chief Enterprise Architect. This systematic chain of events shows a clear progression path and pattern of laying the foundation for the stage 3 activities such as architectural design and enterprise technology selection. Completion of all stage 3 activities lays the foundation for embarking on assimilation stage 4. Similarly, satisfaction of assimilation stage 4 requirements not only creates the foundation for alignment with management objectives, but also paves way for executive level governance policies, driving change and aligning investments with strategic objectives. The framework explicitly requires that agencies demonstrate capacity to effectively execute on elements within one stage prior to engaging in higher level EA assimilation activities (Please see Appendix N). The elements denote the skill level in the application and integration of the EA principles in the day to day management of IT within the agency. The framework indicates that learning through the architecture stages encompasses gradual capability development in both technology and business processes, investment evaluation, leading IT enabled change and process design and execution (Ross et al., 2006). The GAO framework’s objective is thus to deter agencies from pursuing elements in higher assimilation stages before satisfying foundational elements at the lower stages. In line with this the GAO framework is focused on building and measuring the program’s capacity to develop and use an EA over time (Hite, 2005). Thus, the GAO assimilation framework is a capacity and capability development framework, setting performance benchmarks for EA program portfolios (Schekkerman, 2006). 20
At the adopter unit level, the GAO assimilation framework elements are cumulative and incrementally build on capacity and capabilities necessary for assimilation progression. In the framework, all the agencies are automatically considered as being at stage 1 EA assimilation. An agency that satisfies numerous elements within higher assimilation stages without first fulfilling all requirements for stage 1 assimilation will still be ranked as being at assimilation stage 1. As a rule of thumb, all aspects of the particular element must be satisfied by the agency in order to lay claim to having met element requirements. Partial satisfaction would only lead to a “partially satisfied” score and if none of the elements are satisfied, the agency is rated as not having satisfied the element requirements (Please see Appendix E for details). Additionally agencies must satisfy all elements required for one assimilation stage in order to advance to the next assimilation stage. For instance, an agency that satisfies all elements required for stage 3 assimilation ranking, but has yet to satisfy a particular element required at assimilation stage 1 is still considered to be at assimilation stage 1. The framework explicitly requires that agencies demonstrate capacity to effectively execute on elements within one assimilation stage prior to engaging in higher level EA assimilation activities (Please see Appendix N). Ross et al. (2006: 86), in the chapter titled “learning takes time – don’t skip stages” warn of the risks associated with skipping over EA assimilation stages. They argue that learning through the architecture stages encompasses gradual capability development in both technology and business processes and investment evaluation leading IT enabled change and process design and execution. They caution against deployment of advanced technologies in order to skip certain stages. This logic is in 21
alignment with the GAO principles in deterring agencies from pursuing elements at higher assimilation stages before satisfying foundational elements at the lower assimilation stages. EA Adoption and Assimilation EA adoption. The adoption of an IT innovation can be defined as making a decision to use it (as a change) for improving value chain activities (Chau & Tam, 1997). The decision includes acquisition of the technology and allocation of resources. The adoption decision has major significance, as it legitimizes resource allocation required for the deployment of the innovation (Cooper & Zmud, 1990). Adoption stage is deemed a necessary step toward the widespread usage of the technology. In the traditional diffusion of innovation theory the adoption of an innovation begins with an organization becoming aware of the innovation and then evaluating it. In the original version of the Rogers (1995) five stage technology adoption S-curve, adoption ranked highest with contact, awareness, understanding, and trial use/training preceding it. Then the technology adoption lifecycle describes the adoption of the new product or innovation according to the demographic and psychological characteristics of defined adopter groups (Rogers, 1995). Tolbert and Zucker (1983) and Parsons (1951), however, argue that adoption is a multi-dimensional phenomenon that includes the aspect of legitimacy defined by the institutional context. Tolbert and Zucker (1983) define accordingly adoption in institutional context as “the passage of any legal requirement for the institution of procedures.” They however caution that legal requirements do not always ensure adoption (Tolbert & Zucker, 1983). Early adopters are primarily driven by the need to 22
resolve specific problems confronting them whereas late adopters are primarily driven by the desire for legitimacy in the context in which they are located (Rogers, 1995). Other studies have identified and evaluated accordingly a range of variables contributing to success of initial innovation adoption given the institutional context. According to the literature, reform and innovation diffusion have had mixed degrees of success across various institutional contexts (Gelfand, 1975; Griffith, 1974; Thelen, 1972; Wiebe, 1967). Griffith (1974), Rowan (1982) and Tolbert and Zucker (1983) each conducted research studies on reform diffusion. Griffith (1974), who researched the civil service reform imposed on fire and police departments in the state of Ohio in 1884, found that reform adoption stalled until the municipal civil service was given constitutional status. Rowan (1982), researching curriculum reform in public schools, discovered that lack of consensus on the value of reform can lead to failure in adoption or to early rejection. Tolbert and Zucker (1983), researching reform adoption of civil service procedures, found that when adoption was required by the state, the rate of adoption was rapid only within the first ten years. EA assimilation. Assimilation in contrast to adoption can be defined as a series of stages that follow the organization’s initial formal adoption and lead to a widely accepted and pervasive deployment of the system to a point where it becomes a routine as well as an important part of the organization’s activities (Fichman, 2000). Fichman & Kemerer (1999) defines institutionalization as the extent to which an innovation has become a stable and regular part of organizational procedures and behavior that is in line with Tolbert and Zucker’s (1983) original definition. Finally, Ash (1997) defines institutionalization as “routinization” in terms of incorporating the innovation into the 23
daily fabric of the organization. The basic point here is that assimilation as a process of ‘infusion’ is a long drawn process of integrating and institutionalizing the innovation into the operational and social fabric of the organization. Fichman and Kemerer (1999) innovation assimilation model then identifies seven stages which incorporate the innovation from adoption into full scope assimilation: contact; awareness; understanding; trial Use/Training; adoption; Institutionalization and finally Internalization. In the context of EA this can be viewed as a process where the organization becomes initially aware of EA’s potential as an innovation to enhance the organization’s performance. This realization increases the likelihood of EA adoption (Armstrong & Sambamurthy, 1999; Sethi & King, 1994).This initiation stage will be eventually followed by the adoption (Rogers, 1995; Chau & Tam, 1997) where the organization makes a positive decision to adopt the innovation. After the adoption the innovation will undergo a sequence of uptake and infusion steps where it becomes widely accepted within the organization, adapted, routinized, and finally institutionalized (Zhu, Kraemer, & Xu, 2006). Due to the complexity and pervasiveness of EA, its assimilation will have a major impact on an organization’s’ processes, structures, behaviors and culture. This equally demands involvement and commitment from the highest levels of the organization (Schekkerman, 2006: 32). Accordingly, due to its pervasive and transformative impact EA programs and related assimilation processes can be viewed as diffusion and infusion of a radical and complex administrative innovation in an adopter population (Lyytinen & Damsgaard, 2001). As EA is adopted it therefore overcomes major and distinct adoption hurdles as indicated by the idea of assimilation stages and related “stage gates” 24
(Armstrong & Sambamurthy, 1999). Put another way- EA assimilation involves deep and significant change in the organization’s beliefs. Institutional Theory and Innovation Assimilation The innovation assimilation process involves institutional change in the beliefs that underlie and justify organizational activities i.e. something moves from being alien and not accepted to something that is familiar and accepted i.e. legitimate (Scott, 1992; Tolbert & Zucker, 1983). Therefore institutional theory is a highly relevant lens to analyze and understand the assimilation process. Institutional theory defines institutions (such as a complex EA programs) as “social constructions” made up of symbolic elements which include representational, constitutive and normative rule systems (DiMaggio & Powell, 1983). Accordingly, theory identifies three basic types of institutional isomorphism and related origins of changes in beliefs: coercive, mimetic, and normative. These processes or ‘pressures’ also reflect three analytically distinct processes of institutionalization that lead to changes in beliefs (DiMaggio & Powell, 1983). Coercive isomorphism is an attempt by an organization to conform to pressure from institutions on which they are resource dependent (DiMaggio & Powell, 1983). The most common form of coercive pressure is in the disguise of government regulations and policies (Mezias, 1990; Tolbert & Zucker, 1983). Mimetic isomorphism arises from organizations responding to uncertainty by mimicking other organizations either in their industry or market. In mimetic isomorphism, organizations tend to emulate or model themselves after organizations perceived as successful or “legitimate” (Tolbert & Zucker, 1983). This has also been commonly referred to as the “bandwagon effect” (Staw & 25
Epstein, 2000). Normative isomorphism finally is caused by institutional conditions associated with establishment of etiquettes; a “code of conduct”; “code of dressing” which are eventually institutionalized, inculcated and imbibed by the institution membership (DiMaggio & Powell, 1983). This is usually the product of orientation, disposition and control mechanisms associated with social and professional networks, professional associations and membership clubs. I will next review each of these processes in the light of EA assimilation processes. Coercive Pressure The use of the term coercive pressure in this paper means that organizations- in this case federal agencies- are being coerced into moving in the prescribed direction and change their beliefs about the utility of an administrative innovation- in this case EA. This requires not only identification of the required direction but also the ‘target’ change destination (Ashworth, Boyne, & Delbridge, 2005), which in our case is moving up in the EA assimilation levels. Thus “coercive pressure” from an institutional perspective suggests that organizations over time will become susceptible to pressures for conformity and the need for legitimacy due to influence of coercion such as punitive acts towards them (DiMaggio & Powell, 1983; Meyer et al., 1977). Numerous earlier studies show that the presence of coercive pressure is often critical for assimilation of an innovation (e.g. Penttinen & Tuunainen, 2009; Tolbert & Zucker, 1983; Teo, Wei, & Benbasat, 2003). Kaul (2006) in addition note that there is a growing trend for governments to intervene through coercive policies while they promulgate IT innovations within their jurisdiction. King et al. (1994), however, argue that such institutional intervention can also hinder assimilation- by pointing for instance 26
to the lack of clarity of the objectives by the constituent arms of government in formulating coercive policies. They argue that the dilemma of government’s coercive policy is a “microcosm of a broader set of institutional concerns, of which government institutions form an important but incomplete part” (p. 5). King et al. (1994) also acknowledge the absence of research attention to such institutional factors in IS innovation assimilation. This dissertation’s focus on coercive pressure is in part to validate or invalidate a major assumption associated with institutional theory - that government has unlimited capacity to regulate conduct and enforce compliance with its mandates. This argument has inadvertently positioned institutional coercive pressure as the de facto determinant of government mandate assimilation. This dissertation will conduct a time series analysis, to test EA assimilation within three time periods – one under intense institutional coercive pressure and the other two without, and assess the impact of coercive pressure on the assimilation process. The next section discusses EA assimilation in light of these theories. Three Levels of Enterprise Architecture Assimilation I observe three different levels or units of analysis for evaluating EA assimilation. These are the program, adopter unit and population levels2. The program level assimilation refers to the gradual acquiescence of an EA program’s IT governance authority within the organization. In other words, the EA program gradually shedding the “alien” tag, and gaining recognition and legitimacy as an integral part of the organization’s overall management structure. Because of EA’s
2
In both program and adopter unit vs. adopter unit and population analysis there are also significant interactions between these layers. For example, the adopter unit behaviors are influenced by the behaviors of the adopter population (through e.g. mimetic pressure).
27
propensity for radical transformation, EA program assimilation is often fraught with suspicion and frustration. Enterprise Architecture tends to be “viewed as a hostile takeover by program managers and executives who have previously had a lot of independence in developing solutions for their own requirements” (Bernard, 2005: 29). The EA program introduces and deploys radically new standards, policies and associated behavioral changes into organizational routine. Bernard (2005) writes that EA brings new language and planning processes, which like any type of change, can be seen as threatening to those involved and therefore may be resisted (Bernard, 2005). Increased assimilation is viewed as critical to an EA program’s effectiveness (Bittler & Kreizman, 2005). EA program assimilation leads to a sturdy and reliable IT governance structure that includes strategic planning, enterprise architecture, program management, capital planning, security, and workforce planning. EA assimilation at the adopter unit level involves either a single EA program or an EA program portfolio - which progresses through distinct stages or levels that capture the depth and scope of assimilation of EA principles, processes and related behaviors into the adopter unit. The adopter unit EA assimilation levels therefore demonstrate the EA program portfolio’s depth of integration of enterprise-level management standards, processes and capabilities. At the adopter unit level, each EA assimilation stage involves inculcation of “organizational learning” about how to apply IT assets and principles and how to harness business process discipline with strategic capabilities (Swanson & Ramiller, 2003; Ross et al., 2006). An MIT Center for Information Systems Research (CISR) study on EA related budgetary patterns shows, for example, that IT budgets for companies at higher levels of EA assimilation (called maturity by Ross et al. (2006) are 28
at 85% of firms at earlier levels of assimilation, thus indicating higher efficiency in usage of IT assets. In contrast, IT budgets of companies at later stages of EA assimilation are at75% of budgets for those firms within early EA assimilation stages. This may be an indication that transforming effects start to affect the deployment of IT assets as firms move through assimilation stages (Ross et al., 2006). Hence, each EA assimilation level involves additional elements of “organizational learning” about how to apply IT assets and principles and how to harness business process discipline with strategic capabilities (Ross et al., 2006). However this learning is a function of the EA programs. Thus EA program assimilation is essential for enterprise-wide application of the EA principles that are responsible for yielding the benefits for the organization. Finally, at the adopter population level the assimilation process focuses on a set of potential adopters and their emergent properties and how these properties can and will change over time (Lyytinen & Damsgaard, 2011; cf Boonstra & de Vries, 2008). Lyytinen and Damsgaard (2011) observe that within the adopter population “the scope and size of the adopter population depends on the technological, temporal and institutional elements”. They accordingly argue that the assimilation process undergoes the constant shaping of the technology by the adopters and promoters, and structural properties of the networks created by adopters. They accordingly note that adopter populations are different at different stages of innovation assimilation and may have differentiating effects on the adopting unit and related determinants of assimilation. The scope and size of the adopter population depends on the technological, temporal and institutional elements, a view this study will explore further by examining the EA adopter population at three different time points. 29
EA assimilation is often punctuated at the population level by distinct phases which can be viewed as external environmental conditions when viewed by the adopter unit. Several studies have shown recession or increased significance of such determinants during the assimilation process. Tolbert and Zucker, (1983) for example conducted a study on a group of factors that influenced adoption of civil service reform in US cities. The assimilation process (over the population) was divided into four distinct phases. Using data on the adoption of civil service reform by cities from 1880-1935, they found that some factors lost their explanatory power whilst the assimilation process progressed. Some factors maintained their significance through the first three periods and then dramatically dropped in the fourth period. Likewise Wang (2008) in an 11 year longitudinal study of the effects of external pressure on the assimilation of IT innovations found that several factors receded or gained significance at different periods. An assimilation phase thus is a property that can only be applied to populations. It is here defined through a set of organizations that participate - in line with (Lyytinen & Damsgaard, 2011) - in specific type of assimilation activity and have the potential to influence other organizations. Thus an assimilation phase forms an external, temporally framed way of punctuating the innovation trajectory within an adopter population. It refers to distinct temporal (time-based) segments of the assimilation process and related characteristic behaviors observed within the adopter population. What We Know and Do Not Know of IT Assimilation What We Know Table 3 below summarizes the factors known from the literature as affecting EA assimilation. Fichman and Kemerer (1999) provides three categories of factors affecting 30
IS innovation diffusion and assimilation. The first category is comprised of factors that influence technologies and diffusion environments. These include innovation characteristics. Fichman argues that innovations possessing favorable characteristics (such as ease of use or familiar functionality and interfaces) tend to be more attractive and easier to adopt and hence diffuse more rapidly than those with less favorable characteristics. At a macro level, Rogers (1995) argues that it is how the members of a population collectively perceive the characteristics of an innovation that determine its rate of adoption within that population. The other factor is the influence of the propagating institutions. Some highly complex technologies can be moderated by the actions of institutions seeking to propagate those innovations (Eveland & Tornatzky, 1990; King et al., 1994; Swanson & Ramiller, 1997). Institutions mentioned in the literature include technology vendors, consulting firms and user groups which eventually help in determination of the level of resources required for promotion, and enhancement of the technology. The literature also tells us that over time new enhancements are incorporated into the innovation and assimilation barriers are lowered by the introduction of simplified deployment techniques and methodologies (Fichman & Kemerer, 1999; Lyytinen & Robey, 1999; Weill & Ross, 2003; Ross et al, 2006; Lyytinen & Damsgaard, 2011). The environment and the effects of distinct variables influencing EA assimilation may change significantly over time (Lyytinen & Damsgaard, 2011). Institutions, it is argued, have both persuasive and coercive control over the practices, rules and belief systems of those under the institution’s sway (Kimberly, 1998). We also know from the literature that institutions can exert influence via education and socialization processes of individuals, systematic articulation of viewpoints and by 31
rewarding “appropriate” behavior via incentives while punishing “inappropriate” behavior (King et al., 1994). The literature tells us that organizations faced with similar pressures from institutions will eventually resemble each other, or be “isomorphic.” Isomorphic changes within organizations can be reactionary responses to coercive and normative pressures of external institutions, as well as through the mimetic processes, or “modeling,” borne out of uncertainty (DiMaggio & Powell, 1983). TABLE 3: What We Know Researcher Rogers (1995) Fichman & Kemerer (1999) (Eveland & Tornatzky, 1990; King et al., 1994; Swanson & Ramiller, 1997). Rogers (1995)
Sedera & Zakaria (2009)
Fichman & Kemerer, 1999; Lyytinen & Robey, 1999; Well & Ross, 2003; Ross et al, 2006; Lyytinen & Damsgaard, 2011, Chatterjee et al., 2002 Kimberly, (1998). Lyytinen & Damsgaard, 2011 Meyer & Rowan, 1977; Tolbert & Zucker, 1983; 1983, Zhu et al., (2006); Liang et al (2007) and Wang, (2008) (DiMaggio & Powell 1983). Ross et al, 2006; Schekkerman, 2006
Factors that influence EA assimilation Innovation characteristics such as including relative advantage, compatibility, trialability, and observability. Complexity- the more complex an innovation for an organization (e.g due to lack of knowledge and skill) the less the chances of assimilation
Value perception - how the members of a population collectively perceive the characteristics of an innovation that determine its rate of assimilation within that population Knowledge acquisition - acquiring knowledge pertaining to implementation, operation, maintenance, and training can enhance assimilation Lowering of knowledge barriers over time - barriers are lowered by the introduction of simplified deployment techniques and methodologies
Time and Environmental influences. Institutional pressure - institutions have both persuasive and coercive control
Institutional isomorphism- reactionary responses to coercive, normative and mimetic processes Organization’s IT and architectural capacity, or its internal capabilities for IT development
32
What We Do Not Know I observed numerous gaps in our understanding of the antecedents of EA assimilation at three levels. Even though EA assimilation is explicitly or implicitly conceived as a systematic process consisting of multiple stages and phases, an extensive literature search reveals that there are no empirical longitudinal studies on EA assimilation that analyze several of the above layers simultaneously. Most importantly, there are no studies that have addressed EA lifecycle – from adoption to assimilation from a multi-level perspective. This signifies a void in our understanding of EA both as an administrative innovation and as a highly complex multilevel IS phenomenon. For instance at the micro level, even though significant research has been carried out on EA program deployment- especially on its nature, principles and effect- a question which has so far not been adequately addressed includes the drivers of EA program assimilation. Previous studies have focused on and shed light on factors that influence IT units (e.g. Swanson, 1994). However, IS studies and EA in particular has never been researched as an administrative innovation program. Whereas there are numerous IS studies on factors that influence adopter unit assimilation, their contextual focus however are the technologies and processes (e.g. Fichman & Kemerer, 1997, 1999; Zhu et al., 2006a; Bajwa et al., 2004). Few empirical IS studies have shed light on the influence of internal or external factors on complex IS assimilation within the entire organization. The author could only locate a handful of studies that have focused on external factors affecting levels of IS assimilation- mostly in the context of ERP systems (e.g. Bajwa, 2004; Zhu et al., 2006a). Whereas prior studies have extensively covered organizational level factors and their 33
impact on assimilation of innovation, I could not locate a single study that has focused on factors that influence organization wide EA assimilation. Extant meta-analyses of the innovation assimilation literature (e.g. Fichman & Kemerer, 1999; Zhu et al., 2006a, b; Ross et al., 2006) reveals the lack of unified theoretical frameworks that would predict the effects of EA assimilation determinants across phases and stages. Other motivating factors behind this research include the inadequacy of existing models and analyses in explaining the antecedents of EA assimilation. Even though there exists numerous and well conducted case studies on EA assimilation they remain limited in terms of empirical scope (e.g. Ross et al., 2006; Schekkerman, 2006). Moreover, the existing empirical studies on diffusion and assimilation of innovation have tended to focus on specific aspects of the IT innovation and related institutional factors (Fichman, & Kemerer, 1997, 1999; Bajwa et al., 2004; Swanson, 2004; Zhu et al., 2006a; Liang et al., 2007; Wang, 2008). There are no published studies on the antecedents of EA program assimilation. Extant literature reviews have so far yielded neither studies on the antecedents of EA assimilation stages nor the effects of environmental context on the EA assimilation process. Existing empirical studies on the diffusion and assimilation of enterprise IS such as ERP and e-business (e.g. Zhu et al., 2006a; Liang et al., 2007; Wang, 2008) have provided valuable insight into factors that can be conjectured to influence EA diffusion and assimilation. However, I was determined to conduct a “bottom up” investigation to better comprehend the factors from a scholar-practitioner point of view.
34
Research Motivation On one hand the challenges associated with assimilation of EA within the US federal government are well documented, defined and described (e.g. GAO-06-831; GA0-04-40). On the other hand we have several significant gaps in our understanding of how such complex IT innovations are assimilated at the programs, adopting units and populations levels as critically reviewed above. These two observations- one practical and one theoretical- form the key motivation to conduct this study. I will next outline the research scope, goals and methods of the study. Research Scope Using a federal-wide agency EA dataset from 1999-2007, the scope of this study is confined to investigating: 1) the antecedents of federal agency EA program assimilation 2) the antecedents of EA assimilation within individual adopter units (federal agencies) 3) the antecedents of EA assimilation within adopter populations (entire federal EA community) and 4) the effect of changes in the environmental context on the antecedents of EA assimilation (across three assimilation phases – phase 1:1999-2001; phase 2:2002-2004 and phase 3: 2005-2007). The study covers 123 EA programs out of a total of 163 federal agency Enterprise Architecture programs representing approximately 75% of federal-wide EA programs. Section Multilevel Research Study Choice of a Research Design The dissertation used mixed methods research design and methodology to address the above questions (Creswell, 2006; Crotty, 1998). Experimental research, survey research, ethnography, and mixed methods are all specific research designs. This 35
dissertation’s research methodology involved thus collecting and analyzing and ‘mixing’ both quantitative and qualitative data for theory development and validation. This idea of mixing follows the growing trend where researchers collect both quantitative and qualitative data in the same studies (Creswell, 2006) and use related research approaches in combination in order to gain a better understanding of the research problem and have better means to validate the findings. FIGURE 1: Dissertation Triangulation Visual Design Model
36
As shown in Figure 1, which captures this dissertation’s data collection and analysis approach, I was persuaded by the argument that mixed methods research provides strengths that help offset some of the weaknesses of using quantitative and qualitative research alone (Jick, 1979). Jick (1979) and Creswell (2006) argue that quantitative research is weak in understanding the context or setting in which people talk or act, and therefore the authentic voices of participants are “not directly heard in quantitative research.” Jick further argues that quantitative researchers’ own personal biases and interpretations are seldom discussed, because they primarily operate in the background. Qualitative research makes up for these weaknesses as it reveals and puts in the foreground the personal biases and offers ways to validate the findings with rich data.. Other researchers equally point out that qualitative research has its own deficiencies, because the small number of observations and the personal interpretations made by the researcher during the process can create an ensuing bias resulting in difficulties such as in generalization of findings to larger groups. There were several reasons underlying my selection of a mixed methods methodology. The most compelling one was the fact that this was an exploratory study involving a highly complex and institutionally pervasive innovation at multiple levels for which I sought to develop a more generalizable account. At the same time this was an exploratory study - both the research topic and context had never been researched before. Therefore existing methods and theories offered little material guidance how to approach the research problem at the start. Therefore the study resorted to semi-structured interviews to gather data using a bottom-up approach. At the same time the research problem called for answers that go beyond simple qualitative interpretation. I needed 37
quantitative analysis for generalizability and to address the longitudinal nature of research designs. As can be seen from Figure 1, in the context of this dissertation, a mixed method research approach provided several advantages: 1. it provided far more comprehensive multi-level evidence for studying the research phenomenon than either quantitative or qualitative research alone 2. it allowed for multiple sources of data – for instance the qualitative study did not have a set of hypotheses to test, hence data collection via interviews led to identification of new factors, paving way for hypotheses formulation and streamlined quantitative data collection, 3. it allowed for recursive research design (see Figure 1). For instance, upon initial quantitative testing of factors using partial least squares regressions, I was able to target on particular EA programs for further qualitative probing as a form of data triangulation. 4. it helped answer questions that could not have been answered by qualitative or quantitative approaches alone. There were several instances of crossreferencing of data while the author tried to match some of the factors identified in the literature with real life accounts and observations. For instance some of the factors identified in the Ross et al. (2006) typology, in addition to general literature of IS innovation (e.g. Fichman & Kemerer, 1999; Zhu et al., 2006a) could now be “re-confirmed” by re-interviewing some participants from target assimilation categories to ascertain their hypothesized impacts on EA assimilation. Overall, the combined effect of the qualitative and quantitative approaches was greater than its separate parts (Creswell, 2006). The employment of mixed methods research encouraged the use of multiple worldviews or paradigms rather than the typical association of certain paradigms for quantitative researchers and others for qualitative researchers. For instance institutional theory, which provided one of the theoretical lenses through which this dissertation was conducted, recognizes institutional forces as shaping organizational systems. This dissertation, thanks to the deployment of the qualitative
38
study, unveiled that individual actors within organizations can in fact drive behavioral changes within institutions. Structure of the Research Design Table 4 below provides an overview of the research methods used to address the study’s research questions. The last column shows the preferred approach or technique used for the analysis. TABLE 4: Research Methods & Techniques Research question
Research Method
Approach/Technique
RQ 1: What are the antecedents of EA program assimilation? RQ 2: What are the antecedents of EA assimilation within individual adopter units? RQ 3: What are the antecedents of EA assimilation within adopter populations?
Qualitative study Quantitative study
Grounded Theory (Glaser & Strauss, 1995; Corbin & Strauss, 2003) Partial Least Squares regression (Chin, 1998); MODPROBE (Hayes, 2003)
Time series
RQ 4: What is the effect of changes in the environmental context on the antecedents of EA assimilation?
Time Series
Ordinary Least squares regression (Hair et al, 2010); Ordered Logistic regressions Ordinary Least squares regression (Hair et al, 2010); Ordered Logistic regressions
Research question 1 was addressed via a qualitative study (please see Table 4) using semi-structured interviews. I used a semi-structured interview approach, using informal conversations that were guided by a set of common questions. The questions were complemented by probes used by the researcher to elicit detail. Participants were asked about specific events and actions, to avoid generalizations or abstract opinions (Nickles & Weiss, 2004). Research question 2 was addressed via a cross-sectional quantitative study using Partial Least Squares (PLS) regression technique. Research questions 3 and 4 were
39
addressed via a longitudinal quantitative study using ordinary least squares regression and ordered logistic regression techniques. Creswell (2006) argues that mixed methods research is “practical” in the sense that the researcher is free to use all methods possible to address a research problem. It is also “practical,” because practitioners tend to solve problems using both numbers and words, they combine inductive and deductive thinking. This paper’s approach exemplifies Creswell’s argument. For instance it will be seen in Table 9 that this dissertation’s use of multiple methods developed both a theoretical and practical link between the three studies. The table shows that 100% of the qualitative study findings pinpointed the factors that influence EA assimilation from a practitioner perspective. The second study, using the findings as a guide, conducted literature reviews, formulated hypotheses and conducted data collection. The study then conducted confirmatory factor analysis before testing the factors using cross-sectional dataset. The third study extended the second research by conducting a longitudinal quantitative study and testing the factors using time series analysis to establish the explanatory power of the factors over the period under review. The next section provides highlights of the data sampling techniques for the three research studies. Data Sampling The first study, which was qualitative, used both probability and purposive sampling methods as shown in Table 5. For the second and third studies I collected publicly available government data from 123 EA programs out of 163 Federal Enterprise Architecture programs to validate the research model. This represents approximately 75% of federal-wide EA programs. I dropped 17 programs for the following reasons 1) data 40
were not available for some items as they are secret (e.g. several Department of Defense agencies, Central Intelligence Agency, Federal Bureau of Investigations, National Security Agency), 2) some agencies were too small which would bias the sample e.g. the Federal Railroad administration (FRA), the Legal Services Corporation (LSC) 3) some agencies are too large and would bias the sample (e.g. the department of health and human services and the department of transportation as well as the Department of Defense). By using these criteria I included in the sample 123 agencies. Table 5 below highlights the various sampling techniques used in this study and their attributes. TABLE 5: Sampling Techniques Used In Study Sampling Technique Purposive sampling
Subcategory
Description
Sampling to achieve representativeness or comparability
I sent out interview invitations to purposely selected government officials. The criterion used was simple. I looked up organization charts for each of the federal agencies. I then traced back the respondents’ programs based on assimilation ranking – i.e. those that had low assimilation ranking and those with high ranking for comparative analysis. I also divided the respondents based on whether they belonged to one arm of government or not – for instance if the parent arm was the judiciary, the executive or the legislature Once the interview invitation responses were received, either with acceptance or rejection, I then embarked on dividing the names into sub-groups e.g. EA practitioners or executive management I then created clusters based on whether the respondents belonged to a department, agency, bureau or sub agency
Sampling unique or special cases
Probability sampling
Stratified sampling
Cluster sampling
Study Study 1 (Qualitative study)
Study 1 (Qualitative study)
Study 1 (Qualitative study)
Study 1 (Qualitative study)
For the qualitative research study, face to face and telephone interviews lasting between 50 to 90 minutes were conducted. All interviews were recorded and subsequently transcribed. Follow up interviews were conducted as a validation measure 41
(Bryman, 1988; Lincoln & Guba, 1985). These were also transcribed. All email and follow up interview data were appended as addendum to the original interview transcripts. The second and third studies used panel data collected over a 14 month period in multiple stages. Some of the data had been vaulted at the US National archives and Records Administration Agency (NARA). The U.S. government has a five year public data access policy following which the data is sent to the national archives for vaulting. Permission had to be sought from the respective past and present Congressional and Senate committee chairs for release of some of the data. Some of the data were also obtained from annual inspector general reports and agency budgetary proposals. Other data were obtained from the 1999, 2001, 2003 and 2006 GAO EA audit surveys. The survey results were obtained from the National Archives and Records Administration. The rest of the data were obtained from agency web-sites, Code of Federal regulations; gpoaccess.gov and OMB. This next section presents the research highlights. Research Highlights This dissertation is split into three discrete but related studies that address four different research questions. The table below provides a short description of the studies and the accompanying research question(s) that each study addresses.
42
TABLE 6: High Level Description of Studies Study # 1
Research question What are the antecedents of EA Program assimilation
2
What are the antecedents of EA assimilation within individual adopter units?
3
What are the antecedents of EA assimilation within adopter populations? What is the effect of changes in the environmental context on the antecedents of EA assimilation?
4
Description This study was interested in exploring the organizational and institutional factors that influence individual level EA program assimilation within federal agencies. The study elected to use qualitative semi structured interviews to capture individual accounts from EA professional and managers This study was interested in investigating the factors that influence EA assimilation at adopter unit level. The study used Partial Least Square regressions on cross sectional dataset. This study was interested in investigating the antecedents of adopter unit populations over time. This part of the study was interested in establishing the effect of environmental changes on the explanatory power of the antecedents of EA assimilation.
Study Dependent Variables The three studies had a common dependent variable – EA assimilation. However, the definitions and operationalization of EA assimilation at the three levels varied. The table below shows the definition and operationalization of the dependent variable. TABLE 7: Dependent Variable Definitions and Operationalization Study 1
Level Individual program
Dependent variable EA program assimilation
2
Adopter unit
EA assimilation
3
Adopter population
EA assimilation
Definition Moving from alien and unwanted status to embracement and acceptance Degree of application, institutionalization and routinization of EA principles and standards Success in adaptation to prevailing internal and external environmental dispensation or changes
The first study conducted semi-structured interviews with IT and business executives at all U.S. Federal government departments and agencies to assess constraints 43
and inhibitors to EA program assimilation. The study hypothesized organization characteristics such as design, culture and complexity moderated by the organization environment and technology as affecting the assimilation of EA programs. A summary of the study findings are presented in Table 8 below. The section after discusses the findings in detail The second study examined the antecedents of Enterprise Architecture (EA) assimilation, as well as the effects of coercive pressure in accelerating the assimilation process. The study hypothesized five constructs, namely organization complexity, Access to resources, parochialism and cultural resistance, organization scope and management value perception as influencing EA assimilation within adopter units. A summary of the study findings are contained in Table 8. The third study examines (1) the antecedents of EA assimilation phases (2) the impact of environmental changes on the EA assimilation process and (3) the determinants of EA assimilation stages. The study, an extension of the second study, reused the constructs and applied them in a longitudinal study. The study findings are contained in Table 8 below, while details are discussed in the next section. This next section develops a link between the qualitative study findings and the constructs developed for the second and third studies.
44
Major Findings The table below presents the summary of the key findings from the three studies: TABLE 8: Summary of Key Findings Study
Key Finding • •
Study 1
•
•
Study 2
• • • •
Study 3
• • • •
•
Agency culture, size, structure and complexity inhibit E.A. assimilation Framing and labeling of EA as an IT function instead of a strategic business tool reduces its perceived value to the organization, inhibiting its assimilation. Agencies with strong political constituencies are linked to high resistance to change and low EA assimilation, while agencies with weak political constituencies are linked to low resistance to change and higher assimilation Program leader collaborative style and use of creative and aggressive tactics, business knowledge and marketing communications can moderate the effects of agency/department size, complexity and culture on EA program assimilation. Institutional coercive pressure positively moderates EA assimilation Access to resources plays an influential role as a facilitator of EA assimilation. Organizational factors including complexity, access to resources and EA value recognition are significant determinants of EA assimilation The percentage of EA programs languishing at the lower assimilation rankings shrank by 21% during the period when significant coercive pressure was progressively instituted Access to resources and organization complexity are strong determinants across all assimilation phases Parochialism and cultural resistance is a strong determinant only at the early and late phases of assimilation Organizational complexity has determinate effect at all assimilation stages while management value recognition lacks determinate effect across all assimilation stages Coercive pressure has a “jolt” like albeit short lived moderating effect that accelerates the assimilation by reducing the negative effect of organizational complexity and enhancing EA value recognition Support for the theoretical notion of “differently directional effects”: the same determinants may play different roles at different assimilation phases and levels
Study 1: The Antecedents of EA Program Assimilation The first study conducted semi-structured interviews with IT and business executives at all U.S. Federal government departments and agencies to assess constraints and/or inhibitors to EA program assimilation. The results showed that program leaders at lower ranked agencies/departments were more apt than those at highly ranked agencies and departments to describe their organizations as culture dominant, characterized by a high resistance to change. The study also revealed that there are two contrasting ways 45
that EA is perceived within the federal agencies. One perception is that EA is a technical function of the IT group – an activity that is being “forced onto” the agency to benefit technical requirements. Associated with this technocratic perception of EA, a majority of practitioners complained about negative labeling, such as being called “geeks” by superiors and co-workers. An alternative perception is that EA is a strategic business tool ─ an activity that can benefit each bureau with improved performance. The study also found that organization size, culture, structure and complexity are factors affecting EA assimilation. The data showed that entrenched cultural practices, rituals and protocols within the agencies serve as major inhibiters to program assimilation. Structural impediments, such as the number of different constituencies (bureaus and business units) within the agencies were also identified as major impediments to program assimilation. My data identified stark differences between the personal and professional characteristics of program leaders at the agencies and departments with high ranking programs and those at the low ranking programs. These differences include management and leadership style, communication style and personal culture. The table below links the study findings and constructs for studies 2 and 3.
46
Linking Study 1 Findings to Study 2 & 3 Constructs TABLE 9: Link between Study 1 Findings and Studies 2 & 3 Study 1 findings
Program leaders at the highest ranked bureaus employed creative tactics such as publicity, reward and recognition schemes to draw participants and volunteers to their programs, in order to overcome manpower challenges. By contrast, program leaders at the lowest ranked agencies primarily relied on resource deployment from within the agencies. Agency culture, size, structure and complexity inhibit E.A. assimilation progression.
Frustrations were encountered in securing cooperation from the bureaus within the agency. Respondents attributed this failure to cooperate to the disparate missions and diverse legislative and regulatory mandates within the agency. A high resistance to change and low EA assimilation progress was reported by program leaders in agencies with strong political constituencies, while program leaders in agencies with less politically strong constituents reported less resistance to change Framing and labeling of EA as an IT function instead of a strategic business tool reduced its perceived value to the organization, inhibiting its assimilation. EA program leaders at the top ranked agencies exploited their agencies’ vulnerabilities with guerilla tactics such as leaking information to the media and government audit bodies in order to stimulate executive action favorable to their programs. Any hint of negative publicity or threat of punitive action from OMB led to swift action on the part of the agency leaders to blunt the effects
Corresponding Study 2 & 3 construct Access to Resources
Organization Complexity
Organization Scope
Definition of corresponding study 2 & 3 construct The extent to which the organization is capable of securing adequate resources including human capital and money to successfully drive assimilation.
The amount of differentiation among different elements constituting the organization - for example the number of missions, functional units and the number of regulations, as well as their mutual dependencies. The extent and reach of operations affecting the organizations that reside outside its formal boundary.
Parochialism and cultural resistance
The variance of and attitude towards perceptions of changes in the environment often denoted as the “selfish pettiness or narrowness” with regard to external interests, opinions or views
Top management value recognition Coercive Pressure
The perception of the value of an innovation by the management team
The extent of formal pressures exerted on government agencies.
47
Study 2: The Antecedents of EA Assimilation in EA Adopter Units The second study examined the antecedents of Enterprise Architecture (EA) assimilation, as well as the effects of coercive pressure in accelerating the assimilation process. Major study results include: 1) the percentage of EA programs languishing at the lower assimilation rankings shrank by 21% during the period when significant coercive pressure was progressively instituted, 2) institutional coercive pressure positively moderated EA assimilation, 3) access to resources plays an influential role as a facilitator of EA assimilation; and 4) several organizational factors including complexity, access to resources and EA value recognition are significant determinants of EA assimilation. The study, using cross sectional methodology found that institutional pressure had a significant influence on the assimilation process throughout the assimilation lifecycle. Study 3: Antecedents of EA Assimilation Stages and Phases The third study examined (1) the antecedents of EA assimilation phases (2) the impact of environmental changes on the EA assimilation process and (3) the determinants of EA at each assimilation stage. The results indicated that both access to resources and organization complexity are strong predictors of EA assimilation at all three assimilation phases while parochialism and cultural resistance is a strong inverse predictor at the adoption and internalization phases respectively. The evidence shows that organizational complexity has explanatory power at all assimilation stages while management value recognition lacks explanatory power at all assimilation stages. The results suggest that organization complexity is more likely to be a barrier to EA assimilation level advancement from level 2 to level 3 in the absence of coercive pressure. Parochialism and cultural resistance are more likely to be a barrier to EA assimilation level 48
advancement from level 2 to level 3 in the presence of coercive pressure than without. The results also suggest that organization scope is more likely to be a barrier to EA assimilation level advancement from level 4 to level 5 in the absence of coercive pressure. The next section discusses the dissertation limitations. Limitations As with any study, this research has its share of limitations. The context of the study limits its generalizability. The contextual setting is the U.S. federal government, which precludes commercial entities. There have been past arguments that government institutions do possess unlimited capacity to enforce executive interventions. However, in the context of this study, the units of analyses are adopting and assimilating a widely available industry standard that has been shown to be complex and problematic to assimilate even within the private sector (Ross et al., 2006; Schekkerman, 2006; Saha, 2004, 2006). Another dimension to this argument is the considerable controversy over the appropriateness of government policies and actions in the innovation arena. King et al. (1994) question the clarity of objectives articulated in IT-related programmatic statements, which differ by various agencies, commissions and leaders. This particular argument diminishes the effect of the government’s perceived advantage and level- and thus level-sets the playing field with the private sector. The study also had limitations of statistical generalizability. Even though the data sample size was fairly adequate (n=163), it only covered the period from 1999-2007, 3 years after promulgation of EA within the federal government. This may have denied the study the opportunity to investigate the factors that may have been influential at the very early stages of promulgation. 49
For the initial study, I relied on anecdotal data and interviewee accounts. Some of the interviewees were retired or had moved from their former agencies and managerial positions. I thus had to rely on recollections, which in some cases may have been revisionist (Lincoln & Guba, 1994). These accounts were however largely corroborated by audit reports and recollections by colleagues, co-workers and former executives contacted as part of the data collection exercise. Another limitation was the absence of factors that measured the impact of political influence. Institutions, like individuals and organizations are politically and ideologically biased (King et al., 1994). Summary of Key Issues Table 10 below shows a summary of the key issues encountered during each of the research studies and the steps taken by the author to mitigate them. TABLE 10: Summary of Key Issues Study
Issues
Study 1
Respondents asked to recall past events sometimes dating back several years and we recognize that memory can be affected by time Response bias-some respondents painted a positive light and provided socially desirable responses Use of cross-sectional data for regression analysis
Study 2
Study 3
Sample size (n=123) Sample size limitations. For instance couldn’t subdivide dataset further for more granular analysis
Mitigation Triangulation – seeking corroboration from peers and documentary evidence Sought retirees who were unconstrained to shed light on events Used study as CFA for constructs for the longitudinal study Used re-sampling (bootstrapping) Used three subsets of the data
The implications and contributions that the three studies make are captured in chapter V. This next section proposes future areas of research on this topic.
50
Future Research My research focused on factors that influence multi-level EA assimilation within U.S federal Government agencies. The U.S. federal government is financially endowed unlike several other governments in the world, giving it undue advantage in the ability to procure and institutionalize private sector frameworks and assessment tools from top class consulting and outsourcing companies. Future research may want to replicate this study in other less endowed countries that may have promulgated administrative innovations such as EA, and conduct comparative analysis. Future research may also want to replicate this study within private sector conglomerates that may look similar in structure and complexity to the U.S. government and conduct comparative analysis. General Summary This study makes substantial breakthroughs in investigating the factors that can either impede or enhance EA assimilation. First the study invalidated several underlying assumptions associated with institutional theory. Even though (arguably) the U.S. government does possess unlimited capacity to regulate conduct and enforce compliance with its mandate, it has been shown that it is not a sustainable strategy. This dissertation’s findings also rebuff institutional theorists’ claims that mimetic isomorphism among public organizations is rare (Meyer & Rowan, 1977). The dissertation does validate the argument that pursuit of legitimacy is the sole purpose that drives public organizations into embracing reforms and mandates (Tolbert & Zucker, 1983). This is clearly shown as agencies continued adherence to the e-Government Act, even after its lapse in 2006, for pursuit of legitimacy.
51
The author was fortunate to have secured interviews with retired or exiting government officials (the qualitative study was conducted in the midst of a U.S. government transition from President George W. Bush administration to the Obama administration) hence the interviewees were far more candid and unconstrained in their responses. It is equally important to note that EA programs were introduced as a government-wide governance reform and not just IT governance, positioning EA as more strategic than the government functionaries were accustomed to. Past emprirical and longitudinal studies in reform and innovation assimilation have been conducted separately. This dissertation thesis synthesizes the twin phenomena. Although there are numerous studies on administrative reform (Aucoin, 1990; Barzelay, 1995; Bowornwathana, 1994 a, b; Caiden, 1991) there are none on administrative innovation. Conducting this research from a multi-level perspective teases out the finer aspects of EA that have never been clearly articulated in the literature. For instance, as can be seen from Table 7, studying EA programs at a micro level shows that EA program assimilation denotes the efforts to move EA, from the state of being alien and unwelcome to the overall governance framework to being embraced and legitimized as an integral part of the organization governance structure. Studying EA assimilation at the adopter unit level shows that assimilation at the adopter unit level denotes the application and routinization of EA principles within the adopter unit. Studying EA assimilation at adopter population level denotes the adaptation of the population to changing environmental dispensations. So far, no studies have made that distinction. This next chapter covers the qualitative research study that addresses research question 1: what are the factors that influence EA program assimilation? 52
CHAPTER II: INVESTIGATING THE FACTORS THAT INFLUENCE EA PROGRAM ASSIMILATION – QUALITATIVE STUDY OF U.S. FEDERAL AGENCY EA PROGRAMS Introduction According to assimilation rankings issued by the Government Accountability Office (GAO), U.S. Federal agencies and departments have made woeful progress in assimilating Enterprise Architecture programs. Mandated by a 1994 act of congress, EA programs were promulgated to stem waste in government spending on technology. Motivated by a senate report charging lack of value from an estimated $25 billion in annual U.S. government computer related expenditures, the Clinger-Cohen Act directed all U.S. Federal agencies and departments to align their strategic goals and technology needs. Enterprise Architecture programs are designed to have progressive assimilation stages or levels. Assimilation is critical to a program's effectiveness (Buchanan & Soley, 2002). The GAO EA assimilation rankings are the benchmark for EA program assimilation within agencies. A 2006 GAO audit revealed, however, that 84% of federal agencies remained at EA assimilation levels one and two with only 11% having progressed to level three and just 5% attaining levels four and five. In 2008, the federal IT budget exceeded $68 billion, and in striking similarity to the circumstances that led to the passage of the Clinger-Cohen reform Act, the 2009 audit reports from the GAO showed that 585 of the 810 major information technology projects (about 72 percent) faced failure (Grasso, 2009). Numerous congressional and watchdog reports have cataloged the massive losses incurred by the federal government on IT procurement (GAO-08-1015T). Sen. Joseph Lieberman, D-Conn., said, "Federal agencies should be deriving better results from the $60 billion spent annually on information 53
technology. Much of that money is wasted on IT systems that are redundant or obsolete" (Andrues, 2006). The literature is silent about this massive problem of practice. An extant body of normative literature on EA generally (e.g. Saha, 2004, 2006; Bernard, 2005; Schekkerman, 2006; Hobbs, 2006), has focused on EA as an IT or systems discipline and not as either an innovation or a management program. Prior studies on assimilation of innovation within the IS space have addressed factors such as diffusion patterns (Damsgaard & Lytinnen, 1996), process innovation (Mustonen-Ollila & Lytinnen, 2004); and IS assimilation (Fiss & Kennedy , 2009). My study looked at EA programs from both a reform and innovation assimilation perspective, within a unique setting: the U.S. Federal government, where the most wide scale implementation of EA can be found (Bernard, 2005). I conducted semi-structured interviews with IT and business professionals in twenty six Federal agencies and departments to shed light on why some EA programs have attained high assimilation levels while others have stalled. Literature Review In this section I provide a theoretical framework to nest the present study within existing literature. My goal was to explain why some Federal agencies were better able to assimilate their EA programs than others. Because the literature suggests that certain characteristics of an organization – such as culture, structure and complexity -- can affect IS assimilation, I stayed open to that possibility. I begin this section by reviewing literature linking these organizational variables. Thereafter I review several theoretical streams that informed my conceptual framework and the design and conduct of my study. 54
Organizational Characteristics Organization Culture Organization culture − for which more than two hundred and fifty definitions have been proffered (Sackmann, 1991; Kroeber & Kluckhohn, 1952) − can have significant influence on the attitudes and behaviors of organization members (Robbins, 2001). Schein’s (1992) definition of culture as “habits of thinking, mental models, and/or linguistic paradigms” – i.e. “shared cognitive frames that guide the perceptions, thought, and language used by the members of a group and are taught to new members in the early socialization process” suited my inquiry. His suggestion that the most successful way to think about culture is to view it as the accumulated shared learning of a given group, covering behavioral, emotional, and cognitive elements of members’ functioning, linked culture and performance. How an organization’s culture impacts performance has been the focus of considerable research. Kotter and Heskett (1992) conducted a study that challenged the widely held belief that "strong" corporate cultures drive positive business outcomes, demonstrating that while shared values and institutionalized practices can promote performances in some instances, strong cultures can also be characterized by arrogance, inward focus, and bureaucracy -- features that undermine an organization's ability to adapt to change. The goals of a bureaucracy are predictability, efficiency, and stability (Hellriegel & Slocum, 2007). In a bureaucratic culture, managers view their roles as being good coordinators, organizers, and enforcers of clearly defined rules and standards (Hellriegel & Slocum, 2007). Bureaucracy impedes change. The well-documented
55
bureaucratic character of Federal government agencies, I conjectured, might significantly inhibit EA program assimilation. Organizational Structure Organization structure refers to relationships among the “tasks” performed by members of an organization (Hofstede, 1980, 1983) or, as defined by Kotter & Heskett (1992), to certain “organizational arrangements”. Such arrangements, (Kotter & Heskett) explain, “…may call for behavior that is already pervasive in a firm for cultural reasons” (Kotter & Heskett, 1992: 5). Thus conceived organizational structure and culture are inextricably linked. This is despite the two being separately unique phenomena. Structure, like culture (and often in conjunction with it) can impact what organizations do and how they perform. Arguments by Alchian and Demsetz (1972), Garfinkel (1967), as well as Weick, (1969, 1979) that structure does not exist independently of the actions of people − that people create structure, sustain it and terminate it − was pertinent to my study. Crow & Bozeman (1998), building on earlier studies by Snow and Hrebiniak (1980), Bozeman and Loveless (1987) conducted research on the impact of organizational structure and design on performance of the R&D laboratories in the U.S National Innovation System, concluding that performance is a function of the fit between structure and environment. Ford and Slocum (1977) and Tolbert (1985) also studied the relationship between size, technology, and environment and the structure of organizations. The results of this study informed my conceptual framework as I was convinced of the dual factors being critical antecedents to EA program assimilation.
56
Kim (2004), who, as I did, conducted an analysis of all U.S. Federal agencies, rejected the simplistic assumption that they are traditional Weberian bureaucracies. Rather, he found that different agencies with varying policy environments adopt different kinds of structures to accomplish tasks. Agency structures, he argued, differ depending on their policy environments (i.e., regulatory, distributive, redistributive, and constituent) and policy goals. Kim’s demonstration that Federal agency structures vary significantly informed my thinking about why some of them have been more successful than others in assimilating their EA programs. I was consequently sensitive to references to organizational structure in respondent narratives. Organization Complexity Dooley (2002) − who defines complexity as “the amount of differentiation that exists within different elements constituting (an) organization” − notes that organizations are more or less complex as a reaction to environmental complexity, i.e. a complex environment requires a complex organization. If indeed the environment, as argued by Weick (1979), powerfully influences organizational behavior, complexity clearly matters. The literature on organizational complexity has considered it, variously, as a characteristic of the structure or the behavior of an organization. But, as suggested by Fioretti and Vissser (2004), “…complexity matters only because of cognitive problems it gives rise to” – e.g. “the demands it imposes on decision makers concerned with attaining overall organizational effectiveness (Fioretti & Visser, 2004: 2). Evaluated in terms of criteria suggested by the literature (i.e. size, multiplicity of subsystems, inherent nature, etc.), Federal agencies, I conjectured, can, in general, be considered complex
57
organizations. Their complexity, thus, must affect how decisions in and about them are made. Dooley noted that “following the norms of rationality, an organization is more likely to build barriers around its “technical core” if (its) environment is complex.” I wondered to what extent this tendency might affect Federal agency embracement of – or insulation from − mandated IT change. Change implies uncertainty − and Robbins (2003) found that environmental complexity contributes to it. Methods Methodology I conducted qualitative research not only because it allows researchers to get at the inner experience of participants, but also to determine how meanings are formed through and in culture, and to discover rather than test variables (Strauss & Corbin, 1990). Qualitative studies provide insight into people's attitudes, behaviors, value systems, concerns, motivations, aspirations, culture and lifestyles (Ereaut, 2007). At the heart of this study, was establishing the factors that impede or enhance acquiescence of EA programs into the Agency management structure and operations. The Grounded theory approach is a qualitative research method that uses a systematic set of procedures to develop an inductively derived, data-grounded theory about a phenomenon (Strauss & Corbin, 1990). The semi-structured process allowed for flexibility in the delivery of informant stories. At the same time, participant responses influenced the direction of the interviews. This provided the researcher with the opportunity to explore further and adjust the process as required.
58
Sample My sample included executives from all of the U.S. government departments and agencies mandated to implement Enterprise Architecture Programs. Those included in the sample were accessed using the principal researcher’s professional network and ranged from departments with over one million employees and annual budgets totaling over $550 billion to much smaller agencies with fewer than five thousand employees. All of these departments and agencies had implemented EA programs albeit with varying degrees of success. All had been audited at least once during the last 10 years and had an assigned EA assimilation ranking. Requests for interviews were sent via email and regular mail to all twenty-six departments and agencies. I targeted senior ranking officials, with the designations of Senior Enterprise Architect, Chief Architect, Chief Information Officer, Chief Technology Officer, Chief Financial Officer, Deputy Secretary, Department Secretary and Agency Director. I had a one hundred per cent (100%) response rate. Some agencies however, referred us to former EA executives based on my selection criteria that required respondent participation in at least one GAO audit. Given the fact that the study was conducted during an administrative transition, five recently retired Chief Architects who had held the role during EA implementation at their respective departments and agencies were contacted and interviewed. Seventeen EA program executives were long serving Chief Architects with an average tenure of seven years whereas three of the program level interviewees were senior ranking Enterprise Architects. Fourteen agency business executives - one in each of twelve agencies were included in the sample. The remaining fourteen agencies had 59
newly appointed business executives, hence their ineligibility to participate. Federal business executives are U.S Presidential appointees approved by the U.S. Congress to serve for a four-year term. The business executives in my sample included three agency directors, four Chief Financial Officers, two deputy secretaries that serve as Chief Operations Officers, and five Chief Information Officers. The sample included five female and thirteen male EA program executives and nine female and twelve male business executives. Data Collection Twenty-four face to face and fifteen telephone interviews of 50 to 90 minute duration and twenty seven follow up interviews of approximately 30 minute duration were conducted. In some cases, post interview email communication with respondents ensued to clarify or expand on interview responses. In-person interviews were conducted in respondents’ offices or other locales selected by interviewees to provide comfort and confidentiality. Standard qualitative research strategies and protocols (Maxwell, 2005; Glaser & Strauss, 1967) were adhered to in order to limit or reduce biases. All interviews were recorded and subsequently transcribed by a professional service. Follow up interviews were conducted as a validation measure (Bryman, 1988; Lincoln & Guba, 1985). These were also transcribed. All email and follow up interview data were appended as addendum to the original interview transcripts I used a semi-structured interview approach, allowing for informal conversations that were guided by a set of common questions. The questions were complemented by probes used by the researcher to elicit narrative detail and to adjust the process as needed. Participants were asked about specific events and actions, to avoid generalizations or 60
abstract opinions (Weiss, 1994). I began each interview by asking respondents to tell us about themselves and their work. Next I asked specific questions about their respective agencies and EA Programs. I then posed questions centered on preparatory activities for the various audits, reactions to audit results, and any remedial actions in the post-audit period. Finally, participants were asked to recount any unusual incidents, occurrences, and/or and actions associated with EA implementation and audits. The aim was to compare and contrast the various pre-and post audit preparatory activities, the attitude and reactions to the audit exercise and results as well as post-audit activities for each of the agencies. Data Analysis The study employed the constant comparative methodology (Lincoln & Guba, 1985; Glaser & Strauss, 1967). Constant comparative analysis is characterized by concomitant coding and analysis of data in search of themes (Strauss & Corbin, 1998). The labeling or coding process involved identification of passages of text and other meaningful phenomena and categorizing them into similar themes or analytic ideas. The themes and ideas were then comparatively examined and put back in new ways (Strauss & Corbin, 1990). The initial process involved “open-coding” of the interview transcripts. The transcripts and audio recorded interviews were reviewed numerous times to solicit meaning, develop logical categories or themes, and consider classifications in which groups of categories could be assigned without reference to initial conceptual considerations. This process resulted in 602 interview segments, which I next categorized by group. A total of twenty-six categories emerged, which were later 61
consolidated with some eliminated either due to redundancy or data irrelevance. This process eventually reduced my categories to 17. From these categories, seven dominant themes and 9 associated sub-themes emerged that were used in the final analysis. Findings My data produced four major findings and five sub-findings that explain variation in the assimilation of EA programs within Federal agencies and departments. The findings suggest that the framing and labeling of EA as “IT Architecture” as opposed to Enterprise Architecture negatively impacted its assimilation as a strategic planning and budget control utility. The findings also indicate a positive relationship between EA program assimilation levels and individual EA leader characteristics and behavior. EA leader characteristics and conduct appear to moderate the potentially powerful effects of organizational size, culture and structure, as articulated in institutional theory, on progression of EA program assimilation. Finding 1: Agency culture, size, structure and complexity are inhibiters of E.A. assimilation progress Thirty out of thirty executive-level departments and agency respondents identified organization size, culture, structure and complexity as factors affecting EA assimilation progression. Thirteen respondents commented about entrenched cultural practices, rituals and protocols within their organizations that serve as major inhibiters to program assimilation. Eight commented on the number of different constituencies (bureaus and business units) within their respective organizations as major impediments to program assimilation. One EA program executive spoke of the frustrations encountered in securing cooperation from the bureaus within the agency. He attributed this cooperation 62
failure to the numerous disparate missions, as well as diverse legislative and regulatory mandates within the agency. FIGURE 2: Agency Characteristics Agency Size & Complexity “our agency is fairly large, with more than 15 bureaus and different kinds of business functions and business units with their own information technology organizations. So it’s very difficult to have a single architecture that is well integrated, rendering it hard to mature the agency EA program.” (Agency EA)
“The sub-agencies within the department have enormous constituencies (bureaus/business units). They’re very large; they’re very powerful; and they have constituency on the (Capitol) Hill that is substantial. So they don’t have to listen very much to central headquarters. They listen just as they have to – only as much as they have to.” (Former Agency EA)
“our department is very federated and diverseit is so diverse that one size does not fit all in terms of the EA development. Their specific missions and requirements and needs are distinctly different. That’s part of our challenge as an architect is to make it more understood. Everybody wants to sort of accomplish their mission. They’ll do what they feel most comfortable with in doing that.” (Agency EA)
“One of the problems of maturing the EA program is that we have very large agencies – really massive in size – the DOD for example has over 1 million employees and others with over 200,000 employees or something like that, and are basically like a holding company. I mean, all the cats and dogs are in there.” (Agency EA)
“Advancing EA here isn’t easy, well, the Department is a unique agency in that it has a very broad and very dispersed and in fact non-homogenous – it’s like a heterogeneous mission that has like five or six disparate focuses. I mean we have over 900 facilities with people. I mean, we have about 8,500 facilities where there’s equipment. So, those missions, although they have aspects that are related, the actual business activities with the various parts of the Department are involved in are very, very dispersed and very different. And it’s dispersed nationwide.” (Former Dept. EA)
“Show me one large company that deals with multiple, multiple groups, multiple smaller industries, like our numerous constituencies. A large company that’s a conglomerate. I’d like to know how they’re implementing the EA. I don’t think you’ll find one. (Dept. Executive)
Finding 2: Framing and labeling of EA as an IT function instead of a strategic business tool reduces its perceived value to the organization, inhibiting program assimilation There are two contrasting ways of perceiving EA within the federal agencies. One perception is that EA is a technical function of the IT group – an activity that is being forced onto the agency to benefit technical requirements. An alternative perception is that EA is a strategic business tool ─ an activity that can benefit each bureau with improved performance. The EA leaders found themselves facing agencies that misperceived them 63
as being of benefit only to the IT department due to the wording of the Act, which emphasized the technical, regulatory aspect of EA. Associated with this technocratic perception of EA, a majority of respondents (20 out of 23) complained about negative labeling, such as being called “geeks” by superiors and co-workers. Compounding the perception problem seventeen respondents reported that their Programs fall under the CIO’s office, which, they observed, lacks power. Labeling theory, first introduced by Becker (1963), is concerned with how the self-identity and behavior of individuals may be determined or influenced by terms (usually derogatory) used to describe or classify them, and is associated with the concept of a self-fulfilling prophecy and stereotyping. FIGURE 3: EA Framing and Labeling Framing / Labeling “Both the Clinger-Cohen and the EA Government Act of 2002, referred to architecture, not enterprise architecture, actually the culprit is the 1996 Act which used the word “IT architecture.” (Former Agency EA)
“It (EA) was and is still viewed as a compliance exercise. Isn’t it IT anyway? It was not part of the strategic planning at the agency. IT geeks doing strategic planning? Please” (Former Agency Executive)
“Well, I think it (EA) was pretty ill-defined, or undefined in the (Clinger-Cohen) act. The Clinger-Cohen Act of ’96 established the role of the CIO in each agency, and that they were to develop an IT architecture (not enterprise architecture). Congress looked at it after five years, in 1991, and saw that nothing was happening, very few appointments, a few CIOs here and there, but nothing really serious.” (Former Agency CIO)
“The point is that Enterprise Architecture is about the business. It’s not about the IT as it (Clinger-Cohen Act) says. One of the reasons it failed is because at the Department and throughout the federal government, it’s still about the IT. They have not integrated the business with the IT. Really. They think of it as an IT thing, and it’s not. It’s a business thing.” (Agency EA)
“I made him (director) understand they (EAs) weren’t just IT architects, as stated in the (Clinger-Cohen) act; they were enterprise architects so they had to have some knowledge of and ability to use their craft in the business side of the house.” (Former Department CIO)
“for EA to be successful, and meaningful, it will take a new generation of strategic planners that are IT literate. It will not take IT people learning strategic planning.” (Agency Executive Officer)
“She (Agency Director) said, “the people that are in the EA program come up out of a system development culture and IT (sic) geek culture, so they know nothing about business. Why would I let them do strategic planning?” (Agency EA)
“During the annual decisions on IT initiatives, the planning and budgeting process, the EA was never called up to help make those budgeting decisions. It was used at CIO council meetings, however, when we’re talking about IT initiatives, but not about the budgeting decisions. So you have this disconnect between the IT, the technical side, and the business side (Retired Dept. EA)
64
Finding 3: A high resistance to change and low EA assimilation progress was reported by program leaders in agencies with strong political constituencies, while program leaders in agencies with less politically strong constituents reported less resistance to change EA program leaders at lower ranked agencies/departments were more apt than those at highly ranked agencies and departments to describe their organizations as culture dominant, characterized by a high resistance to change. A stovepipe structure, facilitated by strong political constituencies fueled turf wars. FIGURE 4: Agency Culture and Structure Dominant Culture Inefficiency & Wastage
Stove-pipe Structure
“I said to him (Director) “you're going to save $2 million, and he said “we've got an $80 billion budget, why would I be interested in saving $2 million? That's what dribbles off the table at the end of the day for God's sakes. Why do I care about that? You're asking me to change for piddling stuff like this?” (Agency EA)
“Some of these sub-agencies within the departments have enormous constituencies. They're very large; they're very powerful; and they have constituency on the (capitol) Hill that is substantial. So they don't have to listen very much to central headquarters. They listen just as much as they have to - only as much as they have to. So when I said, well, you know, things gotta change. They said, well, like hell they have to change. Who are you?” (retired Agency EA)
“Most agencies have more money than they know what to do with, really, all things being considered. Most of the people who are appointed here are dealing with, you can add three zeros on to anything they've ever dealt with before prior to coming to the federal government. And, add three zeros onto the number of people that they've ever supervised prior to coming to the federal government. So they have enormous money and enormous number of people. They're not used to managing those numbers. And in the federal sector, you get penalized for not spending budgetary allocations.” (Dept. EA)
“It's been a challenge to get it (EA) integrated in here. Never in our stove-pipe culture. If you think about it, if you're helping to make decisions, which means that somebody has to move aside at the table to let you there. Somebody has to give up a little bit of authority. That has been a real challenge because they (bureaus) want to continue to make decisions in the old way. As the Department Enterprise Architect, I don't have direct authority to tell the bureaus to do anything.” (Agency EA)
“Let's say across the federal government we decide that we're gonna get rid of 5 percent of the systems and we're gonna let one of the agencies run it and we're all gonna throw the data into a common data model for this particular function. And we're gonna throw them in the cloud (computing platform). We're gonna reduce costs by 80 percent. We're gonna throw the redundancy on somebody else's server. We're done, right? Guess what happens. No way, Jose. Those individual contractors who lost that business will have their congressmen writing letters very, very soon.” (Agency executive)
“what I see in government is it's not just EA but if you delve into any of the management areas in the federal government, they are so deep and so rich in their thinking and their concepts, but they're stove-piped..” (Agency executive) “..We run EA based on a scorecard approach. Do you have policies in place? Do you have a team? We've run it stovepipe organization by organization.” (Agency executive)
65
Finding 4: Program leader collaborative style and use of creative and aggressive tactics, business knowledge and marketing communications can moderate the effects of agency/department size, complexity and culture on EA program assimilation My data identified stark differences between the personal and professional characteristics of program leaders at the agencies and departments with high assimilationranking programs and those at the low assimilation ranking programs. These differences include management and leadership style, communication style and personal culture. Each of them is discussed below. Finding 4.1. All seven respondents from highly ranked programs evidenced a collaborative leadership style. In contrast, program leaders from the low ranking agencies reported lack of collaboration and talked of frustrating attempts at assimilation of the EA programs, meeting stiff resistance from the bureaus and business units. As the following quote from a former Chief Architect at one of the largest Federal departments illustrates, EA professionals who were successful in achieving high program assimilation recognized the imperative of cross-functional teamwork to effect results. “I did something at the Department that was kind of unusual in government I embraced all of our contractor people as real contributors and real members of the team, no different than anybody else as full partners in collaboration. And so we had contractors on our team, we had feds on our team and it was all one big team” (former Dept. EA) Program leaders from low ranking agencies on the other hand reported failure to effect collaboration. As the following quote from a former agency Chief Architect illustrates, EA professionals at some of the lower ranked agencies had to operate in uncooperative environments, typified by lack of teamwork and collaboration.
66
“..They (bureaus) were all operating independently. There was very, very little, horizontal coordination and no real collaboration.” (Former Agency EA) Finding 4.2. Program leaders at the highest ranked bureaus employed creative tactics such as publicity, reward and recognition schemes to draw participants and volunteers to their programs, in order to overcome manpower challenges. On the contrary, program leaders at the lowest ranked agencies primarily relied on resource deployment from within the agencies. One former EA executive recounted how he crafted an innovative scheme that provided career-enhancing exposure for his EA program’s volunteers by publishing their names and their organizations’ names in the media. Two former EA executives shared how they offered on the job training to volunteers from other agencies and departments, in the process teaching them new skills in Enterprise Architecture, in exchange for labor on the project. FIGURE 5: Reward Schemes Reward Schemes “I got all of the volunteer members of the nationwide EA program team a special recognition award from the deputy secretary signed by the deputy secretary and the CIO, you know, a big really nice walnut plaque with a big gold engraved plaque. And not only did I publish their names, but I published their organizations who gave them time to participate in the project, I published the organization’s names. And it helped their careers and helped them within their organizations to get advancement and better jobs and attention and all that.” (Former Dept EA)
“So I immediately started to leverage people from the agencies and put out a call, and said how would like to spend six months here at the agency and learn the trade of architecture while you’re doing it? We had, I don’t know, 30, 35 people who spent anywhere from three to six months with us, and a couple of them even stayed longer. They learned on-the-job training.” (Former Agency EA)
Finding 4.3. Program leaders at the top ranked agencies exploited their agencies’ vulnerabilities with guerilla tactics such as leaking information to the media and government audit bodies in order to stimulate action favorable to their programs. They also used threats to coerce agency leadership into supporting their programs. One agency 67
executive recalled how unfavorable press coverage made executives, who were political appointees, uncomfortable. Any hint of negative publicity or threat of punitive action from the OMB elicited swift action on the part of the agency leadership to blunt the effects. FIGURE 6: Examples of Guerilla Tactics Threats “I went to one of the IT reporters at the weekly newsletter. And I told him, “I want you to cover this and call out the enterprise architecture in the federal government. And I’m gonna feed you information and you’re gonna use it to publish these articles ‘cause you’re gonna raise people’s awareness and attention about it through the newsletter.” (Former Dept EA)
“and so I said to the director, “if we don’t establish these EA boards (with the bureaus), and start using them, there’s a good chance I’ll have to tell OMB that we’re not using the architecture. And we’ll stop being green (OMB performance rating). And it’s up to you.” (Agency EA)
“I talked to my friend in the inspector general’s office and said, “hey you know a lot of the goals that we set within the enterprise architecture program are not gonna be achieved because we’re not getting our funding anymore and we don’t have the support. So, we’re falling behind. So, you might wanna come audit us on this. Now, I didn’t do that in the open, I did that kind of clandestinely and it was through a friend that I knew worked in the IG’s office. And that’s in fact what happened. It was like calling an audit on myself.” (Former Dept EA)
“So I called the CIO and told him that I was concerned that if we did not take action quickly, that GAO would release this very negative report about the agency. So the CIO called the federal acquisition executive and told him that we had a crisis. The federal acquisition executive convened a meeting the next day to talk about this, and acted on all our EA requests. A few weeks later, the head of the GAO practice agreed to moderate the language in the report because of the commitment the agency had made to improve its architecture.” (Agency EA)
Finding 4.4. Program leaders at the top agencies had a business orientation and communicated primarily in business terminology. In contrast, the program leaders at the lowly ranked agencies mainly came from technical backgrounds and communicated mostly in technical terminology.
68
FIGURE 7: Language Use Communication “one of the things that I was unable to do is I could communicate. That’s one of my strengths. And so communicating and using non-IT terminology. As I said, I never used the word “architecture”. I talked about either business transformation, business performance improvement.” (Agency EA)
“I always communicated in business terms that resonated, I think, better than talking about enterprise service buses or consolidating data centers because, you know, business doesn’t care about that. That’s IT business. That’s IT stuff.” (Former Dept. EA)
“Yeah, it’s really helped that I didn’t come out of the IT business because I think I was better able to speak to the business side of the house as to the value that architecture can, if done properly, can bring.” (Dept. EA)
Finding 4.5. Program leaders at the top ranked agencies -- but none at low ranking agencies -- invested time and effort to marketing and dissemination of information about EA within and without their respective agencies. FIGURE 8: Marketing and Information Dissemination Marketing & Information Dissemination “One of the things that I did was I integrated a lot of marketing and a lot of high level standards. I mean, you look at the documents that we produced in the architecture program, those pamphlets and documents that we produced are better than things that you would buy in the store that had been highly edited. I mean the graphics we chose to put three color graphics inside the documents and we had full color covers at times.” (Former Dept. Chief Architect)
“I was also going out and visiting individual agencies and presenting and showing them best practices from my program at the department and from other things I’d learned through the federal architecture working group. And we really attempted to form sort of one of the early on communities of practice across government for the architects, many of whom were struggling like crazy to get their program afloat. And so the guidance documents helped them a lot, but also a lot of coaching that I did with a lot of the people.” (Former Dept. EA)
“When you try to move government in a particular direction, it takes a lot of repetition and constancy over time and then lots of conferences and speeches. I think I was giving close to 50-something speeches a year, and developing training curricula.” (Retired Dept. EA)
Discussion Intrigued by the wide variation in assimilation of EA within Federal agencies costing hundreds of millions of dollars since mandate promulgation in 1996, I set out to discover why. Informed by the literature, I began my inquiry persuaded that organizational characteristics including culture, structure and complexity might explain 69
disparities in EA assimilation. Institutional theory and assimilation of innovation theory, based on prior studies, strongly suggested that organizational characteristics, instead of individual players’ attributes or motives are determinant forces in innovation assimilation. My data, however, led me in another direction. While it should come as no surprise that these agencies were characterized by strong, complex, bureaucratic cultures, it was not organizational characteristics that explained EA assimilation success or failure. Instead, I discovered, quite surprisingly, the presence of quiet revolutionaries with unorthodox notions about how to overcome organizational impediments to assimilate EA. EA assimilation at the program level, I found, is a function of EA professionals’ behavioral strategies. So, although my findings provided general support for both classical and neo-classical institutional theory (in that the powerful effects of organizational characteristics were revealed to have been impediments to assimilation) it was the individual behaviors and strategies of EA professionals that determined whether EA programs progressed or stalled. My study demonstrated the powerful effects of framing and labeling by EA professionals in both groups. Whereas the EA program leaders were insiders, they were considered “outsiders” in most agencies. Using craft, courage, wit and bravery some EA executives succeeded in re-framing their disciplines – and themselves – antithetically in the minds of agency top management to re-align priorities, lend support and provide funding for their EA programs. The term “tempered radicals” refers to organizational insiders who use their leverage and knowledge of the organization to instigate change. Tempered radical strategy was described by Meyerson and Scully (1995) as “disruptive self-expression” because of its potential to shake up the status quo. Tempered radicals are often at odds 70
with the dominant culture of their organization, and operate on a fault line. My respondent EA leaders enacted a range of strategies including quiet resistance, acting as intercultural boundary spanners, leveraging small wins and collectivizing. Detailed comparative analyses of Meyerson and Scully’s tempered radical characteristics and my respondent strategies are presented in Appendices B and C. EA professionals who succeeded in boosting EA assimilation levels within their Federal agencies used maverick tactics and strategies to address perceptions of EA illegitimacy (Suchman, 1995; Aldrich & Fiol, 1994), fostered by the Clinger-Cohen Act’s framing of EA as “IT architecture”. A frame is defined as the packaging of an element of rhetoric in such a way as to encourage certain interpretations and to discourage others — that is a collection of anecdotes and stereotypes that eventually lead to bias (Goffman, 1974). Tversky and Kahneman (1981) showed that people generally prefer absolute certainty inherent in a positive framing-effect, which offers an assurance of gains. Positive framing effects (associated with risk aversion) result from presentation of options as sure (or absolute) gains and negative framing effects (associated with greater risk tolerance) result from options presented as the relative likelihood of losses. Introduction of EA was intended to instill a “business-like” operating model into the not-for-profit federal agencies. Recognizing the stigma associated with IT, some EA leaders positioned themselves as business rather than technical consultants; i.e. strategic partners rather than service providers. Figure 9 below captures the radical strategies employed by the EA leaders in overcoming the numerous institutional impediments to their Programs’ progress. Starved of critical skilled manpower and financial resources, they exploited the lack of motivation and incentives within the federal government (see 71
Appendix C). Meyerson (2003) observed that sometimes action as a starting point can trigger a wide range of outcomes and further actions. Working in concert with media outlets, they published names and posted photos of volunteers to their “successful” EA projects, a strategy that drew admiration of peers. Meyerson (2003) wrote that small wins include seemingly minute projects that eventually result in concrete, measurable, and visible progress in the process of organizational transformation. Meyerson and Scully (1995) also observed that sometimes the small win becomes an impetus for further and broader action down the line. EA leaders associated with high assimilation level programs also launched creative schemes such as holding high profile ceremonies at which plaques and certificates, signed by federal executives were awarded to volunteers. This drew much needed resources to their programs. In addition to promising career advancement for volunteers to their programs, they used their informal networks to recommend team players to executives, opening opportunities for them in the process (see Figure 9).
72
FIGURE 9: EA Leader Tempered Radical Strategy
Successful EAs’ attitude and risk taking tactics extended to printing their own business cards with titles such as ‘Chief Architect;” titles that were unknown or unheard of within the federal space: “My management wouldn’t assign me the role of chief architect, but I sort of just started calling myself that. And I made up my own business cards. I told all my fellow architects across the government, I said “you know if you’re the person in your agency that’s responsible for developing and promoting and maintaining the enterprise architecture, then you are the chief architect. So, get yourself some cards and print ‘em up and put that title on there.” And so, you know I said, “the worst thing that could happen to you is if somebody were to say to stop giving ‘em out” (retired EA) An interesting aspect of the tempered radical leadership style was applying counter-culture (Roszak, 1968; Nelson, 1989; McKay, 1996, Meyerson, 2003) in combating powerlessness (Kanter, 1979). In most departments and agencies, there was 73
concentration of power within the bureaus and business units. Hierarchical management structure, Turf wars, stovepipe culture and lack of collaboration characterized the prevailing atmosphere within the organizations. Meyerson (2003) wrote “to think in terms of negotiation is to think in terms of competing interests, differing positions and concerns, distinct sources of influence, and alternative framing of issues. Negotiating requires discipline and action: people must participate in shaping how problems unfold” (p. 79). Defying laid down organizational structures; the EA leaders established a shared governance model, co-opting leadership from within the bureaus and business units’ ranks. Not only did they establish a “one-man, one-vote” system based on representation from each bureau or business unit, they also granted veto power and instituted team selfgovernance, such as voting out errant members. Despite the fact that these “boards” and committees were not sanctioned by the federal agencies and departments, they metamorphosed into a strong vehicle for a successful bottom-up approach to the assimilation of the EA programs. One astonishing finding was that despite being the de facto leaders of the boards, the program leaders exercised a democratic leadership style, characterized by a servant leadership model (Greenleaf, 1977; Russell, 2002). Two tactics that the program leaders roundly succeeded in employing were “pushing up the urgency level” (Kotter, 1996) and threats. The program leaders developed a very effective strategy of manufacturing crises in order to trigger action from their management. Some of the strategies included: instigation of impromptu audits by the GAO and OMB; leaking of information to the press or to the twin federal audit bodies, in a bid to expose and embarrass their top leadership into action. By forging 74
strong relationships with exogenous forces within and without the federal government, the EA leaders placed themselves in a strong position to influence budgetary decisions and also heighten the priority of EA within their respective organizations. Recognizing their role as change agents, the program leaders also fashioned themselves as innovation brokers (DiMaggio, 1992), bridging the knowledge gap between the executive and the technical arms of the agencies (Rogers, 1983). By their morphogenic behavior, the EAs were able to recast themselves as consultants (Attewell, 1992), creating a multi-lingual sub-culture in the process, where they could engage with their executive and business unit leadership in “business-speak” and communicate technical jargon with the technical side of the agencies. Their understanding of their leaderships’ political agenda helped in furthering their programs’ causes. For example: “If you said you were going to save a little money; that was not very important to them (Department executives) at all. But if you said you're going enable freeing up more money for the low-income citizens or help farmers or catch products coming across the border that could be really harmful to citizens of the United States, they understood that”(Department EA) By coating communication in familiar themes and goals, the EAs were able to engage the department and agency executives in meaningful discussions that led to better cognition of the role of EA within the agencies (DiMaggio, 1992; Weick, 1979; Levitt & March, 1988). This was an important development because EA, as an innovation, had to be presented in the language of existing institutions by giving it appearance of familiar ideas (McKinley, Mone, & Moon, 1999). Meyerson (2003) argued that it is imperative to proactively transform problems and constantly reframe meaning, and leverage prudent
75
negotiation strategies to help transform issues and open avenues for constructive (collective) action The EAs also embarked on ambitious marketing drives, preparing pamphlets and flyers touting their programs’ goals and achievements. This not only caught the attention of agency executives, but also external decision makers within the federal government. This had the dual effect of network expansion as well as information dissemination. FIGURE 10: Revised Conceptual Model
Conclusion The study suggests that the effects of innovation framing may influence EA program assimilation. The model conjectures business or technology framing of EA as an innovation may influence how and to what extent an EA program assimilates. The study also suggests that the impact of framing on program diffusion is mediated by low value perception and unfavorable labeling. The powerful effects of organizational 76
characteristics moderate the relationship between framing and the mediating factors of labeling and low value perception. The relationship between the mediating variables of low value perception and labeling are moderated by program leader creative and aggressive (guerilla-like) behavior. This suggests that organizational inertia can be overcome by the tactical action of program leaders. It should be pointed out however; guerilla tactics can be a problem and self-defeating over time. They are at best limited time tactics. Implications for Practice and Future Research This study has important implications for government leaders and officials as well as federal EA program practitioners. It also has implications for the practice of enterprise architecture within the public and private sectors –not just in the U.S., but internationally. The manner in which an innovation such as EA is introduced and implemented plays a critical role in its assimilation success or failure (Feder, 1996). Reform diffusion within both the public and private sector can be a highly complex undertaking. Likewise, innovation diffusion continues to draw much scholarly interest, given the rapid rate of promulgation of administrative innovations and rapid adoption of technological innovations. My research focused on the agencies covered under the 1996 Clinger-Cohen Act mandate without differentiating them based on various characteristics and factors. For instance agencies that were not directly under the legislative arm of government could defy the GAO with impunity. Future research should aim to conduct comparative studies of EA program assimilation between agencies and bureaus under each of the three arms of government. Researchers should also give attention to the extent to which formal
77
and informal communication channels affect federal employees’ attitudes toward mandated administrative innovations such as Enterprise Architecture. Limitations of the First Study Several limitations of this qualitative research should be noted. My sample size was small and included only two or three respondents from each of the departments and agencies. Respondents were not randomly selected. They were purposely selected for based on their involvement with agency EA programs. It is possible that including a broader representation of executives from each organization may have revealed different attitudes and beliefs. Including organizations that had voluntarily implemented EA (preferably prior to the 1996 mandate) may have also produced different results. I asked respondents to recall past events sometimes dating back several years and I recognize that memory can be affected by time. I also acknowledge the possibility that some respondents, especially EA professionals, may have been motivated to present themselves in a positive light and to provide “socially desirable” responses. As such I made conscious efforts at every step within the design and conduct of the research to minimize researcher bias, given the principal researcher’s 10 plus years of experience in IT, familiarity with, and/or professional acquaintance with some of the organizations included in the study.
78
CHAPTER IV: INVESTIGATING THE ANTECEDENTS OF ENTERPRISE ARCHITECTURE ASSIMILATION AT ADOPTER UNIT LEVEL: CROSSSECTIONAL QUANTITATIVE STUDY OF U.S. FEDERAL AGENCIES Enterprise Architecture (EA) has over the past two decades emerged as a significant strategic planning approach deployed by Information Technology (IT) management professionals. It is used to translate “an enterprise’s business vision and strategy into effective enterprise change” (Buchanan & Soley, 2002). EA can be viewed both as a management tool and innovation that necessitates an intervention in how the organization utilizes its IT assets and aligns them with business mission and goals. EA programs utilize abstract modeling frameworks to identify and represent interactions between people, process and IT assets within an enterprise. These identified (present or missing) critical relationships are then used to improve cost efficiency, effectiveness and strategic alignment of IT with the organization’s mission (Ross et al., 2006). As a management tool, EA provides guidelines for IT governance and processes that help determine resource alignments such as develop standardized policies, enhance decisions related to IT and business development, and coordinate development activities (Bernard, 2005). Overall, EA can be viewed as a radical administrative innovation due to its novelty, complexity and originality (Damanpour, 1988). Therefore, EA has been viewed as critical task element for information systems executives for some time (Niederman, Brancheau & Wetherbe, 1991). Enterprise architecture is a misunderstood concept among general management literature; because it is perceived only to be an IT, rather than a management issue (Zachman, 1997). Alas, this perception has severely diminished its assimilation and business impact. For example, despite the 1996 government mandate, the 2006 GAO 79
audit found that 48% of federal agencies were still at very early level of EA assimilation, 36% at medium level, 11% at high level , while only 5% at highest levels (Rico, 2006). Similarly, a study by MIT Sloan Center for Information SystemsResearch (CISR) among 456 private companies in North America and Europe showed that 12% of the EA programs were at early levels of assimilation, 48% at medium levels , 34% at high level and only 6% at the highest level . The economic impacts of EA have likewise been humble. In June 2008, the Office of Management and Budget (OMB) reported that 413 IT projects within Federal government—totaling at least $25.2 billion in expenditures for fiscal year 2008 were poorly planned or poorly performing (Grasso, 2009). Likewise GAO’s2009 audit report shows that 585 of the 810 major IT projects (about 72%) faced imminent failure putting the government and taxpayers at risk of losing $26.9 billion (Grasso, 2009). Being a radical administrative innovation, an EA program typically progresses over long periods of time from its initial inception and assimilation to increasingly pervasive levels of assimilation. In EA research these progressive assimilation levels have been denoted as ‘maturity levels’ (Ross et al., 2006). In this paper, I prefer to use the term assimilation in line with extant innovation research. I define EA assimilation herein as the depth of EA innovation integration and its pervasiveness in the organization (Ash 1997)3. This, corresponds with, Sambamurthy and Zmud’s (1996) definition of assimilation as “the success achieved by firms in utilizing the capabilities of IT (in this case EA) to enhance their business performance” and Zmud and Apple’s (1992) definition of innovation infusion as the “comprehensiveness or sophistication of use of an 3
She defines it as “the extent to which the full potential of the innovation has been embedded within an organization's operational or managerial work systems.” (Ash, 1997: 12 )
80
innovation.” Hence, according to my definition reaching distinct, higher EA assimilation levels demonstrates the increase in the extent, the depth and comprehensiveness of the deployment of EA principles and policies suggesting broader and different effects within an organization (Ross et al., 2006).Therefore, organizations can seek to realize EA innovation’s full potential only by assimilating constantly, and over time, distinct innovations related to EA (Wang, 2008). Though significant research has been carried out on EA- especially on its nature, principles and effect- less is known about its assimilation drivers i.e. what factors promote the achievement of different assimilation levels. In particular, questions, which have so far not been adequately addressed, deal with the following issues: 1) How to measure and distinguish different levels of EA assimilation? 2) What factors influence the attainment of different assimilation levels? Failures associated with assimilation of IT innovations have been high profile and in some cases financially ruinous for organizations (Miller, 2000; Xue et al., 2005). Ironically, business value of IT innovations including that of EA can only be realized through successful assimilation (Armstrong & Sambamurthy, 1999). Just like any other complex innovation, EA implementation has a major impact on the organizations’ culture, people and processes, requiring not only huge financial investment but also involvement and commitment from the highest levels in the company (Schekkerman, 2006: 32). It is for this reason that I have elected to study EA in assimilation in organizations. At the same time the topic has attracted only a handful of case studies (Ross et al., 2006; Saha, 2004; Schekkerman, 2006) and we face the non-existence of systematic theory based field studies. I could only locate a single study that had focused 81
on external factors affecting levels of IT assimilation in the context of ERP systems (Liang et al., 2007).Their study found that mimetic pressures had a positive effect on top management beliefs, which eventually had a corresponding positive effect on the organization’s top management participation in the ERP assimilation process. In this paper, I seek to develop a model of EA assimilation to address research questions (1) and (2). EA assimilation is viewed as a radical administrative innovation. Accordingly, I will use as the unit of analysis a progressive EA assimilation process within an organization. The model draws upon assimilation of innovation theories and neo-institutional theory (Dimaggio & Powell, 1983 & Tolbert & Zucker, 1983) .The use of institutional theory emphasizes the presence of “coercive pressure” as a key positive moderator affecting EA assimilation. To validate the proposed model I will investigate EA assimilation within 123 federal agencies during the period of 1999-2008. The EA in Federal government was initiated by the congressional passage of the Clinger –Cohen act of 1996 as a mandate for all federal agencies. Coercive pressure was later instituted through a series of executive interventions by the White House and Congress - such as the 2001 President’s Management Agenda, and the 2002 e-Government act respectively. Therefore a study of Federal Agencies’ assimilation of EA offers an opportune context to study the effects of both organizational level and institutional factors in shaping the EA assimilation. By addressing the two research questions through an empirical study I am able to: (1) identify significant determinants of EA assimilation, and (2) examine the influence of coercive pressure in driving EA assimilation.
82
The remainder of this study is organized as follows. The first section reviews literature on Enterprise Architecture and conceptualizes Enterprise Architecture assimilation concept and discusses factors that affect it. Section 2 outlines the conceptual model of EA assimilation and formulates a set of hypotheses to validate the effects of institutional and organizational factors on EA assimilation. Section 3 introduces the study context, discusses research methodology, and outlines data analysis strategy. The final section conducts the analysis, discusses the results and notes implications for future research Theoretical Framework I will next conduct a literature review related to: (1) the concept of Enterprise Architecture; (2) Assimilation of innovations and factors affecting the assimilation; and (3) Institutional theory and how insitutional factors can be expected to affect EA assimilation Enterprise Architecture Enterprise Architecture in Organizations The idea of enterprise architecture dates back to the mid1980’s. At that time John Zachman, while working at IBM, noted the need to use the idea of a logical construction blueprint (i.e. architecture) as a paragon to define and control the integration of organizational IT systems and their components (Bernard, 2005). Zachman consequently developed a framework that defined key elements of this ‘architecture’, commonly referred to as the “Zachman Framework” (Saha, 2004). Since then, several enterprise architecture frameworks have emerged, including The Open Group Architecture Framework (The Open Group, 1999), the Federal Enterprise Architecture Framework 83
(FEAF-CIO Council, 1999)), the EA3 cube framework (Bernard, 2005), the Department of Defense Architecture Framework (Wisnosky & Vogel, 2004), just to mention a few. Hobbs (2006) defines an enterprise as an “organization performing functions of some scope, complication or risk” i.e. they are highly complex, dynamic and in a state of constant change. In this context architecture forms “a structure or the practice of designing structures”. EA is thereby in its simplest form systematic application of architectural fundamentals to manage the complexity of enterprises (Kappelman, 2010) and EA presents the ‘best practices’ to manage the complications, scope and risk by “bridging the chasm” between business strategy and IT implementation, or between business and IT activities (Kappelman, 2010). Hence EA ‘discipline’ is the “best way humans have invented for managing complexity and change” (Kappelman, 2010: 28) as it “enables managers to manage ubiquitous change and increasing complexity within the strategic and tactical environments their organizations operate” (Kappelman, 2010: 29). In general, EA is an overarching term for enterprise level strategic and operational alignment activity i.e. it covers most of the work that is needed in an organization to align IT assets with its business goals and needs (Nunn, 2007). EA is therefore a “management program” to do the alignment and a documentation method to describe the assets and their relationships. If well done it can provide an actionable, coordinated view of an enterprise’s strategic direction, business services, information flows and resource utilization” (Bernard, 2005: 34). Viewed as a management program, EA provides overall governance and processes that determine resource alignments, develop standardized policies, enhance decision support and oversee development activities (Bernard, 2005: 35). At the strategic level, EA defines how an organization’s core processes, core data 84
and core technologies work together to drive business value in accordance with the Operating Model (Ross et al., 2006). A workable operating model guides the desired level of business process integration and business process standardization for delivering goods and services (Ross et al, 2006). An EA program is supported by frameworks, methods, models and techniques, which help coordinate many critical facets that make up an enterprise (Schekkerman, 2006). Enterprise architects use various business methods and tools to understand and document the structure of an enterprise. In doing so, they produce models which are often broken down into four areas: business, applications, information and technology (Saha, 2004). Hence, viewed as a modeling methodology, EA models five domains across the enterprise: 1) business architecture (a model and definition of how a business unit or significant business process operates); 2) information/data architecture (a guide for the use of data resources); 3) applications architecture (the architecture of an individual application); 4) integration architecture (the design and implementation of methods and solutions to achieve application interoperability) and 5) infrastructure architecture (a guide to selection and configuration of hardware, software and communications components for specific domains (Kappelman, 2010). EA Assimilation Because EA forms an integrative and radical transformation how an organization views and deploys its IT assets and capabilities, EA programs progress through assimilation levels. These levels capture the depth and scope of deploying EA principles and integrating related behaviors into organizational routine. Increased assimilation is viewed as critical to an EA program’s effectiveness (Bittler & Kreizman, 2005). An MIT 85
Center for Information Systems Research (CISR) study on EA related budgetary patterns shows, for example, that IT budgets for companies at higher levels of EA assimilation (called assimilation by Ross et al., 2006) are at 85% of firms at earlier levels of assimilation, thus indicating higher efficiency in usage of IT assets. In contrast, IT budgets of companies at later levels of EA assimilation are at75% of budgets for those firms within early EA assimilation levels. This may be an indication that transformatory effects start to affect the deployment of IT assets as firms move through assimilation levels (Ross et al., 2006). Hence, each EA assimilation level involves additional elements of “organizational learning” about how to apply IT assets and principles and how to harness business process discipline with strategic capabilities (Ross et al., 2006). There are several widely published models of EA assimilation including National Association of State Chief Information Officers Enterprise Architecture Assimilation Model (NASCIO, 1999), the Government Accountability Office EAMMF (CIO Council, 1999), the Extended Enterprise Architecture Maturity Model (E2AMM) by the Institute For Enterprise Architecture Developments (Schekkerman, 2005). I will next review the MIT and GAO models, because they are the most well-known and widely adopted in characterizing EA assimilation levels. MIT Assimilation Model The MIT CISR assimilation model is a holistic model of drivers and levels of EA assimilation. It incorporates both changes in capacity- i.e. IT assets- and capability building – i.e. ways of deploying IT assets for strategic ends. These changes characterize distinct learning levels and applications of EA principles and techniques as part of the assimilation process offering new value benefits at each assimilation level (Ross et al, 86
2006). The levels are strictly ordered so that companies are viewed to undergo four progressive levels in terms of how to apply the EA principles. These four levels have been derived based on several case studies in the private sector. These levels are called: 1) business silos, 2) standardized technology, 3) optimized core and 4) business modularity (Table 11- Ross et al., 2006). TABLE 11: Learning Requirements of EA Levels (Ross, Weill & Robertson 2006) Business silos IT Capability
Local IT applications
Business Objectives
ROI of local business initiatives Individual applications
Standardized technology Shared technical platforms
Reduced IT costs
Key management capability
Technology enabled change management
Who defines applications
Local business leaders
Shared infrastructure services Design and update of standards; funding shared services IT and business unit leaders
Key IT governance issues
Measuring and communicating value
Establishing local/regional/global responsibilities
Strategic implications
Local/functional optimization
IT efficiency
Funding priorities
Optimized core Companywide standardized processes or data Cost and quality of business operations Enterprise applications Core enterprise process definition and measurement Senior management and process leaders Aligning project priorities with architecture objectives Business operational efficiency
Business Modularity Plug-and-play business process modules Speed to market strategic agility Reusable business process components Management of reusable business processes IT business and industry leaders
Defining, sourcing, and funding business modules Strategic agility
As shown in Table 11, as the company advances through the four levels, its IT strategy takes on increased importance while the business value of IT increases (Ross et al., 2006). Ross et al. (2006) also argue that EA capability assessment becomes critical to 87
establishing the organization’s ability to implement new technologies and innovations that are pivotal for its strategic effects. At assimilation level 1, companies focus their IT investments on delivering solutions for local business problems and opportunities. The role of IT is to automate specific business processes. IT investments are justified on the basis of cost reduction. At level 2, companies shift some of their IT investments from local applications to shared infrastructure. In this level companies establish technology standards intended to decrease the number of IT platforms they manage. At level 3, companies move from a local view of data and applications to en enterprise view. The role of IT at this level is to facilitate achievement of company objectives by building reusable data and business process platforms. At level 4, management refines and increasingly modularizes the processes that were digitized in the third level. The role of IT is to provide seamless linkages between business process modules. The GAO Assimilation Framework The GAO assimilation framework is focused on building and measuring the organization’s capability to develop and use an EA (CIO Council, 1999). Capability is defined as the organization’s ability in managing IT assets throughout the value chain and developing core competencies (Weill & Ross, 2003). According to the Innovation Value Institute (IVI, 2005), IT capability is typically measured using an assimilation framework and provides a concise management roadmap to optimize business value derived from IT investments. Most capability maturity frameworks (such as the Carnegie Mellon University Software Engineering Institute’s Capability Maturity Framework Integration (CMMI)) consist of a five-level maturity model that is used to organize and structure the 88
framework for mapping IT improvement efforts. Thus, the GAO assimilation framework just like the CMMI is a maturity framework defining assimilation and performance benchmarks for EA programs in Federal government (Schekkerman, 2005). A Framework integrates the taxonomy of the viewpoints and methods that are necessary to satisfy the needs or concerns of the EA stakeholders (Saha, 2004). The purpose of an EA Framework is to define the information needed to describe an EA and/or the methods for governing, creating and exploiting the EA (Bernard, 2005). The continuous approach uses in contrast to the staged approach, Capability Levels (CL) for describing the state of improvement. The difference between assimilation and capability levels is that a capability level only classifies the ability of an organization within a certain Key Performance Area (KPA), e.g. IT security or maintenance of EA deliverables, whereas an assimilation level classifies the overall ability of an enterprise level process, e.g. EA management or software development. Thus, an assimilation level is derived from the capability levels of the KPAs. The section below details the capability levels of the two main assimilation models that are built around capability frameworks. According to Hite (2005) at level zero and one either an enterprise does not have a plan to develop and use enterprise architecture, or it has plans that do not demonstrate an awareness of the value of having and using enterprise architecture. Hence there is no assimilation. At level two, an enterprise recognizes that EA is a corporate capability by vesting accountability for it in an executive body that represents the enterprise. At level three, an enterprise focuses on developing architecture products according to the selected framework, methodology, tool, and established management plans. At level four, an enterprise has completed its EA product, implying that the architectures and plans have 89
been approved by the EA steering committee (established at level 2), or an investment review board. At level five, an enterprise has secured senior leadership approval of the EA plans and a written institutional policy stating that IT investments must comply with the architecture, unless granted an explicit compliance waiver. At level five, the enterprise also tracks and measures EA benefits and adjustments are continuously made to both the EA management process and the EA plans (Saha, 2006) Overall the GAO framework has 31 elements (which I will use subsequently to measure assimilation of EA over time (see Appendix E).These elements are indicators of different EA assimilation levels and their values can be mapped to distinct assimilation levels (CIO Council, 1999). The elements are also cumulative and incrementally build on capabilities necessary for EA assimilation progression. The GAO framework has 12 elements or Key Performance Areas, which define requirements to objects. These requirements can be found at the levels 2, 3, and 4 within the critical success attribute of demonstrating the satisfaction of commitment. Please see Figure 11 below for the details on the core elements or KPAs and associated attributes. These KPAs are cumulative as can be seen from Figure 11 below. For instance as can be seen below, the level 1 assimilation level has a single element.
90
FIGURE 11: GAO Assimilation Framework Elements and Attributes
91
As a rule of thumb, all aspects of a particular element must be satisfied in order to lay a claim that the agency has met the requirements and can move to a higher level. Every reached assimilation level is the foundation for the next level and thus cannot be omitted. Partial satisfaction would only lead to a “partially satisfied” score, and if none of the elements are satisfied, the organization is rated as not having satisfied the element requirements (See Appendix N for details).Therefore, in the GAO framework, all organizations are considered initially as being at level 1. Even an organization that satisfies several elements within higher assimilation levels without first fulfilling all requirements for all elements of level 1 assimilation will still be ranked as being at assimilation level 1. For instance, an agency that satisfies all elements required for level 4 assimilation ranking, but has yet to satisfy a particular element required at assimilation level 1 is still considered to be at assimilation level 1. Comparing MIT and GAO Frameworks The two assimilation frameworks share a lot in common in terms of activities and objectives. In particular, both assimilation frameworks 1) support capability development; 2) support capacity development; 3) support process improvements; 4) provide EA effectiveness benchmarks; 5) illustrate benefits (current and projected), and 6) set goals for the next assimilation level. For instance the MIT framework’s level 3 objectives are aligned with the GAO assimilation level 5 objectives. The MIT assimilation framework level 2 activities and objectives are aligned with the GAO assimilation framework’s level three activities and objectives. MIT CISR assimilation level 4 objectives focus on modularity (autonomy) in investment decision making and alignment within business unit impact. This closely matches the objectives of GAO 92
framework level 5 objectives. In terms of activities the MIT framework level 1 activities and objectives correspond to those in level 2 within the GAO framework. The GAO framework level 4 activities correspond to the MIT assimilation level 2. A synthesis of the commonalities and differences are shown in Table 12 below TABLE 12: Comparative Analysis of MIT Assimilation Framework and the GAO EAMMF Level 1
2
3
4
5
MIT CISR
GAO EAMMF
Business Silos: IT Investments focused on solution delivery for local business problems and opportunities. Role of IT is to automate specific business processes. IT investments justified on basis of cost reductions and ROI. Standardized technology: IT Investments shift from local applications to shared infrastructure. Management of technology standards is key to this level. Investments justified on the basis of cost reductions and compatibility. Optimized Core: IT investments shift from local applications and shared infrastructure to enterprise systems and shared data. Investment justified on the basis of digital options, standardized processes and flexibility. Business Modularity: IT investments shift towards achieving strategic agility through customized or reusable modules. Seamless linkages between business process modules and greater autonomy/discretion at business unit level for building or buying modules. Investment justified on the basis of business impact and digital options.
Creating awareness: Either an enterprise does not have plans to develop and use architecture, or it has plans that do not demonstrate an awareness of the value of having and using architecture. Building EA foundation: An enterprise recognizes that the EA is a corporate asset by vesting accountability for it in an executive body that represents the entire enterprise
Developing EA products: An enterprise focuses on developing architecture products according to the selected framework, methodology, tool, and established management plans. Completing EA products: An enterprise has completed its EA product, meaning that the products have been approved by the EA steering committee (established at level 2) or an investment review board, and by the CIO.
Leveraging EA to manage change: An enterprise has secured senior leadership approval of the EA products and a written institutional policy stating that IT investments must comply with the architecture, unless granted an explicit compliance waiver. Also at level five, the enterprise tracks and measures EA benefits or return on investment, and adjustments are continuously made to both the EA management process and the EA products
93
Both frameworks advocate that a strong new foundation is laid at every assimilation level prior to embarking on the next level. Ross et al (2006) therefore argue that the fulfillment of all requirements of one assimilation level (capabilities) is in effect foundation building for the organization’s capacity to fulfill requirements for the next assimilation level. Likewise, The GAO assimilation framework explicitly requires that agencies demonstrate capacity to effectively execute on elements within one assimilation level prior to engaging in higher level EA assimilation activities. Ross et al. (2006: 86) book’s third chapter is titled therefore “learning takes time – don’t skip levels” as to warn of the risks associated with skipping over levels. They argue that learning through the architecture levels encompasses gradual capability development in technology and business processes, investment evaluation leading to IT enabled change and process design. This logic aligns with the GAO directives deterring organizations from pursuing elements in higher assimilation levels before satisfying foundational elements at the lower levels. The Ross et al. (2006) capability development pattern is also evident within the GAO assimilation framework element taxonomy. For example, a particular assimilation level 2 element is setting up of an implementation committee as a prerequisite to the setting up of the EA program office. This is then followed by the appointment of a Chief Enterprise Architect. This systematic chain of events shows a clear progression path of laying the foundation for the subsequent assimilation level such as architectural design and enterprise technology selection. For instance, completion of all assimilation level three activities lays the foundation for embarking on assimilation level 4 activities. Similarly, satisfaction of assimilation level 4 requirements not only creates the foundation 94
for alignment with management objectives, but also paves way for executive level governance policies, driving change and aligning investments with strategic objectives. Ross et al. (2006: 88) write that: “The objective of the enterprise architecture is not so much to achieve a particular end state as it is to recognize what direction the company is going. Transitioning through the architecture levels allows companies to rack up benefits.” Institutional Theory and Innovation Assimilation Institutional theory defines institutions as “social constructions” made up of symbolic elements which include representational, constitutive and normative rule systems that influence how organizations and their members respond to environmental pressure (DiMaggio & Powell, 1983). The theory identifies three basic types of institutional pressure: coercive, mimetic, and normative, which reflect three analytically distinct processes of institutionalization (DiMaggio & Powell, 1983). Institutional studies posit that firms obtain legitimacy by conforming to dominant practices i.e. pressures within their institutional field and thus seeking to create isomorphism with their environment (Tolbert & Zucker, 1983; Scott, 1992). Coercive isomorphism is an attempt by an organization to conform to pressure from institutions on which they are resource dependent. The most common form of coercive pressure is from government regulations, policies and funding decisions (Mezias, 1990; Tolbert & Zucker, 1983). Mimetic isomorphism arises from organizations responding to uncertainty by mimicking other organizations either in their industry or in the market. In mimetic isomorphism, organizations emulate themselves after organizations perceived as successful or “legitimate” (Tolbert & Zucker, 1983). This has also been commonly referred to as the 95
“bandwagon effect” (Staw & Epstein, 2000). Normative isomorphism is caused by institutional conditions associated with establishment of etiquettes; a “code of conduct”; “code of dressing” which are eventually institutionalized, inculcated and imbibed by the institution membership (DiMaggio & Powell, 1983). This is usually the product of orientation, disposition and control mechanisms associated with social and professional networks, professional associations and membership clubs. Thus “coercive pressure” from an institutional perspective suggests that firms or organizational units are susceptible to compliance pressures for conformity and the need for legitimacy (DiMaggio & Powell, 1983; Meyer & Rowan., 1975). Several studies have established that coercive pressure is a critical factor affecting the assimilation of radical innovations (Tolbert & Zucker, 1983). I build on this foundation while developing below a model that identifies organizational conditions under which coercive pressure may drive assimilation. More specifically I maintain that coercive pressure will either have a dampening or accelerating impact on the relationships between the antecedents and levels of EA assimilation depending on the nature of the relationship. Research Model and Hypotheses Overview of the Research Model Drawing upon assimilation of innovation and institutional theories I propose next a model that incorporates several common factors examined in the current IS/IT literature affecting innovation assimilation (Fichman, 2000; Liang et al., 2007; Wang, 2008). The research model is grounded on the concept of assimilation levels and has been formulated with an eye to detect significant organizational factors that influence specifically innovation assimilation. The model posits that: organizational complexity, 96
organization scope, parochialism and cultural resistance and management value perception (see Figure 12) are major factors affecting EA assimilation. Appendix D provides a summary of literature references utilized for all constructs. The research model (Figure 12) also includes some second order constructs access to resources, organization complexity and scope- along with some first order constructs, - management value recognition, as well as parochialism and cultural resistance. I thus surmise that these organizational factors capture the main causes affecting EA assimilation as identified in the past innovation (Fichman, 2000; Zhu et al., 2006; Liang et al., 2007; Wang, 2008). A description of each construct is outlined in Table 13 below. FIGURE 12: Research Conceptual Model
97
TABLE 13: The Study Constructs Constructs Access to resources
Organization Complexity Organization Scope Parochialism and cultural resistance Top management value recognition Control variable Size
Definition The extent to which the organization is capable of securing adequate resources including human capital and money to successfully drive assimilation. The amount of differentiation among different elements constituting the organization for example the number of missions, functional units and the number of hregulations, as well as their mutual dependencies The extent and reach of operations affecting the organizations that reside outside its formal boundary. The variance of and attitude towards perceptions of changes in the environment often denoted as the “selfish pettiness or narrowness” with regard to external interests, opinions or views The perception of the value of an innovation by the management team
The number of employees in the organization
As the model shows, I also posit that two moderating factors influence the effects of the organizational antecedents: 1) access to resources, and 2) coercive pressure. In general I posit fourteen hypotheses associated with the effects of identified factors on EA assimilation levels. As I develop next the formal hypotheses related to the model, I will provide further justification for the logic how the antecedents affect the level of EA assimilation. Therefore, I will next review the content of each construct and reasons for its inclusion in the model and state associated hypotheses. Access to Resources I follow Higgins’ (2005) extended definition of resources to include funding for research & development, technology and human capital with knowledge on the innovation to enable assimilation. The access to resources factor is a multidimensional second order construct with two first order constructs: funding and skilled resources. I will use this construct as both as an antecedent to EA assimilation as well as a moderator 98
on the relationship between other antecedents and EA assimilation. I use the construct as an organization level construct to cover aspects such as IT budgets and resource allocations that can have a direct effect on assimilation. I will next characterize the content of the two components of access to resources: funding and skilled resources. Funding. Funding is defined as the easy accessibility or ready availability of financial resources. This definition is in line with Fichman and Kemerer (1999) and Zhu et al., (2006b). My argument is that an organization’s ability or inability to secure funding has causal effect on its access to vital resources for EA assimilation. Several studies have shown evidence that lack of or abundance of financial resources is linked to the overall picture of the resources that organizations have at their disposal for assimilation of innovations (Fichman & Kemerer, 1999). For instance Downs and Mohr (1976) found that financially strong organizations are particularly well positioned to adopt high cost innovations, which subsequently translated into ability to summon auxiliary resources for its assimilation. An organization’s financial capacity, whether weak or strong has also been shown to strongly influence its access to vital resources for investments in critical technical skills such as consulting and advisory services (Attewell, 1992). Other studies that have focused on the effect of funding have equally shown, for instance that large IT budgets, which are associated with ease of access to funding, are positively linked to greater IT benefits realization (Salmans, Kappelman, & Pavur, 2009), an argument earlier proffered by Katz and Shapiro (1986). However, as stated earlier in this paper, the realization of IT benefits is only possible through its assimilation and usage (Armstrong & Sambamurthy, 1999), meaning that the funding must be readily available throughout the lifecycle. This argument was further advanced by Rogers (1991) 99
who also found that easy access to funding, which enables the engagement of consulting, advisory and service firms can help lower knowledge barriers for the focal innovation, an important aspect of its assimilation. King et al. (1994) and Rogers (1995) separately established that complex technology based innovations require easy and ready access to financial resources in the form of strong financial sponsorship or outright funding subsidies for their diffusion. Based on the above arguments, I therefore argue that funding has a causal link to the access to critical resources construct. Skilled resources. Skilled resources is defined as the availability of human capital with the technical, political and managerial know-how necessary for bridging the knowledge gaps between the executive and the technical arms of the organization (Rogers, 1983). Technical skills in particular are not only an important prerequisite for successful IS adoption (Kwon & Zmud, 1987), my argument goes further to link their availability to an organization’s overall resource pool. For instance, numerous studies have shown that the level of investments in skilled resources can greatly influence innovation assimilation (Chatterjee et al., 2002). Skilled resource factors such as education level, number of technical specialists and professionals have been shown as wielding considerable influence in driving innovation assimilation (Damanpour, 1991) and Cooprider and Victor (1993) in separate studies also found evidence that high levels of IS and business unit knowledge can also enhance IT assimilation in organizations. Another important dimension of skilled resources is knowledge transfer of the innovation. For instance, abundance or scarcity of pre-existing knowledge related to the focal innovation, and legacy knowledge can have major influence on assimilation (Attewell, 1992; Cohen & Levinthal, 1990; Fichman & Kemerer, 1997a). Extending this 100
discussion, Fichman and Kemerer (1997b) write that organizations must be prepared to invest in mechanisms that facilitate knowledge acquisition during the assimilation process such as hiring of "mentors”. In the federal government upon promulgation, EA lacked concrete methodology or detail, with most artifacts and guidance documentation (such as the Federal Enterprise Architecture Framework-FEAF) being developed retrospectively (Andrues, 2006). The above arguments show that both funding and skilled resources do have a causal effect on the access to resources construct. However, there is also a symbiotic relationship between the two indicators as the organization needs money to hire the skilled resources. On the other hand, having slush funding but lacking skilled resources would not be beneficial to the organization in assimilating its innovations. Based on the above arguments, it is apparent that: Hypothesis 1a. Access to resources is positively related to EA assimilation. Parochialism and Cultural Resistance Parochialism and cultural resistance are defined as negative attitude towards changes or an innovation. Parochialism in particular has been associated traditionally with large and bureaucratic organizations (Weber, 1922) as goals of a bureaucracy are predictability, efficiency, and stability (Hellriegel & Slocum, 2007).Parochialism fosters arrogance and inward focus -- features that will undermine an organization's ability to change (Kotter & Heskett, 1992). Because EA implementations involve business transformation which introduces huge organizational change including new governance structures, business processes and new ways to make sense of the organization (Schekkerman, 2005: 31; Ross et al., 2006; Hargadon & Douglas, 2001), organizations 101
with high levels of parochialism and cultural resistance are not well positioned to assimilate EA. When Information Systems innovations threaten an existing organization’s culture, the resistant forces may sabotage implementation by either rejecting it or it will manipulate it in order to match the existing culture (Litwinenko & Cooper, 1994). It is thus not surprising that with EA threatening predictability and stability, the effects of parochialism and cultural resistance may be significant. I posit that parochialism and cultural resistance will impede EA assimilation and suggesting that: Hypothesis 1b. Parochialism and cultural resistance is inversely related to EA assimilation. Top Management Value Recognition Value Recognition is a dual process of assimilation and accommodation of the value of innovation and involves constant modification of information to fit into what is already known (Eby, Molnar, & Shope, 1998). Tolbert and Zucker (1983) and DeLone & McLean, (2003) found that executives and community leaders are also an important source of information that shapes other executives’ perceptions of value. Tallon, Kraemer and Gurbaxani (2005) have found similar results. Hence, failure of the top management to recognize the value of the innovation as communicated by the peers and other media is a good indication of a lack of knowledge and skills of how to best implement it (Fichman & Kemerer, 1999). Moreover, not only do the CEO's perceptions toward IT matter, but also the sense of importance they attach to it (Liang et al., 2007).A recent global study of 659 CEOs conducted by the London School of Economics showed that only 25 per cent had value appreciation of IT. The GAO report (GAO-06-831) of 2006 faulted agency top management’s failure to recognize and embrace EA in strategic decision-making for its
102
dismal assimilation. I thus posit that positive value recognition of EA among top executives is critical to EA assimilation leading to the next hypothesis: Hypothesis 1c. Higher management value recognition is positively related to EA assimilation. Organization Complexity Organization complexity is defined as the amount of differentiation among distinct elements constituting the organization including the number of missions, functional units and the number of regulations, as well as their mutual dependencies.In this study I define organization complexity as a multidimensional construct with three formative components: 1) complexity of mission heterogeneity, 2) complexity of administrative multiplicity and 3) complexity of the legal and regulatory framework. I will next characterize these three components. Mission heterogeneity. Mission heterogeneity refers to the degree of dissimilarity of mission areas or mission focus within an organization. Studies such as the one by Zhu et al. (2006a) found evidence that complexity in organizations, as epitomized by dissimilar or heterogeneous missions encountered major difficulties in IT innovation assimilation. Ross et al. (2006) also note that organizations segmented into multiple strategic and specialized business units typically end up with different strategies thus complicating their efforts in maturing EA. High mission heterogeneity can also lead to the organization being in a state of flux with discordant views or “differentlydirectioned” effects as Fichman, (2001) calls it, further compounding complexity. Members of the business units tend to view their respective units as being autonomous, mission focused, rational, efficiency-seeking operations (Boddy, Boonstra, & Kennedy, 103
2006). However, Fichman (2004) writes that studies of IT innovation have so far failed to find support for this hypothesis (Fichman & Kemerer, 1997a; Grover & Goslar, 1993; Nilakanta & Scamell, 1990; Zmud, 1982). Overall I make the assumption that higher mission heterogeneity leads to increased level of organizational complexity. Administrative multiplicity. Administrative multiplicity refers to the extent to which administrative and decision-making authority is dispersed within an organization (Aiken & Hage, 1971; Ross et al., 2006). Administrative multiplicity within organizations can be attributed to duplicative managerial roles and tasks leading to complexity associated with a dysfunctional organization structure (DeLucca & McDowell, 1992; Gentile, 1996). A dysfunctional organizational structure with a variety of managerial roles can impact teamwork as departments within the same organization are unwilling or unable to cooperate. This may not only result in toxic working relationships, but employees within the departments may end up lacking a sense of camaraderie. Administrative dysfunctions have been shown to result in structural inertia which retards IT innovation assimilation (Zhu, Kraemer, & Xu, 2006). EA assimilation entails adaptation of existing IS, business process redesign, or other far reaching adjustments to organizational structure, all of which are dependent on the prevailing authority and decision-making structure (Barua, Konana, Whinston, & Yin, 2004, Zhu, 2004, Chatterjee et al., 2002). Overall this study posits that higher administrative multiplicity increases organizational complexity. Complexity of the legal and regulatory framework. This factor is defined as the scope and number of environmental and regulatory restrictions influencing the organization, including government and consumer regulations (such as the Patriot Act, 104
HIPPA, SOX), legal (Intellectual Property & contract enforcement) and industry policies (Cyber security and nuclear data). Dasgupta, Agarwal, Ioannidis and Gopalakrishnan (1999) confirmed that highly restrictive government policies led to lower IT assimilation (see also Zhu & Kraemer, 2005). The number of industry and environmental regulations and restrictions imposed on organizations have also been associated with both higher complexity and therefore an impediment to assimilation (Liang et al., 2007). Overall, in the past organization complexity has not only been associated with increased variance and diversification of business processes,but also within complex organizational structures, all of which have been identified as having negative effects on IT innovation assimilation (Zhu et al., 2004: 42). Strassmann (2005) moreover argued that the benefits enjoyed by higher IT budgetary allocations will eventually face erosion unless EA is deployed to contain increased complexity in IT services and governance. I find Strassmann’s argument reasonable and persuasive, but also observe that increased complexity at any part of the organization will negatively influence EA assimilation4. Overall, I posit that organization complexity will have negative effects on EA
assimilation. Hypothesis 1d. Higher organization complexity will inversely influence levels of EA assimilation. Organization Scope Organization scope is defined as the geographical extent of a firm’s operations (Zhu et al., 2006). In this paper’s context it is narrowly defined by the challenges and obstacles associated with organizations operating in heterogeneous geographies. I
4
At the same time complex organizations are the ones who will most likely benefit most of EA so the effects of complexity are highly paradoxical.
105
recognize however that the organization’s scope is multidimensional in nature and can vary, among others, along geographical, market based, economic, political or jurisdictional dimensions. Thus, organizations operating in multiple regions, counties, states and nation-states have to contend with a larger number of requirements based on local economic conditions, political and administrative priorities and even in some cases religious influence (Bainbridge, 2003). For instance Gurbaxani and Whang (1991) found evidence that companies faced steep increases in transaction costs when they expanded into heterogeneous markets. The effect of organization scope on innovation assimilation was first identified in the technology-organization environment (TOE) framework (Tornatzky & Fleischer 1990) which studied contextual factors that influence e-business assimilation. The multifaceted TOE framework identified three aspects of a firm’s context known to influence assimilation of a technological innovation. But of the three aspects, this study concerns itself with the organizational context which refers to organization factors such as scope (Zhu et al., 2006b). Two formative components are expected to affect an organization’s scope: jurisdiction and autonomy. Both factors are known from previous studies to considerably affect the firm’s global operational scope (Damanpour, 1991). Jurisdiction. Jurisdiction is defined as the span of geographic and operational authority including administrative, technological, political and economic influence (Lusthaus, Anderson, & Murphy, 1995; Datta & Nugent, 1998). In large organizations, departments can face different geographical, operational and environments challenges (Lawrence & Lorsch, 1967). For example, sovereign states and governments often set up restrictions and regulations that ultimately affect the organization’s operational scope 106
(Burgess & Fried, 2002). The legislative and judicial arms of the various sovereign states can limit operations of global organizations for political, cultural or safety reasons (Kraemer et al., 2006). This can severely impact the organization’s operational scope as they accede to or comply with the statutory requirements. Studies have shown that organizations operating within multi and cross-country jurisdictions face scope restrictions when diffusing innovations (Teece, 1980, Zhu et al., 2006). In a classic case of jurisdiction’s influence on organization scope, Kraemer et al. (2006) found evidence that IT diffusion occurred unevenly across countries within varying jurisdictions, as the pace of infrastructure development and political dynamics played a major role in determining the pace of diffusion. The study also found that jurisdictional differences shaped both the factors and diffusion patterns of e-business within developed and developing countries (Zhu & Kraemer, 2005). Overall higher variability and range of jurisdictions have a causal effect on organizational scope. Autonomy. Autonomy is defined as the “degree to which an organization is free to make decisions with respect to its own operations” (Molnar & Rogers, 1976: 62). In general, it refers to the degree of self-governance in decision making. Whereas every organization is subject to some form of environmental constraint, the degree of autonomy to pursue organizational interests can be greatly hampered (Wang, 2008). In such cases lack of autonomy directly influences the organization’s operational scope. For example, resource dependent firms and thus less autonomous firms are usually subjected to higher levels of micro-management by sponsoring organizations (Mizruchi & Stearns, 1994). This limitation in autonomy ultimately affects the organization’s scope boundary. An organization’s autonomy can also have impact on its ability to innovate (Powell & 107
DiMaggio, 1991; Fichman, 2001). Prior research has also shown that an organization’s dependency on external entities can impact its capacity to diffuse innovations (DiMaggio & Powell, 1983). High dependence on external resources constrains also organizational abilities for IT innovation (Wang, 2008). It is thus evident that an organization’s autonomy affects the scope of its operations. Therefore I posit that Hypothesis 1e. Larger Organization scope will negatively influence EA assimilation. Moderating Effect of Access to Resources There are several challenges associated with access to resources that position it as an influential moderator of the hypothesized relationships (H1a-H1e). For instance, organizations have been known to invest in technologies to force a cultural change (Senior & Fleming, 2006). I do expect that easy access to resources will have a moderating effect on the powerful forces of parochialism and cultural resistance. Studies have also shown that the availability of sufficient resources heavily influences assimilation of innovations by availing funds to engage top management in the decision making and implementation process (Rogers, 1995; Fichman, 2000; Liang et al., 2007; Kappelman, 2010). The focus of resource accessibility as conceptualized in this paper however goes beyond funding and skilled resources. It addresses the overall impact of the distribution pattern of the twin resources and the overall impact within the organization. It has been shown, for instance, that the scarcity of slack resources is associated with uneven distribution of technical expertise (Rogers, 1995). This can have a major impact on the organization’s operational scope, in aspects such as resource dependency, infrastructure development and economic influence (Zhu et al., 2005). Extant reviews 108
show that resource-related factors enhance or restrict an organization’s capabilities for investing in new technologies (Zhu & Kraemer, 2005). Therefore, I also use access to resources as a moderating factor to examine the impact that resource availability has on the causal relationships between identified antecedents and EA assimilation. Thus my next set of hypotheses: Hypothesis 2a. Increased access to resources will decrease the effect of negative relationship between Parochialism and cultural resistance and EA assimilation. Hypothesis 2b. Increased Access to resources will increase the effect of positive relationship between management value recognition and EA assimilation. I follow the Strassmann (2005) argument which casts doubt on the widely held perception that larger IT budgets would automatically lead to improved IT activities. Without an organizing framework such as EA, which Strassmann describes as creating structure and order, complexity and cost of IT tend to rise eroding the IT budget advantage. This leads me to the next hypothesis: Hypothesis 2c. Increased Access to resources will decrease the effect of the negative relationship between Organization complexity and EA assimilation. As argued before, organizations thinly spread out and operating in multiple environments, states and regional markets have to contend with the requirements of local economic conditions, political and administrative priorities. I argue that unfettered access to resources can in fact help mitigate challenges associated with highly restrictive operating environments (Zhu et al., 2006). Past research has also shown that resource abundance can have a drastic impact on ease or rate of assimilating an innovation (Fichman, 1999; Fichman & Kemerer, 2002). Accordingly, I posit that resource abundance can in fact mitigate the debilitating effects of limited organization scope. A 109
munificent environment fosters opportunities for sustained growth, whereas slack resources can remove more easily the barriers to diffusion of innovations (Dess & Beard, 1984; Rogers, 1995). For instance Zhu, Kraemer and Xu (2006) found evidence that economic environments positively shape innovation assimilation. I therefore argue that the factors that make up the economic environment; namely financial and human capital do have a moderating effect on the hypothesized causal relationships between organization scope and EA assimilation. This leads to my next hypothesis that: Hypothesis 2d. Increased Access to resources will decrease the effect of negative relationship between organization scope and EA assimilation Moderating Effects of Coercive Pressure Coercive pressure is defined as exertion of “sanction-laden” pressure from government regulators and policy enforcement agencies (Liang et al., 2007). Such pressure is exerted primarily via proxy institutions such as regulatory bodies. In the case of the U.S. government, regulatory agencies such as the Government Accountability Office (GAO) and the Office of Management and Budget (OMB) serve as the regulatory proxies to drive mandatory compliance. The effect of coercive pressure, especially of government regulations and directives is well researched (e.g. Hoque & Hopper, 1994; Ansari, Bell, & Lundblad, 1992). Both studies found evidence that the coercive pressure from central banks directly affected daily banking activities and operations. Banks complied with each and every operational directive, assimilated systems, behaviors and decision-making techniques prescribed by the regulatory bodies without question. The study revealed that such coercive pressure inadvertently pushed the banks into “legitimacy seeking” over efficiency gains. In this study I use coercive pressure as a 110
moderator on the relationship between the model’s antecedents and EA assimilation. Several studies have shown that organizations under the sway of institutional or government control view the pursuit of legitimacy as important as the actual performance (Meyer & Scott, 1983). Liang et al. (2007) used coercive pressure as a direct antecedent for ERP assimilation and surprisingly found that coercive pressure did not have any direct impact on innovation assimilation. In contrast, Wang (2008) used coercive pressure as a mediator variable in his longitudinal study. Wang found that coercive pressure had a significant mediating effect on assimilation of innovation. It should be noted however that Wang used standalone institutional factors as independent variables, as opposed to merging them into higher level constructs, as my study does. According to Baron and Kenny (1986: 1176), a mediator variable "accounts for the relation between the predictor and the criterion". A moderator variable on the other hand is defined as "a qualitative or quantitative variable that affects the direction and/or strength of the relation between an independent and dependent or criterion variable" (Baron & Kenny, 1986: 1174). My study was interested in establishing not only the predictive power of the antecedents, but also how different values of the moderator affected the strength of the relationship between the study’s antecedents and the DV. I do not exclude the possibility that my moderator variables may be a hybrid, in which case it acts as both a mediator and a moderator, also called "Mediated Moderation" (Baron & Kenny, 1986). However, the focus of this paper as stated above is to establish the “dial-button effect” of institutional coercive pressure on the relationships between the IVs and the DV. I therefore posit that
111
coercive pressure has a moderation effect on the relationship between the organizational antecedents and EA assimilation. This leads me to the next hypotheses: Hypothesis 3a. Higher levels of coercive pressure will positively moderate the relationships between Access to resources and EA assimilation. Liang et al. (2007: 9) writes that “top management team members are the focal point of coercive pressures” and are forced into driving the assimilation, regardless of whether they believe in the benefits of the innovation or not. I have yet to come across a single study that showed a true change in cultural resistance once coercive pressure was exerted. The studies conducted so far show strong evidence that the coercive pressure yields compliance, but only anecdotal evidence of its affect in reduction of parochialism and cultural resistance. However organizations experiencing coercive pressure are forced to continually demonstrate to the regulators that they are acting on “collectively valued” purposes meaning that indirectly, reluctant executives and employees are forcefully indulged (Meyer & Rowan, 1977). The threat of imminent job losses as a result of the company hiring external staff and consultants to replace reluctant employees and managers eventually weakens the resolve of the resistant groups (Senior & Fleming, 2006). Organizations succumbing to such pressure find it thus necessary to ensure their continued access to resources that their organizations need for survival, in the process blunting the effects of parochialism and cultural resistance. It is my contention, based on arguments by Hoque and Hopper (1994) that coercive pressure can positively moderate the relationship between parochialism and cultural resistance and EA assimilation. I’m persuaded by Liang et al. (2007: 9) who argue that coercive pressure forces managers and employees as resource benefactors, into submission, thereby negating the influence of 112
resistance to the mandated or prescribed innovation. It is in supporting this argument that I propose that coercive pressure will compel acquiescence, by streamlining the organization’s focus on compliance and satisfaction of the regulators’ requirements over the narrow interests of parochialists and culturally resistant individuals within the organization’s ranks. I therefore propose that: Hypothesis 3b. Higher levels of coercive pressure will decrease the effect of Parochialism and cultural resistance on EA assimilation. Liang et al. (2007) found that the top managers' thoughts and actions were effective channels for advancing institutional compliance. However, coercive pressure eventually shaped top managers' participation in the ERP assimilation process thereby positively affecting ERP assimilation. Following Liang et al. (2007), I contend that coercive pressure plays a moderating role by permeating the organizational boundary and shaping the top management’s value recognition, leading to successful EA assimilation. Thus this leads to my next hypothesis. Hypothesis 3c. Higher coercive pressure will positively moderate the relationship between management value recognition and EA assimilation. Liang et al. (2007) also argue that the continued evolution and institutionalization of an IT innovation eventually leads to increased coercive pressure in the forms of new laws, regulations, as well as professional and industry standards. I adopt that line of argument and contend that as the number of regulatory, legal and industry regulations and pressure grow, they not only increase organization complexity but further inversely impacting EA assimilation. This leads to my next hypotheses: Hypothesis 3d. Higher coercive pressure will decrease the negative influence of organization complexity on EA assimilation. 113
Hypothesis 3e. Higher coercive pressure will decrease the negative influence of organization scope on EA assimilation. Research Design and Data Collection Data Sample and Scales I collected publicly available government data from 123 EA programs out of 163 Federal Enterprise Architecture programs to validate the research model. This represents approximately 75% of federal-wide EA programs. I dropped 17 programs for the following reasons 1) data was not available for some items as they are secret (e.g. several Department of Defense agencies, Central Intelligence Agency, Federal Bureau of Investigations, National Security Agency), 2) some agencies were too small which would bias the sample e.g. the Federal Railroad administration (FRA), the Legal Services Corporation (LSC) 3) some agencies are too large and would bias the sample (e.g. the department of health and human services and the department of transportation as well as the Department of Defense). By using these criteria I included in the sample 123 agencies. The list of agencies is included in Appendix P. TABLE 14: Data Details and Sources Research sample Data sources
Years in scope
123 /163 Federal EA Programs (Please see Appendix P) 42 with assimilation levels 4 & above/ 81 with assimilation 3 & below • • •
Government Accountability Office (GAO) EA assessment results GAO – Survey results on EA challenges -Mgmt, culture, etc Office of Management and Budget (OMB) – Agency budgets
2002– 2008
Five distinct data sources were used to collect a panel data set that covered all constructs included in the research model (Table 14). Hence all constructs were operationalized using as proxies data items and indicators that were publicly available or 114
obtained from GAO or OMB. A large number of items were obtained from GAO EA assimilation measurement survey (GAO-04-40; GAO-04-798T; GAO-06-831), which will be explained next. Construct Operationalization and Scales A summary of scales used to measure each construct and their respective data sources are shown in Table 14.The data were collected in multiple stages with some of the data having been vaulted at the U.S. National archives. The U.S. government has a five year public data access policy after which the data is sent to the national archives for vaulting. Congressional committee approval had to be obtained for release of some of the data. Some of the data were also obtained from annual inspector general reports and from budgetary proposals which contain a great amount of operational detail. Many of the scales in the final data set were operationalized using data proxies from the GAO EA implementation survey. This was obtained from the National Archives and Records Administration (NARA). The rest of the data was obtained from agency web-sites, Code of Federal regulations; gpoaccess.gov and OMB. I will next explain how GAO survey was collected as this data set forms the nexus of the used data for analysis. In general, the GAO collects and analyzes data for the congressional decision making (Keenan & Mauch, 1986). Unlike pollsters and market researchers, GAO evaluators rarely do national surveys- only government-wide surveys. Hence, the GAO seeks information only from members of special populations, such as federal and state government employees, welfare recipients, or agency executives (Keenan & Mauch, 1986). Most of the data are collected via assessments, surveys and interviews. However, the periods between surveys can be quite long (multiple years). At the same time the 115
program assessments and audits are also comprehensive. The questionnaires are mostly prepared by the internal Program Evaluation and Methodology Division (PEMD). The questionnaires ask key demographics for figures, statistics, amounts, and other facts which the study used to build proxies for organizational characteristics of the agencies (Keenan & Mauch, 1986). Since 1999 the GAO has been conducting an EA program survey and interviewed all federal agencies about their usage, benefits and challenges to EA implementation. The measures and items of the scale have remained the same throughout the three audits. A sample of the EA survey instrument used is shown in Appendix M. I especially have included indicators, which were selected as proxies for each construct. Most of the data from the survey instrument covered five constructs: value recognition, funding, parochialism and cultural resistance and skilled resources and assimilation levels. TABLE 15: Scales and Measures Overview Construct/variable EA Assimilation
Computation Points scale (1- 5). Computed as:1 = EA at lowest assimilation; 2 = EA at low assimilation level; 3 = EA at medium assimilation level; 4 = EA at high assimilation level; 5 = EA at highest assimilation level. Please see Appendix N for how this scale is computed
Organization Complexity Mission Points scale: 1=1-5 missions ; 2 = 6-10 heterogeneity missions ; 3 = 11-15 missions ; 4 = 15-20 missions ; 5 = 21-25 missions ; 6 = 26-30 missions; 7 = 31-35 missions ; 8 = 36-40 missions ; 9 = 41-45 missions ; 10 = 46-50 missions Administrative Points scale: 1-10 units (1); 11-20 units (2); multiplicity 21-30 units (3); 31-40 units (4) 41-50 units (5); 51-60 units (6); 61-70 units (7); 71-80 units (8); 81-90 units (9); 91-100 units (10).
Years 2001-2008
2001-2008
2001-2008
Sources GAO audit results / National archives (NARA)/ Inspector general reports and GAO congressional reports Agency websites: data such as number of missions per agency; number of functions (using organization
116
Legal & Regulatory framework
Organization Scope Autonomy
Jurisdiction
Access to resources Funding
Skilled resources
Top Management Value recognition
Points scale: 1-20 regulations (1); 21-30 regulations (2); 31-40 regulations (3); 41-50 regulations (4); 51-60 regulations (5); 61-70 regulations (6); 71-80 regulations (7); 81-90 regulations (8); 91-100 regulations (9); 100200 regulations (10)
2001-2008
charts; number of physical facilities, such as buildings, office locations, military bases, laboratories and field offices, satellite operation centers etc
Points: 1 = standalone bureau; 2 = Bureau under sub-agency, 3 = Bureau under department; 4= Sub-agency under agency; 5=agency under department; 6= independent agency; 7=cabinet level department Points scale (1=International, 2=Multi-state or national, 3=Single state or local authority).
2001-2008
1.Code of Federal regulations: data on number of legal and regulatory requirements by agency 2. gpoaccess.gov Agency websites: on parent agencies and departments for each of the bureaus and sub-agencies
Likert scale:1= very great extent; 2= great extent; 3= moderate extent; 4= some or little extent; 5= no extent Likert scale:1= very great extent; 2= great extent; 3= moderate extent; 4= some or little extent; 5= no extent
Survey years 1999, 2002, 2006 Survey years 1999, 2002, 2006
Likert scale:1= very great extent; 2= great extent; 3= moderate extent; 4= some or little extent; 5= no extent
Survey years 1999, 2002, 2006
2001-2008
1.National Archives and Records Administration : With permission from Congressional chair 2.GAO:1999, 2003 and 2006 EA survey results 4.National Archives and Records Administration : With permission from
117
Parochialism and cultural resistance
Likert scale:1= very great extent; 2= great extent; 3= moderate extent; 4= some or little extent; 5= no extent
Survey years 1999, 2002, 2006
Coercive Pressure (EAAF)
3-Point scale based on aggregate 3-color rating: Red (1) = non -compliant; yellow (2) = partially compliant and green (3) = fully compliant. Please see Appendix G for details Points scale {1 = (1-25k ) 2 = (25k-50k) 3 = (50k-75k) 4 = (75k-100k) 5 = (100k-125k) 6 = (125k-150k) 7 = (150k - 175k) 8 = (175k200k) 9 = (200k-225k) 10 = (225k - 1m)}
2002-2008
Size
2001-2008
Congressional chair GAO:1999, 2003 and 2006 EA survey results 3.National Archives and Records Administration : With permission from Congressional chair GAO:1999, 2003 and 2006 EA survey results OMB/NARA; E-gov.gov
Agency websites/portal
Construct Definition and Operationalization Detailed descriptions of each construct and its operationalization will follow. Detailed definitions are summarized in Appendix D. Overall the final constructs included three formative second order constructs (Petter & Straub, 2007). In formative constructs, the items or indicators form an index and related variance underlying the construct; with the causal direction from the items to the latent formative construct (Petter & Straub, 2007). Dependent Variable EA assimilation. EA assimilation was measured as the extent to which the organization has progressed through its EA assimilation process from initial awareness to full institutionalization (Fichman, 2001). I used the GAO Enterprise Architecture 118
assimilation ranking as an ordinal scale measure to indicate the assimilation level. The scale is computed as an index from several indicators reflecting the extent to which different aspects of EA discipline have been integrated into the agency’s operations. The GAO assimilation framework, which is mapped to this scale, is a self-explanatory scorecard, used for the assessment of the EA assimilation process. The scorecard uses a combination of two complementary methods used to determine an organization’s assimilation level. The first method is focused on the weighted mean assimilation level, whereas the second method is focused on the percentage achieved at each assimilation level for the thirty one EA core elements. This measure was derived from the EA program audit data as contained in the GAO assessment reports (GAO-04-40; GAO-04798T; GAO-06-831). The GAO EA assimilation measure uses a five point scale captured in a Guttmann scale (Schuman & Presser, 1981) and it shows the consecutive levels at which an EA program has been formalized and integrated into the business operations of an agency (Bernard, 2005). As noted measure is composed of 5 assimilation levels which are composed of 4 critical success attributes and 31 core elements that help link the attributes to the assimilation levels. Appendix N enlists the core elements that were used for the scoring and determining the assimilation level. Independent Variables Access to resources. Access to resources is a multidimensional construct encompassing the availability of financial and skilled human capital resources. I used the data from GAO surveys in 1999, 2001 and 2003. 1) Funding: The instrument questionnaire asked executives to identify the extent to which lack of funding impacted their organization’s EA assimilation. The responses were tapped with an inverse five 119
point Likert scale, with (1) being “to a very great extent”; (2) “to a great extent”; (3) “moderate extent”; (4) some or little extent and (5) being to “no extent.” 2) The instrument questionnaire asked executives to identify the extent to which lack of skilled staff impacted their organization’s EA assimilation. The responses were tapped with an inverse five point Likert scale, with (1) being “to a very great extent”; (2) “to a great extent”; (3) “moderate extent”; (4) some or little extent and (5) being to “no extent”. Please see Appendix M for details on the survey instrument and Table 15 for details on the scale. Parochialism and cultural resistance. The data for this construct was obtained from the GAO surveys of 1999, 2001 and 2003. The questionnaire asked executives to identify the extent to which parochialism and cultural resistance impacted their organization’s EA assimilation. The responses were tapped with an inverse five point Likert scale, with (1) being “to a very great extent”; (2) “to a great extent”; (3) “moderate extent”; (4) some or little extent and (5) being to “no extent”. Please see Appendix M for details on the survey instrument and Table 15 for details on the scale. Management value recognition. The data for this antecedent was derived from the GAO surveys of 1999, 2001 and 2003. The instrument is in Appendix M. The questionnaire asked executives to identify the extent to which lack of top management understanding of the importance and value of enterprise architecture impacted their EA assimilation. The responses were tapped with an inverse five point Likert scale, with (1) being “to a very great extent”; (2) “to a great extent”; (3) “moderate extent”; (4) some or little extent and (5) being to “no extent.” Please see Appendix M for details on the survey instrument and Table 15 for details on the scale. 120
Organization complexity. The mission heterogeneity indicator was operationalized as the number of missions an agency had. I used a 10-point scale as shown in Table 15 above. The data were obtained and coded from the agency’s website. The administrative multiplicity indicator was operationalized as the number of agency managers with decision-making authority. The data were gathered from organization charts, job descriptions and Inspector-General reports. Again, I used a 10-point scale to gain enough variance. The Legal and Regulatory framework factor was operationalized as the number of legal and regulatory requirements an agency had to comply with or enforce for the industries within its purview. This data was obtained from the code of federal regulations (gpoaccess.gov) and agency websites. I also used a 10-point scale for this factor in order to tap into adequate variance. Organization scope. The organization scope construct was operationalized as a second order formative construct composed of jurisdiction and autonomy. Jurisdiction was operationalized as the extent of operational coverage: whether the organization operated in a single state or had multi-state, national or international presence. Using a 4 point scale, an agency was scored as international, if it had international operations and facilities; nation-wide if it operated in multiple states or local if it operated within a single state or local authority (see Table 15 on the points scale for this factor). Autonomy was defined as the degree of presence of authority, independent decision-making and resource independence. I used the government tier structure to tap into this phenomenon. Federal government tier 1 agencies are cabinet level departments and agencies whose executives are appointed by the executive office of the President and have direct reporting relationship with the White house. Tier 2 agencies would be independent agencies, tier 3 121
agencies are non-executive agencies, tier 4 agencies are sub-agencies within the tier 2 agencies and tier 5 agencies are bureaus within either the tier 3 or tier 4 agencies or subagencies nested within the top tier agencies. The underlying conceptual logic was that cabinet level and independent agencies enjoy more decision-making autonomy than those with multi-layered reporting relationships. Coercive pressure. This factor was defined as the extent of political, regulatory and controlling pressure exerted on the government agencies. This study uses the OMB audit scores as the proxy for coercive pressure. The OMB’s annual EA assessment is 80% comprised of enforcement of mandatory requirements that need to be met by the agency. Agency budgetary requests face automatic rejection unless they demonstrate 1) strict compliance with these mandates and 2) specific year-to-year performance improvements as stipulated by the OMB. Agencies have to show measurable results from their EA programs, such as tangible cost savings, cost avoidance by way of technology, datacenter, service and facility sharing, in addition to general EA program maturation in order for their budgetary requests to be approved. The OMB also holds agency senior management accountable for EA implementation and the agency heads are required to take personal responsibility for EA governance. Any deficiencies identified by the OMB have to be addressed “hands on” by the agency executives. The OMB has powers to sanction, such as rejecting or slashing the budgetary requests, if the agencies are noncompliant, a situation agency heads would like to avoid at all costs. In addition, the agency EA performances, including areas of deficiency are displayed on the EGovernment portal, for congressional and citizen consumption. As can be seen from Appendix G, the OMB scores agencies based on level of compliance using a color-coded 122
scheme (red means non-compliant; yellow means partially compliant and green means fully compliant). This was computed using the weighted scores which were then aggregated into 3-color coded rating scorecard by the OMB. The study adapted and converted the color codes into a 3-point scale mapped as: Red (1) = non -compliant; yellow (2) = partially compliant and green (3) = fully compliant Controls Organization size. Organization size has been consistently found to influence innovation diffusion (Rogers, 1995); as well as assimilation of innovation (Tornatzky & Fleischer, 1990; Damanpour 1996). Larger organizations usually possess more slack resources and greater capacity for assimilating innovations (Kimberly & Evanisko, 1981; Damanpour, 1991). Slack resources also help large organizations to seize opportunities and implement innovations (Armstrong & Sambamurthy, 1999). Fichman (2000) writes that size serves as a proxy for other positively related variables, such as scale, wealth, specialization, and slack resources. This factor was operationalized as the number of employees within an agency. I used a 10 point scale for this factor (please see Table 15). Statistical Analysis Method I used partial least squares (PLS) path modeling developed originally by Wold (1982) to estimate the model. PLS was specifically chosen for this study for the following reasons: 1) the phenomenon under investigation was relatively new and required the development of the measurement model, 2) the focus of the study rendered the predictive power of the model more important than the parameter estimates, 3) the data not only had sample size deficiency, but also did not meet the independence and normal distribution requirements (Marcoulides & Sanders, 2006) 123
PLS is a nonparametric estimation method. And as argued above, PLS is more tolerant towards violations against normality assumptions, and is therefore regarded to be the most suitable for the analysis of small samples (Wold, 1988). Again, as mentioned in the above section the choice of PLS was that my sample size was admittedly small (n = (123). I therefore resorted to resampling (bootstrapping) in PLS (Efron & Tibshirani, 1993) to overcome sample size limitations and create reliable estimates for all estimated parameters. At the same time PLS does not offer significance tests based on statistical distributions. However, a jackknife procedure packaged in the PLS software (Lohmoeller, 1989) can be used to calculate the standard deviations for parameter estimates and generate a t-approximate for those estimates. In PLS, the path coefficients are thus viewed as standardized regression coefficients; with the loadings being similar to factor loadings. PLS offers three additional benefits for the statistical analysis given the complexity of the model and the size and nature of the data set. First, in PLS analysis relationships among latent variables are estimated and tested within the context of a measurement model that explicitly assumes measurement errors in the observable variables, whereas traditional multiple regression assumes that variables have been measured free of errors, an assumption that in the social sciences is questionable (Fornell, 1983). Second, PLS enables one to perform combined regression and factor analysis within the same statistical step, because factors or latent variables are created as linear combinations of observed indicators in the first level which are then used in simultaneous partial regressions with the latent variables through an iterative procedure (Wold, 1982). Third, PLS generates a variety of reliability and validity statistics that can be calculated 124
in the context of the theoretical measurement and structural model. In contrast, in the traditional regression procedures the estimation of such statistics (e.g. Cronbach's alpha) is independent of the model being tested. Finally, the measurement model in PLS can be defined to be either reflective or formative. Scale reliability and validity can be assessed via confirmatory factor analysis (CFA), which can be performed using the partial least squares (PLS) approach. CFA is more appropriate than alternative approaches such as exploratory factor analysis in areas with strong a priori theory and pre-validated measurement scales. Measurement Model I followed (Straub et al., 2007) guidelines to estimate the validity and reliability of formative constructs. To establish content validity, I first conducted in-depth literature reviews to scope the domain of the constructs. The literature reviews confirmed that I had three multidimensional formative constructs and that the indicators were good formative items (Straub et al., 2004).Using content validity, (I consulted widely with practitioners and experts from the Federal Government EA community as well as IS scholars, my conclusion was that the organization complexity construct had three formative indicators: legal and regulatory framework, administrative multiplicity and mission heterogeneity). The organization scope construct was also multidimensional and had two formative indicators: jurisdiction and autonomy. The access to resources formative construct had two formative indicators: funding and skilled resources. I next evaluated the statistical validity of the constructs using correlation weights within the constructs using PLS (Chin, 1998). I ran the measurement model with the three formative constructs and the two standalone antecedents included in the model. One 125
indication of poor measurement model with formative constructs is high multicollinearity (Straub et al., 2007). Multicollinearity is an undesirable property in formative models as it causes estimation difficulties (Diamantopoulos & Winklhofer, 2001). These estimation problems arise because a multiple regression links the formative indicators to the higher level construct. Multi-collinearity for each construct was measured with the variance inflation factor or VIF (Straub et al., 2007). Typically VIF values exceeding 10 are regarded as an indication of strong multicollinearity (Allison, 1999), though some also suggest a cut-off point of 3.The following are the VIF scores for the organization complexity indicators: Mission heterogeneity (1.261); administrative multiplicity (1.089) and legal and regulatory framework (1.102). For the access to resources construct, the following were VIF scores: Funding (3.462) and skilled resources (1.305). The organization scope construct had the following VIF scores for its indicators: autonomy (1.182) and jurisdiction (1.182). Typically indicators found to have high VIF, become candidates for removal (Hair et al., 1995). I concluded that these values were acceptable and there was no threat of multi-collinearity and therefore had no reason to drop any of the items. After that a full estimation of the measurement model was done (Table 15 shows the final model). Please see Appendix I for the correlation matrix used in estimation. Table 16 indicates the formative item regression weights that determine each “indicator-relevance”- the significance of each indicator’s weight in forming the underlying formative construct. Table 16 also contains the path coefficients showing each path’s loading, mean, standard deviation, standard error and t-statistic. The table also reports the R2 results i.e. how much each item explains the variance in the 126
established construct. Based on the values I concluded that the measurement model was acceptable as the indicators had little multi-collinearity, accounted for high level of variance in the construct and were significant. The low AVE and CR values for some of the constructs are acceptable as the items were part of formative construct. TABLE 16: Measurement Model Construct Organization Complexity
Access to Resources
Organization Scope
Causal indicators Mission heterogeneity Administrative multiplicity Legal and Regulatory Framework Funding Skilled resources Autonomy
Mean
Std Dev
N
Std. error
Loadings
Tolerance
VIF
R2
AVE
CR
1.86
1.445
685
8.266
.656
.793
1.261
.186
.230
.237
15.81
12.178
685
2.88
.576
.918
1.089
.074
12.79
9.411
685
4.17
.495
.907
1.102
0.92
0.48
0.844
685
21.361
.891
.289
3.462
0.705
.785
.832
0.73
1.3
685
9.566
.598
0.766
1.305
0.184
2.09
0.894
685
12.236
.730
.867
1.182
0.116
.680
.372
.846
1.182
Jurisdiction .414 1.86
0.715
685
15.474
0.153
I was satisfied that the measures were reliable and the constructs evidently accounted for at least 50% of the variance. Even though the reliability coefficients as shown in Table 16 were 0.832 for access to resources; 0.237 for organization complexity and 0.372 for organization scope respectively, I was nonetheless satisfied with the results since there is no benchmark or defined standard measure for formative measures. The AVEs were well above 0.50 for each of the constructs. The measurements were thus reliable as stated before because the constructs accounted for at least 50% of variance. Inspection of the formative item measures satisfied us that multi-collinearity was not a problem. I was generally satisfied that several loadings were high. I followed Chin (1998) argument that, covariance based estimates such as reliability and AVE were not applicable for evaluating my model’s formative constructs. Instead, I relied on the path 127
weights of the various indicators to check, if they significantly contributed to the constructs. The measurement of the constructs was thus conducted by examining significance of the path weights. All the path weights were significant at the .01 level, suggesting that they contribute significantly from different paths to form the constructs in the model. As stated earlier, I did not have to resort to item trimming to improve any of the constructs. I also noted that the control variable, size had significant positive correlations with administrative multiplicity; positive correlations with mission heterogeneity, positive correlations with legal and regulatory framework. However, size also had strong negative correlations with jurisdiction and a strong negative correlation with autonomy. To meet the sample size requirement for PLS analysis recommended by Chin (1998), I conducted testing using models with and without the control variable, meaning the analysis was run using five PLS models (Liang et al., 2007). The organization complexity construct had three incoming links from the first order formative constructs and another link from the moderator variable. Having dropped 17 data points, I had a sample size of 123, which meant that I was well within the limits as EA assimilation can have at most seven incoming links (Liang et al., 2007). For the moderation testing I used Prof. Hayes MODPROBE script, a popular SPSS Syntax by Prof. Andrew F. Hayes of Ohio State University. I followed the standard computational procedures by using macros for probing the interactions in OLS regression (Hayes & Mathes, 2009). The MODPROBE script estimates both the model coefficients and the standard errors in the model including among others, the predictor variables, the
128
product, and any additional variables (covariates) to estimate the model’s dependent variable (Hayes & Matthes, 2009). MODPROBE was the best suited for this analysis because the model had both first and second order constructs as moderating variables. In PLS the second order construct is formative in nature in the sense that first order items forms the second order construct. In this regard you cannot link a first order to another first order item (Vinzi et al., 2010). In my model, access to resources was a 2nd order construct. The script automatically creates the interaction term by multiplying the formative moderator construct’s indicators by the indicators from each of the independent variables’ indicator variables. In the case of the MODPROBE script, the interaction term is displayed in the summary. As stated earlier, I used a three phased approach for interaction testing, in addition to five separate models that will be explained below. The first model tested for direct effects of the antecedents on the dependent variable; two models tested for moderation effects without control variables while the other two models tested for moderation effects with the control variable. The first model was used to test for all the direct effects and had all the independent variables as well as the control variable but none of the moderating variables. The second model was used for testing the moderating effects of access to resources on the two second order constructs without the control variable, size. In the third model the control variable (size) was introduced. The fourth model was used to test for the moderation effect of coercive pressure first order construct with no control variable. The fifth model had coercive pressure as the moderator, all the antecedents and the control variable, size. 129
Findings The estimated path loadings, significance levels, and R2 values are presented in Table 16. The R2 value for EA assimilation was 70.9% in the initial model; which is relatively high and gives confidence for the validity of the model. The R2 statistic, however, did record a slight increase to 71% upon introduction of the control variable during the moderation testing. This observation is discussed further in the post-hoc analysis section. FIGURE 13: Regression Results with Access to Resources Moderation
130
Direct Effects (H1a-H1e) Hypothesis H1a was supported (β=0.841; p-value=0.00) confirming that access to resources does have significant impact on EA assimilation. Hypothesis H1b that parochialism and cultural resistance would negatively influence EA assimilation was supported. Though the path approached significance (β=0.527; p-value=0.06) the sign of the beta was positive. Hypothesis H1c that high management value perception would positively influence EA assimilation was supported (β=0.543; p-value=0.04). Hypothesis H1d that high complexity would retard EA assimilation was supported (β=0.042 p-value=0.001). Hypothesis H1e that organization scope would retard EA assimilation was supported (β=0.027; p-value=0.007). Results H1a – H1e TABLE 17: Hypothesis H1a-e Results Hypothesis
Description
H1a
Access to resources is positively related to EA assimilation. Parochialism and cultural resistance is inversely related to EA assimilation. High management value recognition is positively related to EA assimilation. High organization complexity is inversely related to EA assimilation. Organization scope will negatively influence EA assimilation.
H1b H1c H1d H1e
Supported Y/N
Beta
P-value
Supported
0.841
0.00
Supported
0.527
0.06
Supported
0.543
0.04
Supported
0.042
0.001
Supported
0.027
0.007
Results for Moderation Hypotheses H2a – H2d Hypothesis H2a that access to resources would positively moderate the relationship between parochialism and cultural resistance and EA assimilation was supported (β=-0.483; p-value=0.0072). Assimilation levels are surprisingly high with 131
increased access to resources even when parochialism levels go up. In contrast, and as expected, assimilation levels decrease as the level of parochialism and cultural resistance goes up, and when access to resources is low. Hypothesis H2b that access to resources would positively moderate the relationship between managerial value recognition and EA assimilation was partially supported (β=-0.1635; p-value=0.0797). The beta was negative, however with the alpha level = 0.05, the p-value was only significant at the 90% confidence level. Assimilation levels go up with increased access to resources and the effect of value recognition is amplified. In contrast, with low access to resources increase in value recognition leads to slightly lower assimilation. Hypothesis H2c that access to resources would negatively moderate the relationship between organization complexity and EA assimilation was not supported (β=-0.0027; p-value = 0.8528). This was a rather interesting result as it shows that availability of skilled manpower and financial resources can in fact positively neutralize the effect of organization complexity on EA assimilation. In light of the result in Hypothesis H1d, I re-tested this moderation by adding the control variable just to examine the interactions. The p-value for size as a control variable =0.7532 which was insignificant. The interaction plot shows that the effect of organizational complexity on EA assimilation is not affected by access to resources. Hypothesis H2d that access to resources would negatively moderate the relationship between organization scope and EA assimilation was not supported (β=0.0384; p-value=0.7562). The interaction plot shows that the effect of organizational scope on EA assimilation is not affected by access to resources.
132
Test of Hypotheses H2a – H2d Moderating effect of Access to Resources (please see Appendix K for output tables) TABLE 18: Moderation testing results for hypotheses H2a-H2d Hypotheses – Access to Resources would: (positively) moderate H2a
H2b
H2c
H2d
Parochialism and cultural resistance and EA assimilation (positively)moderate Management value recognition and EA assimilation (negatively) moderate Organization Complexity and EA assimilation (negatively) moderate Organization scope and EA assimilation
Supported Y/N
B
se
P
Exp (B)
Wald
Partially supported
-0.483
0.0812
0.0072
1.1686
3.6855
Partially supported
-0.1635
0.0933
0.0797
1.1776
3.0718
Not supported
-0.0027
0.0144
0.8528
0.9973
0.0344
Not supported
-0.0384
0.1238
0.7562
0.9623
0.0964
FIGURE 14: Regression Results with Coercive Pressure Moderation
133
Results for Moderation Hypotheses H3a – H3e TABLE 19: Moderation testing results H3a-H3e Hypotheses - Coercive pressure would: (positively) moderate H3a H3b
H3c
H3d
H3e
Access to Resources and EA assimilation (negatively) moderate Parochialism and cultural resistance and EA assimilation (positively) moderate Management Value Recognition and EA assimilation (negatively) moderate Organization complexity and EA assimilation (negatively) moderate Organization Scope and EA assimilation
Supported Y/N
B
Se
P
Exp (B)
Wald
Supported
0.5403
0.1852
0.0035
1.7164
8.5137
Supported
-0.2401
0.0734
0.0011
0.7865
10.7074
Supported
-0.2697
0.0843
0.0014
0.7636
10.2308
Not supported
-0.0076
0.0201
0.7046
0.9924
0.1438
Supported
-0.3042
0.1017
0.0028
0.7377
8.9469
Hypothesis H3a that coercive pressure would positively moderate the relationship between access to resources and EA assimilation was supported (β=0.5403; pvalue=0.0035). The interaction plot (Appendix K) shows that higher levels of coercive pressure leads to higher assimilation levels with both lower and higher access to resources. Hypothesis H3b that coercive pressure would negatively moderate the relationship between parochialism and cultural resistance and EA assimilation was supported (β=-0.2401; p-value=0.0011). The interaction plot shows that higher coercive pressure leads to higher assimilation levels with both lower and higher levels of parochialism. Hypothesis H3c was supported (β=-0.2697; p-value=0.0014). The interaction plot shows that higher coercive pressure leads to higher assimilation levels with both lower and higher levels of value recognition. Surprisingly though, with higher management value recognition the effect does not increase. Hypothesis H3d- - - was not 134
supported (β=-0.0076; p-value=0.7046). The interaction plot shows that higher coercive pressure leads to higher assimilation levels with lower and higher levels of organizational complexity. Surprisingly with higher complexity the effect does not increase whereas with low coercive pressure assimilation levels increase as complexity goes up. Hypothesis H3e - coercive pressure can mitigate the retarding effect of organization scope on EA assimilation-was supported (β=-0.3042; p-value=0.0028). The interaction plot shows that higher coercive pressure leads to higher assimilation levels with lower and higher levels of organizational scope. Surprisingly with higher scope the effect does increase with higher level of coercive pressure whereas with low coercive pressure the assimilation level remains largely the same. Please see Appendix K for the interaction plots and output tables. Overall, inclusion of the control variable does not add any significant value to the interactions. PostHoc Analysis The result for hypothesis H2d was a major surprise given its implication that unfettered access to financial and skilled resources can in fact positively mitigate the retarding effect of organization scope on EA assimilation. This ran contrary to the result in Hypothesis H1e. The result remained unchanged even after introduction of the control variable. There was no significant change in the R2of the model which remained at 70.9%. Most notable however was that the inclusion of the control variable (size) led to reduction in the regression weight between the organization scope construct and EA assimilation in the reverse direction. i.e. when controlled for size, every unit increase in EA assimilation led to reduction in organization scope by 0.1%. That means that when controlled for size, higher EA assimilation directly led to decreased organization scope. 135
Hypothesis H3e that coercive pressure mitigates the retarding effect of organization scope on EA assimilation-was supported. The model R2 statistic however recorded a slight increase to 71% upon introduction of the control variable. Similarly the moderating effect of coercive pressure on the relationship between organization scope and EA assimilation reversed directionally such that any unit increase in coercive pressure reduced the organization scope by 0.1%. There were some major differences in the moderation effect of both the access to resources and coercive pressure constructs. Whereas coercive pressure as theorized systematically increases assimilation levels, the effect of access to resources was varied, having an effect on some relationships and having no effect at all on other relationships. Discussion and Conclusion This research study set out to address the following question: 1) What organizational factors influence the level of EA assimilation? A secondary objective was to examine the effect of coercive pressure in accelerating the assimilation of EA. Using a dataset on EA assimilation on EA in federal agencies; I tested the effects of nine organization level factors on the level of EA assimilation. This research is the first ever to study the factors that influence EA assimilation and also the first that conducted analysis on the influence of coercive pressure on the EA assimilation process. To lend further coherence to the conceptual model, I identified the U.S. Office of Management and Budget (OMB) as the primary government agency which moderates the effect of both internal and external organization factors on the EA assimilation process. The moderation offered by the OMB is useful for explaining the variability in the level of 136
EA assimilation even across those organizations embedded in similar institutional contexts. In the study, I used the OMB’s coercive pressure as a moderating variable as opposed to a direct antecedent. Previous studies that focused on coercive pressure have used it as a direct antecedent (e.g. Liang et al., 2007) who found that it was not a significant predictor. Other studies such as Wang (2008) used coercive pressure as a mediator and found that it does play an influential mediation role. I set out to test the OMB’s coercive pressure as a moderator in my study and found evidence that it has significant moderating effect on a number of relationships. The major highlight of the study’s findings is the moderating effect of coercive pressure. The most astonishing result was the positive moderating effect of coercive pressure on parochialism and cultural resistance, which confirmed a popular belief that coercive pressure could forcefully defang the insidious effect of parochialism and cultural resistance in assimilating innovations in organizations. This finding is consistent with my earlier discussion about the main source of coercive pressure for EA assimilation – i.e. EA being a government mandate and backed by a powerful regulatory agency (OMB) in the context of this study. This may suggest that stiff resistance to innovation assimilation can indeed be tamed via the use of coercive pressure despite its potent threat to existing structural order and cultural inertia. It is instructive though that Hypothesis H2a was only partially supported, adding a new dimension to the argument on the potency of parochialism and cultural resistance as a retarding factor in EA assimilation. Although I did not have the data showing the number of external resources such as consultants and service organizations, an informed guess may be that organizations that had encountered 137
high parochialism and cultural resistance may have successfully diluted its effect by using their easy access to financial resources to hire skilled resources (such as external contractors) to develop and assimilate EA. The second result I will discuss is the moderating effect of coercive pressure on the relationship between management value recognition and EA assimilation. The hypothesis was supported; meaning that as top managers were coerced into assimilating EA, the resulting publicity and recognition from the executive branch of government may have influenced the change in value perception of EA. One possible explanation for this result may be found in the study which found that executives and thought leaders are an important source of information that shapes other executives’ perceptions of IT value (DeLone & McLean study, 2003). The OMB dispatched management experts to work with top managers at non-compliant agencies, which may have led to removal of knowledge barriers. This action may have also led to the top agency managers being persuaded and cajoled into appreciating the value of EA. Another explanation could be from studies in legitimacy (Suchman, 1995) that have consistently argued that legitimation eventually prevails over management resistance to innovations. The surprise result however was that size, when introduced as a control variable positively reversed the effect of access to resources moderation on the relationship between top management’s value recognition of EA and EA assimilation. This suggests that in larger organizations, top management may view EA positively as it works to contain complexity. This is consistent with Ross et al. (2006) who argue that at the medium assimilation level top management are turned into “believers in EA” after witnessing its ability to contain both complexity and costs. 138
The results also did support my argument that organization scope would have a strong inverse relationship to EA assimilation. My results support previous studies that have linked autonomy and jurisdiction as significant negative influencers of both diffusion and assimilation of innovations (e.g. Rogers, 1985; Fichman, 2000; Liang et al., 2007). My results provide a sound basis for further investigation using multinational firms and conglomerates with global operations, partnerships and/or alliances in future studies. However, both access to resources and coercive pressure positively moderated the relationship between organization scope and EA assimilation. This may mean that skilled resources and money can help overcome hurdles associated with organization scope limitations. It could also mean that coercion driven compliance may yield positive results in assimilation of an innovation even when encumbered by scope limitations. There was a strong positive correlation between coercive pressure and autonomy. This may suggest that the less autonomous organizations were more susceptible and submissive to coercive pressures than their autonomous peers. Another interesting result was that access to resources had a positive moderating effect on organization complexity and EA assimilation. The Salmans et al. (2009) study that found a link between larger IT budget and higher IT investments, with resultant increase in EA activities was of interest in this study. I had sided with Strassmann (2005), who cast doubt on the simple argument that large IT budgets would lead to improved IT activities. Like Strassmann I was convinced that without EA, which he described as “creating structure and order”, complexity and costs tend to rise, in the process eroding the big IT budget advantage. As stated earlier, Kappelman (2010) argued that EA is continually being deployed as a
139
strategy to tackle growing organization complexity. The results of this study backed up that argument. The moderation results show that an increase in organization scope will lead to a decrease in EA assimilation levels. This finding supports the Zhu et al. (2006a) finding that organizations operating in diverse geographies faced a daunting task in evenly assimilating innovations, largely due to uneven infrastructure and dissimilar legal, economic and political regimes. The moderation result showing that increased coercive pressure will lead to decreased organization scope was a rather fascinating result. This suggests that organizations can overcome encumbrances related to organization scope in order to accede to coercive pressure. This finding supports Suchman’s legitimation theory (1995) that organizations will endeavor to overcome all obstacles in an effort to secure legitimacy. Implications for Practitioners and Managers This research has unique implications for IS practitioners, especially for EA practitioners and managers. Many of this study’s findings offer guidance to management and EA practitioners. The moderating role of coercive pressure clearly highlights that decisive and sustained coercive pressure plays a significant role in assimilating EA in large and complex organizations. The research on IT assimilation remains sporadic and scarce, despite its high importance and reputation as problematic (Swanson, 2004). As organizations continue to grow globally and inorganically, so too are challenges in assimilation of IT and EA in particular. Contemporary organizational factors such as complexity, high industry regulations (targeted at individual business units) business unit autonomy and jurisdiction continue to pose significant challenges. Whereas prior studies 140
have extensively covered organizational level factors and their impact on assimilation of innovation, I could not locate a single study that has focused on factors that influence EA assimilation in particular. My study sought to test factors attributable to anecdotal evidence from case studies such as Ross et al. (2006), Kappelman (2010), Brynjolfsson and Kemerer (1996). The study was also interested in exploring the influence of coercive pressure (DiMaggio & Powell, 1983) on EA assimilation. My study results suggest that highly complex organizations may require higher degree of investment in EA, in addition to investment in simplification, in order to realize successful EA assimilation. There are several practical implications for IS practitioners. One is to ensure that they focus investments on tackling complexity for successful EA assimilation. Second is that EA assimilation planning should take into account jurisdictional factors (such as uneven landscape) regional infrastructure, politics, regulations as well as the impact of parochialism and cultural resistance. Thirdly, and most importantly, the results show that securing top management EA value appreciation is critical for securing vital resources necessary for EA assimilation. The results also show that investment in skilled resources may yield better results than funding in driving successful EA assimilation. Limitations Admittedly my sample size was small (n=123/ 106). Small sample sizes do not permit a researcher to detect low valued structural path coefficients (Chin & Newsted, 1999). However, considering that the U.S. government has the largest number of EA programs globally (163 programs), the study’s sample comprised 75% of the available federal programs for research in a uniform setting and context. Another major limitation 141
was the research setting - the U.S. federal agencies. The presence of private sector data may have probably produced different results. I also recognize that the government has the advantage in terms of capacity and mechanisms to sustain regulatory institutional pressure that private sector organizations may not. This may have made the results less generalizable. Park and Luo (2001) argue that government agencies still “exert significant influences on business policies and practices.” Future Research Future research may want to extend and/or replicate this study to cover private sector organizations, such as conglomerates and multinational corporations, which may be facing similar challenges in assimilating their EA. Most importantly, future research may want to examine the longitudinal effect of the factors on EA assimilation. For instance, what is the relationship between the predictive powers of the antecedents on the levels of EA assimilation over time? What are the differences in the predictive power of the antecedents before and after institution of coercive pressure?
142
CHAPTER IV: INVESTIGATING THE DETERMINANTS OF EA ASSIMILATION ACROSS ASSIMILATION LEVELS AND PHASES: A QUANTITATIVE LONGITUDINAL STUDY Introduction Being a radical administrative innovation, Enterprise Architecture typically progresses over long periods of time from its initial inception and assimilation to increasingly pervasive forms of assimilation. As Zhu, Kraemer and Xu (2006) argue it is important for Information System (IS) scholars to understand the key factors that influence IT innovation assimilation. IT Innovation assimilation has recently become a significant research topic, because only deep IT assimilation enhances operational efficiency and competitive agility (Zhu & Kraemer, 2002, 2005). In this paper, EA assimilation is defined as the depth of EA innovation integration and its pervasiveness in the organization (Ash, 1997; Sambamurthy & Zmud, 1996)5; higher EA assimilation levels reflect cumulative increases in the extent, the depth and comprehensiveness of the deployment of EA principles and policies. It is expected to result in broader and distinct organizational effects over time (Ross et al., 2006) and organizations can therefore realize EA innovation’s full potential only by increasing its assimilation levels (Wang, 2008). According to the extant literature, any innovation assimilation process is influenced by both internal and external factors (Rogers, 1995: 376–383). External elements often do have a significant effect on the rate and success of assimilation. Tolbert and Zucker, (1983); Zhu et al., (2006a); Liang et al. (2007) and Wang, (2008) all show that institutional and environmental pressures influence assimilation rate over extended 5
She defines it as “the extent to which the full potential of the innovation has been embedded within an organization's operational or managerial work systems” (Ash, 1997: 12).
143
periods of time. Internally, factors such as the organization’s IT and architectural capacity, or its internal capabilities for IT development have been shown to influence EA assimilation (e.g. Ross et al., 2006; Schekkerman, 2005). Capacity in the context of EA assimilation refers to the organization’s present technical and managerial abilities to deploy IT whereas capability is "future" oriented. It is what the organization is expected to do in future with EA. However, few empirical studies have shed light on the influence of either the effects of internal or external factors on the EA assimilation. For the purposes of our study EA assimilation is not only long process; it is also a punctuated one in that it is guided by distinct phases if viewed from external, environmental view point, and stages, if viewed from an internal view point. For the purposes of this study, an assimilation phase is an external, temporally focused way of punctuating the innovation trajectory of a given innovation in an adopter population. To this end it refers to characteristically distinct temporal (time-based) segments of the assimilation process generated by prevailing essential differences among the adopter population. An assimilation stage, in contrast, is a way to punctuate internally the assimilation process into distinct ‘environments” or ‘gestalts”. An assimilation stage therefore refers to a set of internal distinct and often designated and benchmarked capacities possessed by the adopting unit over its course of assimilating the innovation. These capacities are path dependent and can therefore be rank ordered as expressed in widely adopted EA assimilation scales. In addition, each assimilation stage assumes a distinct and unique set of skills and competency requirements that need to be fulfilled prior to progression to the next stage. Overall, EA assimilation phases and stages both are sequentially ordered (Rogers, 1995; Ross et al., 2006). 144
Extant meta-analyses of the innovation assimilation literature reveal the lack of unified theoretical frameworks that would predict the effects of EA assimilation determinants across phases and stages (Fichman & Kemerer, 1999; Zhu et al., 2006a; Ross et al., 2006). Curiously, I could neither locate a single empirical study that would have examined the influence of assimilation phases and stages on antecedents of EA assimilation- a research gap that this study will be seek to bridge6. These observations provide the motivation for this study. I want to develop a more integrated model to investigate how factors that influence EA assimilation change across assimilation stages and phases and test this model using a longitudinal panel data set. This study is thereby an extension of the Makiya and Lyytinen (2011) study, which used cross-sectional data to identify general internal and external determinants of EA assimilation. In this paper I will extend this study by conducting a longitudinal analysis of the determinants of EA assimilation on population level. I will specifically analyze the effect of changing environmental influence and to this end use the presence of coercive pressure as a moderator across EA assimilation determinants to evaluate its true effect across assimilation phases and stages. This paper has three separate but related objectives. The first objective is to examine the change of determinants of EA assimilation across phases in the same adopter population. I follow here Lyytinen and Damsgaard (2011) who argue that the attributes and characteristics of adopting units and populations differ and should thus be analyzed separately. The adopter population assimilation process is temporal, with both 6
Several studies exist that address internal temporal effects of innovation adoption e.g. (Rogers, 1995) and IS assimilation (Fichman & Kemerer, 1999), Only Ross et al. (2006) and Schekkerman (2005) captured (albeit anecdotally) the joint influence of internal and temporal (environmental) effects on the EA assimilation.
145
the adopted technology and set of potential adopters changing over time (Lyytinen & Damsgaard (2011) cf Boonstra & de Vries, 2008). I will examine in particular how the determinants have differential effects at different assimilation phases. That is, the same determinants may have “differently directional effects,” depending on the phases of assimilation within the population (Fichman, 2000). The second objective is to examine the specific effect of the presence or lack of coercive pressure on assimilation (Rogers, 1995; Zhu et al., 2006). The Makiya and Lyytinen (2011) study showed that coercive pressure influences EA assimilation directly and through moderation. However, the study did not examine the impact of coercive pressure within the different ‘phased’ populations and compare the effects of different environmental contexts where the pressure was present or not on how other factors influence assimilation. This study will therefore introduce coercive pressure as a moderator and compare the predictive power of the determinants pre and post promulgation of coercive pressure. The third objective is to examine the determinants of assimilation stage advancement within individual adopting units. Zhu et al. (2006) argue that existing research has primarily focused on a single assimilation stage - adoption hence little is known about the determinants of the other assimilation stages. Both Fichman (2000) and Zhu et al. (2006b) suggest that the post-adoption stages of assimilation are especially worthy of a focused study. The main assumption underlying this objective is that the cumulative effect of organization learning and internal capabilities development influences what determinants are significant at various assimilation stages (Fichman & Kemerer, 1999; Ross et al., 2006; Lyytinen & Damsgaard, 2011). Organizations learn 146
how to generate value from IT through cumulative capability development (Robey et al., 2000; Ross et al., 2006). These benefits multiply as the organization continues to expand its learning about the application of EA management practices through consecutive stages. Objectives (1); (2) and (3) have a common thread of recognizing the role of environmental attributes, knowledge as well as the skills as important contributors of the assimilation. The knowledge and skills elements include endogenous (cumulative capacity development) elements while exogenous factors cover things like access to resources or influences of isomorphism or coercive pressures. The changing environmental dispensation is also a learning attribute as the organization learns to incorporate new environmental elements into its response repertoires during the assimilation (Fichman, 1999; Schekkerman, 2005; Lyytinen & Damsgaard, 2011). Higher EA benefit realization is achieved by ‘educating’ the organization executives on the IT related business opportunities that ongoing EA assimilation and improved IT effectiveness offer. The literature suggests that the negative perception of EA dissipates, as the management learns from failures of others (Lyytinen & Robey, 1999). Motivated by the issues identified above, this paper seeks to study the following research questions: 1. What are the specific determinants of EA assimilation phases and how do their effects vary at different phases? 2. How does institutional change impact the EA assimilation process? 3. What are the specific determinants that affect EA assimilation stage advancement and how do their effects vary at different stages?
147
To address these questions, the study develops a theoretical model how phases, stages and institutional context influence EA assimilation. The model specifies five EA assimilation factors, five assimilation stages and three assimilation phases. The model is then tested using a unique longitudinal data set of 123 EA programs. The remainder of this paper is organized as follows. The next section reviews the theoretical foundation of the study. Section 3 describes the study’s conceptual models and hypotheses. Section 4 offers a description of the research design and data collection methodology. Section 5 reports on the hypotheses testing results. Section 6 is the discussion and conclusion, while section 7 reviews the limitations of the study. Finally section 8 makes proposals for future research. Theoretical Foundation Over the past twenty five years, innovation researchers have increasingly shown interest in organizational and institutional factors that influence the level of IS innovation. Bajwa et al. (2004) notes, however, that many researchers have proposed stage models for IS products and their assimilation without however, explicitly addressing some of the key determinants that may influence innovation behaviors during specific phases given the extended scope of innovations and the outcomes of these phases. This section revisits the theoretical roots of and empirical evidence belying the perspectives on temporal aspects of innovation diffusion and assimilation and then discusses appropriate theoretical perspectives to make sense of this phenomenon. To this end, I conduct literature reviews related to: (1) the theory of innovation assimilation; (2) impacts of environmental change on the assimilation; (3) the configuration analysis typology and temporal assimilation phases; (4) Environmental jolts and coercive 148
pressure; and (5) theory of EA assimilation stages. The literature reviews highlights the main arguments, perspectives and theories that pertain to the study’s theoretical foundation. The literature reviews for (4) and (5) are a continuation of the discussions in the Makiya and Lyytinen (2011) study. Configuration Analysis Typology Configuration analysis is a new approach devised by Lyytinen and Damsgaard (2011) for analyzing the adoption and assimilation of inter-organizational information systems (IOIS). Configuration analysis conceptualizes diffusion or assimilation as a complex evolutionary multilevel system that must be probed systematically. The typology introduces two key theoretical concepts, namely adopter configuration7 and adopter populations that this study draws upon. We will here only adopt the concept of adopter population. An adopter population is defined as the “set of all organizations that have participated (or could have participated) in at least one adopter configuration” (p. 5). Lyytinen and Damsgaard (2011) note that adopter populations are different at different stages of innovation assimilation and may have differentiating effects on the adopting unit and related determinants of assimilation. They observe that the scope and size of the adopter population depends on the technological, temporal and institutional elements, a view this study will explore further by examining three adopter populations at three different time points. Next I conduct a review theoretical perspectives on IS assimilation.
7
An adopter configuration is defined as a “set of interrelated adopters united by an organizing vision and associated key functionality, which determine the structure, mode of interaction and appropriation available for the participating organizations” (p. 4).
149
Theoretical Perspectives on IS Assimilation Assimilation of innovations not only takes often a long time but also occurs in periodic phases (e.g. Fichman & Kemerer, 1999, Tolbert & Zucker, 1983; Wang, 2008). Studies have shown recession or increased significance of determinants of assimilation as time progresses. Tolbert and Zucker, (1983) for example conducted a study on a group of factors that influenced adoption of civil service reform in U.S. cities over a four phases using data on the adoption of civil service reform by cities from 1880-1935. They found that some groups of factors lost their predictive power as the assimilation process progressed. The results showed that whereas city characteristics were strong predictors of assimilation in the early periods, the later assimilation periods were instead related to institutional definitions of legitimate structural form, rendering city characteristics ineffective predictors of the adoption. Wang (2008) conducted an 11 year longitudinal study on the effects of external pressures on the assimilation of ERP systems in Fortune 1000 companies. He examined both the effects of pressures from organizations' exchange partners and pressures from its institutional environment. He found that several factors receded or had significant effect at different periods of time. The results also showed that between 1997 and 1999, three factors associated with ERP assimilation turned significant one after another, while those related to the institutional environment were only significant in and after 2002. Other key variables including the study’s control variables were significantly associated with assimilation in 1996-1998, but not afterwards. Wang’s study provides important insights into the role that critical role of environmental change, particularly the institutional change on the assimilation process. 150
This finding is partially supported by the Makiya and Lyytinen (2011) study which found that coercive pressure had significant influence on the EA assimilation process. However, the study neither addressed the effect of population characteristics nor its influence on the factors that influence individual units. This paper will address this void by examining the interaction between population characteristics and individual units. This approach is justified by the Lyytinen and Damsgaard (2011) argument that population characteristics interact with the determinants of individual adopter units - that is, characteristics of assimilation phases influence the determinants of EA assimilation. Borrowing from these previous studies, this study defines three temporal phases of assimilation. These three separate time periods will be used as proxies for three conceptual phases related to assimilation at the population level. The next section introduces, defines and characterizes these assimilation phases. Assimilation Phases There are three theoretical assumptions underlying the idea of identifying assimilation phases. The first is the assumption that as time progresses from the initial to deeper levels of assimilation, there is more widespread adoption and use of EA among all organizations. This leads to better knowledge sharing and understanding of the innovation (Rogers, 1995; Fichman & Kemerer, 1999; Lyytinen & Robey, 1999; Lyytinen & Damsgaard, 2011). For example many firms fail to achieve deep usage beyond initial adoption in the early stages due to lack of related knowledge and external support (Chatterjee et al., 2002; Lyytinen & Damsgaard, 2011). The second assumption is that the characteristics and nature of the EA will change over time, as new enhancements are incorporated into the tools and technologies based on learning by doing (Fichman & 151
Kemerer, 1999; Ross et al., 2006; Lyytinen & Damsgaard, 2011). Hence the barriers related to assimilation will be different at different phases. The third assumption is that as diffusion advances, so too do the deployment techniques and methodologies via introduction of frameworks and assessment mechanisms by new industry players (Fichman & Kemerer, 1999; Lyytinen & Robey, 1999; Weill & Ross, 2003; Ross et al., 2006; Lyytinen & Damsgaard, 2011). This all suggests that the environment and the effects of distinct variables influencing EA assimilation may change significantly over time (Lyytinen & Damsgaard, 2011). Consistent with earlier technology assimilation literature (e.g. Fichman & Kemerer, 1999; Zhu et al., 2006) the study’s initial temporal phase is adoption – defined as making the decision to use the innovation for value chain activities (i.e. allocating resources and physically acquiring the technology). However, the author disagrees with Fichman and Kemerer’s (1999) description of this phase. They view it internally as a “single event in the assimilation process – typically acquisition or commitment to use the innovation” (p. 10). The literature suggests that this phase is characterized by a number of activities and outcomes and is largely influenced by industry or peer innovation adoption (Zhu et al., 2006). Zhu et al. credit such environmental influences mainly from the industry and business entities as key drivers of assimilation at this phase. For instance, Teo, Wei, and Benbasat (2003) found that firms’ adoption of network technologies was significantly influenced by increased adoption by peers within the same industry. However, adoption decisions can also be imposed. For instance, Lyytinen and Damsgaard (2011), note that organizations may adopt an industry standard, because they may have been mandated to adopt it as part of a national imperative (community wide 152
configuration). Whatever the adoption context, the literature suggests that this phase is typically marked by inconsistency, ambiguity and chaos. For instance Fichman and Kemerer (1999) write that after adoption, the firm and its members usually do not have sufficient knowledge to leverage the system, and misalignments occur between the new technology and the user environment. Lyytinen and Damsgaard (2011) write that new standard is often an insurmountable obstacle, and that many adopters never manage to evolve beyond their original vision and structure, and subsequently, a large number of ‘false starts’ are needed before a viable innovation emerges victorious. Extant literature reviews suggest also that at this phase organizations usually encounter uncertainty around mandatory and regulatory requirements (e.g. DiMaggio & Powell, 1983). Due to the uncertainty, managers struggle to mimic successful ones; hence mimetic pressure often drives the assimilation. An exemplification of mimetic pressure is captured by the Gartner Hype curve, which uses hype cycles to show how and when technologies move beyond the hype and eventually becoming widely accepted. According to Gartner, hype cycles enable executives to decide on suitable entry points for adoption, usually by learning from and eventually mimicking successful early adopters. The second phase is the institutionalization phase – defined here as the period when the innovation is forcibly infused and instituted into the adopter population (Hoque & Hopper, 1994; Zhu et al., 2006; Liang et al., 2007). The literature shows that at this phase organizations are under pressure to invigorate their innovation assimilation process due to extant knowledge available and pressure from their peers. The trigger for the invigoration is often duress or pressure from the regulatory bodies or the emergence of an early majority of adopters that demonstrate its value (Ansari, Bell, & Lundblad 1992; 153
Rogers, 1995). Rogers (1995) identified five adopter categories: innovators, early adopters, early majority, late majority, and laggards. Rogers characterizes early adopters as opinion leaders who are widely respected in their social circle. He describes the early majority as “deliberate” adopters, the late adopting majority as “skeptical” about the value of an innovation, and laggards as “traditional” adopters. Early adopters serve as opinion leaders who eventually persuade others to adopt the innovation by providing subjective evaluative information (Rogers, 1995). As the executives from late majority and laggard groups begin to understand the value of the innovation and get the sense of missed opportunities they resort to forcibly driving adoption, mainly via coercive pressure (Teo et al., 2003). The third temporal phase is routinization - defined as the phase where innovation (EA) has become widely used (late majority has joined) and where it eventually becomes an integral part of the most organization’s activities (Fichman & Kemerer, 1999; Zhu et al., 2006). This is similar to achieving a status where for a majority of organizations routinization has become the status of the innovation (DeLone & McLean, 1992). Extant literature suggests that assimilation activities at this phase are often influenced by the forces of normative isomorphism (DiMaggio & Powell, 1983). In the context of this study, normative pressure can be attributable to professional or peer influence and can be driven by the desire for acceptance by others within the population of those who have not adopted the innovation. It is therefore expected that as organizations begin to closely resemble each other in the population late adopters will enjoy the advantage of widespread knowledge of the innovation and experiential economies of scale (Tolbert & Zucker, 1983). 154
EA Assimilation Stages The EA assimilation lifecycle includes complex sets of tasks such as adaptation of existing IS, business process redesign, and other far reaching adjustments to organizational structure, all of which may have a unfavorable impact on the prevailing and future authority and decision-making structures (Barua et al., 2004, Zhu et al., 2004, Chatterjee et al., 2002). Because EA forms internally an integrative and radical transformation how an organization views and deploys its IT assets and capabilities, EA deployment progresses in its assimilation life cycle internally through distinct assimilation levels. These levels capture the depth and scope of deploying EA principles and integrating related behaviors into new organizational routines. Overtime organizations must demonstrate capacity to effectively execute on elements within one assimilation stage prior to engaging in higher level EA assimilation activities. As Ross et al. (2006: 86) argue, learning through the architecture stages encompasses gradual capability development in technology and business processes, investment evaluation leading to IT enabled change and process design. Learning is path dependent while at different stages different learning content and scope becomes relevant for the innovation to move forwards. At the same time increased assimilation is viewed as critical to an EA program’s effectiveness (Bittler & Kreizman, 2005). Several taxonomies of stages for EA assimilation have been proposed. Each such stage is defined through activities that characterize that assimilation stage (Ross et al., 2006). These sets of activities aim at improving the IT organization’s effectiveness, mapped against identified clear business change and transformation capabilities. For instance, as the assimilation reaches the advanced stages the focus shifts towards change 155
management and business transition frameworks while the early stages emphasize activities that increase efficiency of resource utilization. As the company advances through the assimilation stages, its IT strategy takes on increased importance as the business value of IT increases (Ross et al., 2006). Ross et al. also argue that new types of EA capabilities assessment become critical in establishing the organization’s ability to implement technologies and innovations that are pivotal for its strategy. In the next section I briefly define each assimilation stage and provide high level characteristics and activities associated with each stage. Key stages are depicted in Table 20. TABLE 20: EA Assimilation Stages Stage
Description/characteristics
Stage 1 Establishment of EA institutional commitment and direction
This stage is characterized by unpredictable, rather ad-hoc and unstructured investment processes (Schekkerman, 2005). Organizations perceive the role of IT as being to automate specific business processes. At this this stage use of IT is chaotic, ad hoc, and heroic. It forms the starting point for an evaluation ofnew processes often brought about by automation (Szyszka, 2009; Ross et al., 2006). IT Investments focus on solution delivery for local business problems and opportunities. IT investments are justified on basis of cost reductions and ROI. This stage is characterized by the establishment of basic technology and project selection capabilities, driven by development of selected criteria such as project selection, benefit and risk evaluation, and an awareness of the organization’s priorities (Schekkerman, 2005). This stage includes development of capabilities such as project management methodologies, and process discipline (Szyszka, 2009). IT Investments shift from local applications to shared infrastructure. Management of technology standards is key to this stage. Investments are justified on the basis of cost reductions and compatibility. The most dominant practice at this stage is the deployment of enterprise resource planning (ERP) packages as the core of operations and business process standardization. This stage witnesses radical business transformation and new governance models. The focus is on the establishment of IT investment portfolio selection and criteria as well as alignment with the organization’s mission, strategies and goals, followed by specific project outcomes (Schekkerman, 2005). IT investments shift from local applications and shared infrastructure to enterprise systems and shared data. Investment justified on the basis of digital options, standardized processes and flexibility.
Stage 2 Creation of the management foundation for EA development and use
Stage 3 Optimization of IT via development and use of EA products for enterprise transformation
156
Stage 4 Expanding and evolving the EA and its use for targeted business results
Stage 5 Leveraging the EA to manage change management.
This stage is characterized by enhanced business modularity and strategic agility through customized or reusable modules (Ross et al, 2006) using evaluation techniques to improve IT investment processes and portfolio (Schekkerman, 2005). Senior executives are actively involved in identifying, harnessing and alignment of their organization’s IT capabilities and turning them into possibilities based on the business strategy. IT investments shift towards achieving strategic agility through customized or reusable modules. Seamless linkages between business process modules and greater autonomy/discretion at business unit level for building or buying modules. Investment justified on the basis of business impact and digital options. At this assimilation stage organizations leverage EA to manage change. At this stage, the IT organization is firmly positioned within the corporate levels of the organization and IT is active player in business strategy formulation. Organizations at this stage have written policies governing IT investments and alignment with organization strategy and mission are paramount. Senior leadership approval is sought for the EA products. Also at this stage, the enterprise tracks and measures EA benefits or return on investment, and adjustments are continuously made to both the EA management process and the EA products
This next section conducts literature reviews on the effect of environmental jolts on assimilation of innovation Theoretical Perspectives on Environmental Jolts Several studies have established that institutional environmental jolts are a critical factor in the assimilation of radical innovations (Tolbert & Zucker, 1983, Hoque & Hopper, 1994; Wang, 2008). The term coercive pressure means that organizations are being coerced into moving in the prescribed direction over time. This requires not only identification of the required direction but also the ‘target’ change destination (Ashworth, Boyne, & Delbridge, 2005). Thus coercive pressure from an institutional perspective suggests that firms or organizational units are at some point of time susceptible to compliance pressures for conformity and the need for legitimacy and they will be punished if they do not do so (DiMaggio & Powell, 1983; Meyer & Scott., 1983). The Makiya and Lyytinen (2011) study indicated that coercive pressure moderated the relationships between EA assimilation and 1) management value recognition and 2) 157
organization complexity. However, the study did not examine the impact of sudden environmental change where the coercive pressure was suddenly made present during the assimilation process. A number of researchers highlight the coercive influence of government directives and regulations on an organization’s behaviors in a variety of different environments (Ansari, Bell, & Lundblad, 1992; Hoque & Hopper, 1994). The Hoque and Hopper study indicates that coercive pressure emerges suddenly and unexpectedly and may at times cause more confusion than aid the acceleration of the assimilation process. Indeed we know very little of the effects of such environmental jolts on the assimilation process. In this study, I will extend the findings of the Makiya and Lyytinen (2011) study to cover the three temporal phases and to tease out the differentiating effects of introducing coercive pressure during the assimilation process. I will also control for the different levels of normative and mimetic pressure by examining the effect of the presence of coercive pressure over three phases with the intent to capture the impact of abrupt change during the assimilation process as the regulatory environment changes suddenly (Lyytinen & Newman, 2008). This next section introduces and describes 1) the research model and 2) the study hypotheses Research Model and Hypotheses I will next develop and test three separate conceptual models as shown in Figures 15, 16 & 17 to address the following questions: 1. What are the specific determinants of EA assimilation phases and how do their effects vary at different phases? 2. How does institutional change impact the EA assimilation process?
158
3. What are the specific determinants that affect EA assimilation stage advancement and how do their effects vary at different stages? The first model (Figure 15) will be used for addressing research question 1. The study will focus on the impact of temporal environmental changes as EA progresses through the three temporal phases (t1, t2 and t3). The second model (Figure 16) will be used to examine the influence of coercive pressure on the determinants of EA assimilation. The study will introduce coercive pressure as a moderator variable and test its effect on the relationship between the determinants and EA assimilation across the three temporal assimilation phases. The third model (please see Figure 17) will be used to address research question 3. The objective is to examine the determinants that influence advancement to the higher level EA stages. In general I posit eleven hypotheses that will guide the prospective statistical analyses. Five hypotheses are associated with the first model, three are associated with the second model and three are associated with the third model. In the next section, as I develop the formal hypotheses related to the model, I provide further justification for the logic belying the predictions or conjectures. RQ I: The Change of Determinants of EA during Distinct Assimilation Phases The research model draws upon the determinants of EA assimilation identified in the Makiya and Lyytinen (2011) study. However, the model extends the study to cover three temporal phases: t1 (adoption), t2 (institutionalization) and t3 (routinization) as outlined above EA determinants that are significant during a particular phase can be expected to influence activities undertaken in that phase, and/or promote the successful progression 159
of EA assimilation of that phase. In other words, the variations in determinants in the positive direction could be viewed as predicting success of EA assimilation, while variations in the negative direction create barriers that prevent deeper EA assimilation. To this end I will specifically examine the effects of determinants of EA assimilation identified by the Makiya and Lyytinen (2011) study at the three temporal phases: adoption phase t1 (1999-2001); institutionalization phase t2 (2002-2004) and routinization phase t3 (2005-2007). The study’s determinant include : organizational complexity, organization scope, parochialism and cultural resistance and management value perception (see Figure 15). The research model is composed of some second order constructs - access to resources, organization complexity and scope - along with some first order constructs - management value recognition, as well as parochialism and cultural resistance. I thus surmise that these antededents capture the main causes affecting EA assimilation as identified in the past innovation studies during the three phases (Fichman, 2000; Kraemer et al., 2006; Liang et al., 2007; Wang, 2008).
160
FIGURE 15: Research Conceptual Model 1
I intend to postulate direct effects of the determinants on each assimilation phase. I will then compare the effects of these antecedents at the three temporal phases. In general I posit eleven hypotheses associated with the effects of identified factors on the EA assimilation process. As I next develop the formal hypotheses related to the model, I will further justify the logic of how the determinants affect the level of EA assimilation given the assimilation phase. Therefore, I will next review the content of each construct and reasons for its inclusion in the model and state associated hypotheses.
161
TABLE 21: Study Constructs and Definitions Constructs
Access to resources
Organization Complexity
Organization Scope
Parochialism and cultural resistance Top management value recognition Control variable Size
Definition The extent to which the organization is capable of securing adequate resources including human capital and money to successfully drive assimilation. The access to resources construct has two first order constructs: funding and skilled resources. Funding is defined as the availability of adequate financial resources for EA assimilation. Skilled resources is defined as the availability of human capital with the technical, political and managerial know-how necessary for bridging the knowledge gaps between the executive and the technical arms of the organization. The amount of differentiation among different elements constituting the organization for example the number of missions, functional units and the number of regulations, as well as their mutual dependencies. The organizational complexity construct is multidimensional and has three causal components: mission heterogeneity, administrative multiplicity and finally legal and regulatory framework. Mission heterogeneity refers to the degree of dissimilarity of mission areas or mission focus within an organization. Administrative multiplicity refers to the extent to which administrative and decision-making authority is dispersed within an organization. Legal and Regulatory framework is defined as the environmental and regulatory restrictions of the organization, including government and consumer regulations. The extent and reach of operations affecting the organizations that reside outside its formal boundary. Organization scope is defined through two components: jurisdiction and autonomy. Organization jurisdiction is defined as the span of geographic and operational authority including administrative, technological, political and economic. Autonomy is defined as the degree to which an organization is free to make decisions with respect to its own operations, resources and budgets. The variance of and attitude towards perceptions of changes in the environment often denoted as the “selfish pettiness or narrowness” with regard to external interests, opinions or views The perception of the value of an innovation by the management team
The number of employees in the organization
Access to Resources Evidence shows that financial resources are linked to the organizations’ capacity for assimilating innovations (Fichman, 1999). Financial and skilled human resources are important ingredients for an organization’s capacity. However, the impact of resource abundance or scarcity is felt most at the early assimilation phase. Adoption of a risky innovation in its early adoption is a financially intensive undertaking. The literature suggests that most innovation projects fail due to financial constraints, or lack of skilled resources to implement them at this stage (Rogers, 1995; Fichman & Kemerer, 1999; Zhu et al., 2006). In line with this Downs and Mohr (1976) found that financially strong 162
organizations are particularly well positioned to adopt and assimilate high cost innovations. Several studies have shown that commensurate with decreasing complexity and improved technical depth, the importance of money and skilled manpower decrease in importance as the IS assimilation advances. An organization’s financial capacity, whether weak or strong has been shown to strongly influence its access to vital resources for investments in critical technical skills such as consulting and advisory services (Attewell, 1992). Two separate studies established important links in support of Attewell’s claim. Chatterjee et al. (2002) established that the level of investments in skilled staff is linked to successful innovation assimilation. Katz and Shapiro (1986) and later Salmans et al. (2009) established a positive link between fluid funding of IT and greater benefits realization. Based on the above arguments, I posit that access to resources will have significant determinant power at the initial phase of assimilation, but then dwindles in significance as the assimilation advances. Hypothesis 1a. Access to resources will have a positive but declining effect as EA assimilation progresses through the assimilation phases. Parochialism and Cultural Resistance Studies have shown that parochialism and cultural resistance are particularly strong upon introduction of a new IS innovation, with the resistance mainly coming from employees and managers with sentimental attachment to certain technologies and systems as well as established behaviors (Boddy, Boonstra, & Kennedy, 2003). For instance Ross et al. (2006), write that the expected benefits of EA assimilation may be delayed as the managers will not only have to learn new behaviors, but ample time is required to learn and adjust to the new behavioral dispensation. Evidence shows that parochialism and 163
cultural resistance are usually high when the assimilation is undergoing internalization and new organization structures and behaviors take root. According to Ross et al (ibid), at later phases of assimilation, in order to fill knowledge gaps, organizations hire external resources such as consultants or employees with advanced knowledge of the technologies or systems. The resulting loss of power coupled with a plethora of unwelcome changes particularly in operations or organization structure are expected to further fuel the retarding effect of parochialism and cultural resistance. It is on this basis that I do expect parochialism and cultural resistance to have predictive power throughout the assimilation lifecycle, hence: Hypothesis 1b. Parochialism and cultural resistance will be inversely related to EA assimilation levels at assimilation phase t1 and t3. Top Management Value Recognition This study’s prediction is based in part on literature reviews which suggest that managers tend to experience considerable ambiguity about the value of new innovative technologies (Weick, 1990). Failure to recognize an innovation’s value may be a reflection of top managements’ lack of knowledge of how to implement it (Fichman & Kemerer, 1999). The responsibility bestowed upon the managers is also predicted to eventually dilute appreciation of the innovation. Salmans et al. (2009) argue that it is not enough for the top management to simply tout the value appreciation of an innovation but must also assume ownership and ensure its successful implementation. Tallon, Kraemer and Gurbaxani (2005) found that top managers played an important role in sustaining momentum at later phases of innovation diffusion. However, few managers appreciate the duress and sustained scrutiny, a factor that I conjecture will impact value recognition 164
at later phases. It is therefore predicted that the sustained coercive pressure will diminish the impact of value recognition of EA. It is thus expected the effect of management value recognition will wane as EA advances from phase t2 to t3. Therefore: Hypothesis 1c. Management value recognition will have a positive but decreasing effect on assimilation levels across all assimilation phases. Organizational Complexity Organizational complexity is defined as the amount of differentiation that exists within different elements constituting the organization. As EA assimilation advances, managers will struggle to manage ubiquitous change and increasing complexity within their strategic and tactical environments (Kappelman, 2010). The study posits that organizational complexity will remain a significant antecedent throughout the phases of the assimilation lifecycle. At the adoption phase, complexity is associated with structural and technology impediments (Rogers, 1995). For instance duplicity in business units, managerial tasks and technology implementation. At the institutionalization phase, it is expected that coercive pressure from the regulator will increase the number of regulations and compliance requirements, increasing complexity (Hoque & Hopper, 1994). At the routinization phase, it is expected that organization changes associated with the new operating models and processes will exacerbate complexity (Ross et al., 2006). It is thus my expectation that organization complexity will influence EA assimilation throughout the assimilation lifecycle, thus: Hypothesis 1d. Organization complexity will be inversely related to assimilation levels across all assimilation phases.
165
Organization Scope The literature suggests that organization scope will be a determinant at the adoption phase due to EA’s potent threat to existing authority and autonomy structures (Ross et al., 2006; Lusthaus, Anderson & Murphy, 1995; Datta & Nugent, 1986). At the institutionalization phase, it is expected that jurisdiction will emerge as the strongest barrier as changes to existing regulatory frameworks pave way for new legal and regulatory requirements (Zhu et al., 2006). At the routinization phase, it is expected that EA’s new organizational and governance structures will run into barriers such as decision rights (Weill & Ross, 2003). I thus posit that: Hypothesis 1e. Organization scope will be inversely related to assimilation levels across all assimilation phases RQ 2: The Impact of Coercive Pressure This part of the study will address research question 2, whether introduction of coercive pressure influences the predictive power of the determinants of EA assimilation. This way the study will tease out in particular the effects of introducing coercive pressure during the latter two phases (t2 and t3) which helps compare and contrast the effects of the EA assimilation determinants pre and post coercive pressure promulgation.
166
FIGURE 16: Research Model 2
Based on the findings of the Makiya and Lyytinen study (2011), as well as extensive literature (e.g. Granlund & Lukka, 1998), I do predict that coercive pressure will negate the impact of organization scope. Under coercive pressure, managers from various regions and business units will be strong armed by the regulator into driving assimilation. I also predict that coercive pressure will negate the effect of access to resources as funding decisions will be driven by the regulator and not local managers. Finally coercive pressure will negate the effect of parochialism and cultural resistance as managers will be under intense and sustained scrutiny and pressure from the regulator and will ignore these personal reasons for not engaging in the innovation. However, the 167
Hoque and Hopper (1994) study showed that the sustained coercive pressure exerted by regulators eventually led to voluntary routinization of the imposed activities and operations at later assimilation phases. The Hoque and Hopper study provides an important theoretical lens for formulating the next set of hypotheses. I posit that coercive pressure loses potency over the long term, i) it will have strong impact only at phase t2 but then ii) lose much of its impact at phase t3. Therefore: Hypothesis 2a. Coercive pressure will significantly reduce the inverse effects of access to resources, parochialism and cultural resistance and organization scope, at t2 but will have decreased effect at t3. It is expected that the regulator will successfully strong-arm top management to use EA at t2 for driving down IT costs and compliance with regulatory mandates. Regulators imposing pressure on subject organizations tend to hoist up and reward compliant managers, making it attractive for reluctant or skeptical managers to acquiesce to the regulator for career advancement or survival. However, such a strategy may not be sustainable, as the innovation becomes routinized within the organization. It is expected that at t2 most senior managers will be able to use IT to articulate value of EA using the language of the business to executives - a feeble but positive sign that management value recognition is taking root. Ross et al. (2006: 74) write that business managers begin appreciating the value of EA as it standardizes IT, reducing risk, and the cost of shared services, reliability, security and improvements in development time. As the benefits become apparent, business unit managers begin recognizing and appreciating the value of EA. At t2, coercive pressure will equally coerce managers into cooperation and collaboration enhancing decision-making in critical areas such as portfolio selections (Schekkerman, 2005). The Ansari et al. (2002) and Meyer and Rowan (1977) studies 168
indicated that organizations experiencing coercive pressure have to continually demonstrate to the regulators that they are acting on collectively valued purposes. As the organization moves to t3, it is expected that the effect of coercive pressure will diminish, as managers simply view EA as a compliance exercise and not as a value add tool. At this phase, most managers begin delegating rather than continuing with the “hands on” approach witnessed at t2. It is well documented that organizations experiencing coercive pressure for elongated periods of time eventually succumb and adhere to regulatory requirements in a perfunctory and detached fashion (Meyer & Rowan, 1977). The Liang et al. (2007) and Bajwa et al. (2004) findings that, at later assimilation phases some top managers dutifully participate in the assimilation process despite having deep seated skepticism leads me to conjecture that coercive pressure will lose potency at t3. Hypothesis 2b. Coercive pressure will significantly amplify the positive effects of management value recognition at t2 but then have decreased effect at t3 The Hoque and Hopper (1994) and Liang et al. (2007) findings that top managers can be successfully coerced into driving assimilation leads to my prediction that coercive pressure will positively enhance management value recognition of EA. Liang et al. (2007) found that regulatory satisficing eventually converts top managers into effective channels for the coercive pressures. Liang et al. (2007) also argue that the continued evolution and institutionalization of an IT innovation eventually leads to increased coercive pressures in the forms of new laws, regulations, professional and industry standards, thus exacerbating complexity. However other studies (e.g. Meyer & Rowan, 1977; Greve, 2000) have shown that organizations experiencing coercive pressure have to 169
continually demonstrate to the regulators that they are acting on collectively valued purposes in a proper and adequate manner. The Liang et al study also found that top managers' thoughts and actions were effective channels for advancing institutional compliance and that some type of coercive pressure eventually shaped top managers' participation in the ERP assimilation process thereby positively affecting its assimilation. Hence, even complex organizations will be successfully coerced into driving and assimilating EA. Results from the Makiya and Lyytinen (2011) study indicate that coercive pressure positively moderates the relationship between EA assimilation and organization complexity. Leaning on this finding, I posit that coercive pressure will successfully impose standardization and contain structural inertia thereby enhacing assimilation at t2, but having diminished impact at t3. Therefore, I posit that: Hypothesis 2c. Coercive pressure will significantly reduce the inverse effect of organization complexity at t2 but then have decreased effect at t3 RQ 3: What are the Determinants of Movement Between Assimilation Stages? This section examines the determinants of moving between EA assimilation stages as EA advances from one stage to the next. It is important to note that for this study the previous assimilation stage will also be used as a control in addition to being used as the dependent variable. This will enable the use of the current assimilation stage as a control in order to analyze the effect of other variables in moving to the next stage. Assimilation stage 1 will also be held as the default assimilation stage, meaning that the results will not be reported for this stage. The study will also use t1 as the default and as a control for phase t2. The limitations within the dataset are behind this approach. For instance at t1, the number of 170
EA programs at the high levels of assimilation were negligible, meaning that the study would run into n-power restrictions. In order to account for the environmental effect, the study will run two sets of analysis. First the study will run the regressions exclusive of coercive pressure moderation for t2 and t3. This will enable the examination of the direct effects of the determinants on the assimilation stages. Then, the study will introduce coercive pressure as a moderator and examine the impact of its inclusion on the predictive power of the determinants. Overview of the Research Model This research model reuses the constructs from the previous quantitative study. The research model is grounded on the concept of assimilation stages and has been formulated with an eye for detection of significant organizational factors that influence the movement between assimilation stages. The model posits that: organizational complexity, organization scope, parochialism and cultural resistance and management value perception are major factors affecting movement between EA assimilation stages. I posit that the same factors may have “differently directioned effects,” depending on the stages of assimilation (Fichman, 2000).
171
FIGURE 17: Research Conceptual Model 3 Access to Resources
Coercive Pressure
EA Assimilation
Funding Skilled Resources
Parochialism
Stage 5
Stage 5
Stage 4
Stage 4
H3c -
H3d+ Value Recognition
H3e+ Stage 3
Organization complexity
H3a-
Mission heterogeneity Administrative multiplicity Legal & Regulatory Framework
Organization scope
Stage 3 Size
H3
Stage 2
Stage 2
Stage 1
Stage 1
t2
t3
t1
t2
b-
Jurisidiction Autonomy
EA stages
The next section provides the formal hypotheses related to the model, and provides justification for the logic of how the determinants affect the EA assimilation stages. Hypotheses H3a-3e The following hypotheses predict that some of the determinants will either gain or lose predictive power as the assimilation process advances from one stage to the next. Access to resources is predicted to wield predictive power only at the early stage of EA assimilation. The literature suggests that early stages of assimilation have high investment requirements (such as funds for technology acquisition and acquisition of 172
skilled human capital), making access to resources a strong determinant factor. At the early stages of assimilation, financial benefits are hard to demonstrate. Not only are skilled resources such as consultants expensive, retention of old employees also bloats the costs. As Ross et al. (2006: 73) argue, it “is not the frustration of isolated systems that usually drives management from this (first) stage. It’s the cost.” Ironically, IT investments are usually justified on the basis of cost reductions. At this stage though, business silos compete for capital funding using locally focused cost-benefit analysis (Ross et al., 2006: 73). The struggles associated with financial value are expected to continue at both stages 2 and 3. According to Ross et al. (2006), IT investments shift from local applications and shared services to enterprise systems and shared data and so too does the financial clout of business unit IT leadership. Thus: Hypothesis 3a. Access to Resources is only positively related to EA between assimilation stage 2 and 3 at both t2 and t3 The earlier research study suggests that organizational complexity is very influential throughout the assimilation lifecycle. This result was hardly surprising, in light of the abundance of empirical support. Majority of the literature reviewed suggests that complexity will be a major impediment as EA advances from stage one to two for instance, as the organizations make the transitions from the silo operations to the standardized operation mode (Ross et al., 2006). Studies have also shown that competencies and skills garnered at lower stages such as stage 2 and 3 often prove insufficient to sustain momentum at advanced assimilation stages. For instance the Ross et al. (2006) typology suggests that organization complexity is particularly potent upon advancement from stage 1 to stage 2 as the organization changes focus from individual 173
business unit focus to enterprise-wide standards. The typology also suggests that upon advancement to stage 3 multi-business unit and organization unit processes get optimized and organization changes start to take root. According to extant literature a key activity that contributes to complexity’s determinacy at advanced stages is increased IT outsourcing, with organizations with weak internal capacity and capabilities looking to leverage outsourcing companies’ advanced capabilities. The literature also suggests that complexity retains determinacy at stage 4, as modularization and flexibility grow both locally and globally (Ross et al., 2006). However at stage 5, complexity manifests itself in disruptive organization change and investment prioritization (Bernard, 2005). Parochialism and cultural resistance are predicted to be determinants at all stages of the assimilation lifecycle. According to Ross et al. (2006), the local business focus at stage 1 renders the transition to stage 2 difficult as managers accustomed to the old order lead resistance to changes. The most prevalent form of parochialism and/or cultural resistance at stages 1, 2 and 3 is in the arena of decision rights (Weill & Ross, 2003). The IT governance structures at the lower EA stages place considerable clout in the hands of business unit managers. This clout comes under threat as the organization moves up the assimilation levels and new governance structures dilute their influence. According to Schekkerman (2005) at stage 2, the organization carries out radical changes such as the “de-selection” of obsolete, high risk or low value IT investments” which may not be popular within the organization. As the organization advances from stage 2 to 3, (Ross et al. 2006), the management focus on optimization often exposes skill limitations forcing companies to either outsource or hire more external resources, which doesn’t bode well for parochial interests. According to Schekkerman, (2006) the advancement from stage 3 174
to 4 often results in management engaging in high risk strategic experiments as the company tries to modularize its IT, in the process introducing uncertainty – an environment not well suited to staff with parochial or cultural biases. This could further entrench parochialism and cultural resistance as the organization implements new organization structures associated with advanced EA stages. Therefore: Hypothesis 3b. Parochialism and cultural resistance and Organizational complexity are inversely related to EA between assimilation stages 2 and 3 and stages 3 and 4 at both t2 and t3 The literature suggests that organization scope will only have determinacy at stage 2. Due to the siloed nature of the IT operations, jurisdiction and autonomy are expected to be critical factors when EA advances from stage 1 to stage 2. Ross et al. (2006) write that the silo mentality characteristic of stage 1 creates an insular operational and budgetary focus. At stage 2, even as the organization moves to standardized operations, autonomy and jurisdictional clout will be challenged via legal and regulatory requirements. It is predicted that the transition from stage 1 to 2 will encounter setbacks as jurisdiction and autonomy will continue to weigh down assimilation progress placing highly autonomous and jurisdictionally inclined business units at loggerheads with the regulators. Thus: Hypothesis 3c. Organization scope is only inversely related to EA between assimilation stage 4 and stage 5 at both phase t2 and t3 Research Design and Data Collection Research Context: Federal EA Programs The empirical context for this paper is the U.S. federal government. The United States government has globally the most wide-scale EA implementation, with over 163 175
EA programs (Bernard, 2005). I chose this sample and context, because it was the best suited for the research objectives and the research model for the following reasons: 1) the data enables testing of all eleven hypotheses 2) public availability of data for most of the factors, 3) EA was formally introduced as an administrative innovation, 4) the U.S. government has the capacity and mechanism to institutionally enforce mandate compliance. Research Methodology The hypothesized research models (Figures 15, 16 & 17) were tested empirically using three separate statistical analyses. The first study examined the direct effect of the determinants of EA assimilation over three phases (t1-t3) and conducted a comparative analysis of the three statistical outputs. The second study examined the impact of coercive pressure on the relationship between EA assimilation and determinants at phase t2 and t3. The third study examined the factors that influence EA assimilation stages over the three assimilation phases. Below are the descriptions of the three studies RQ 1: The Determinants of Assimilation Phases The study initially tested the direct effects of the determinants of EA assimilation at the three phases without the influence of the moderating variable, coercive pressure. The study controlled for past assimilation phases – for instance when regressing for time point t2, the study used t1 as a control and likewise for regressions at t3. For this part of the study, ordinary least squares (OLS) regressions was the preferred regression technique.
176
RQ 2: The Impact of Coercive Pressure on Determinants of EA Assimilation The study introduced coercive pressure as a moderator variable at phase t2 and t3 to examine the effect of its presence on the determinacy of the variables. The study used OLS regressions and also controlled for previous temporal phases RQ 3: The Determinants of EA Assimilation Stages The study controlled for past assimilation phases and stages in order to amplify the effect of the determinants at each assimilation stage. A detailed description of multinomial logistic regression is provided in the next section of the paper. Data Sample and Scales The study re-used the longitudinal data sample and scales contained in the Makiya & Lyytinen (2011) study. All five constructs used for the second paper were reused for this study. All five constructs were measured at phases t1, t2 and t3, while coercive pressure was measured only at phases t2 and t3. Operationalization The following section outlines how each construct was operationalized. Details on construct operationalization are covered in the second paper. Dependent Variable: EA Assimilation EA assimilation was measured as the extent to which an organization has progressed through its assimilation process from initial awareness to full institutionalization (Fichman, 2001).
177
TABLE 22: Construct Operationalization Constructs
Access to resources
Operationalization The extent to which the organization is capable of securing adequate resources including human capital and money to successfully drive assimilation. The number of mission areas or concentrations an organization had.
Years Survey years 1999, 2002, 2006
Sources 1.National Archives and Records Administration: With permission from Congressional chair
2001-2008
Agency websites: data such as number of missions per agency; number of functions (using organization charts; number of physical facilities, such as buildings, office locations, military bases, laboratories and field offices, satellite operation centers etc 1.Code of Federal regulations: data on number of legal and regulatory requirements by agency 2. gpoaccess.gov Agency websites: on parent agencies and departments for each of the bureaus and subagencies 3.National Archives and Records Administration: With permission from Congressional chair GAO:1999, 2003 and 2006 EA survey results
Organization Complexity
The expanse of operational coverage and the degree of authority and resource independence.
2001-2008
The variance of and attitude towards perceptions of changes in the environment often denoted as the “selfish pettiness or narrowness” with regard to external interests, opinions or views The perception of the value of an innovation by the management team
Survey years 1999, 2002, 2006
Organization Scope
Parochialism and cultural resistance
Top management value recognition
Coercive Pressure
Size
The extent of formal pressures exerted on government agencies. The number of employees in the organization
Survey years 1999, 2002, 2006
2002-2008
4.National Archives and Records Administration: With permission from Congressional chair GAO:1999, 2003 and 2006 EA survey results OMB/NARA; E-gov.gov
2001-2008
Agency websites/portal
178
Statistical Analysis OLS estimation was used to derive the regression variate for each observation in the data set. As such, the predicted value represents the total of all effects of the regression model and allows the residuals to be used as extensively as a diagnostic measure for the overall regression model (Hair, Black, Babin, & Anderson, 2010). The study met two key assumptions for the OLS estimates to be unbiased and have minimum variance. The study assumed the absence of multicollinearity and heteroskedasticity by checking to see, if the adjusted R2 was high and if the t-stats were low; in addition to checking, if the correlation coefficients were high. Using VIF, I was satisfied that there was no threat of either multi-collinearity or heteroskedasticity. Logistic regression is best used in research problems that involve a single categorical dependent variable and several metric or non-metric independent variables (Hair et al., 2010). The multinomial logit model is more effective when the dependent variable is polytomous (Aldrich & Nelson, 1984). In a multinomial logistic regression model, the estimates for the parameter can be identified compared to a baseline category (Long, 1997). This study uses logistic regression analysis to identify the relationships between the study’s independent variables (organization complexity, organization scope, access to resources, management value recognition and parochialism and cultural resistance) and the dependent variables (EA assimilation stage). The choice of multinomial logistic regression was based on its ability to estimate the individual effects of continuous or categorical independent variables on categorical dependent variables (Wright, 1995).
179
Findings Next I detail the results of the hypotheses testing. The table 23 below provides the model fit statistics for testing of hypotheses 1a-1e. TABLE 23: Model Fit Statistics Phase
Years
t1
99-01
t2 (no coercive pressure) t2 (with coercive pressure) t3 (no coercive pressure) t3 (with coercive pressure)
02-04
R Square
Adjusted R Square
Std. Error of the Estimate
R Square Change
F Change
0.627
0.613
0.748
0.627
3.436
0.001
0.753
0.746
0.958
0.753
1.129
0.151
0.756
0.747
0.944
0.756
1.504
0.062
0.638
0.634
1.144
0.638
1.239
0.005
0.593
0.588
1.149
0.593
1.475
0.008
P-Val
02-04
05-07
05-07
At time period t1, the model was significant with p-value=0.001 at α=0.05 with R2= 62.7%. The significant factors were organization complexity (β = 0.214; p= 0.002) and parochialism & cultural resistance (β = 0.113; p= 0.084). The R2 ranges from 0 to 1 and can be interpreted as the percentage of the variance in the dependent variable that is explained by the independent variables. The adjusted R2 statistic is the same as the R2 except that it takes into account the number of independent variables. The F statistic is the ratio of the explained to the unexplained portions of the total sum of squares, adjusted for the number of independent variables and the degrees of freedom. At t2, the model was insignificant without coercive pressure (p-value=0.154) with R2= 75% and all second order items had insignificant relationship with EA assimilation. Upon inclusion of coercive pressure, the model regained significance with p=0.062 and 180
R2=75.6%; along with coercive pressure (β=0.214, p-value=0.037) and organization complexity (β=0.184, p-value=0.08). It was therefore apparent that coercive pressure had a positive impact on the relationship between the determinants and EA assimilation. There were 6 estimated variables namely Access to resources, organization complexity, organization scope value cognition and parochialism. The n=118. At t3, the model was significant with or without coercive pressure moderation. The resulting model with or without coercive pressure had (p-value=0.005) and (p=0.008) and R2=64% and 59% respectively. Results for hypotheses H1a – H1e This next section reports on the ordinary least squares statistical analysis. TABLE 24: Hypotheses H1a-e Results Temporal Phase
t1 Beta
t2
Hypothesis
Supported Y/N
P-val
H1a
Supported
0.173
0.002
H1b
Supported
0.073
H1c
Not supported
H1d
Supported
H1e
Not supported
-0.072
Beta
t3 P-val
Beta
P-val
0.153
0.148
0.232
0.064
0.084
0.107
0.216
0.201
0.001
0.002
0.979
-0.089
0.516
0.162
0.347
0.173
0.002
0.184
0.08
0.232
0.064
0.216
-0.052
0.687
0.155
0.306
Hypothesis H1a was supported. Access to Resources has a positive significant relationship with EA assimilation at t1, t2 and t3. Hypothesis H1b was supported. Parochialism and cultural resistance had negative determinate power at t1 and at t3. Hypothesis H1c was not supported. Management Value recognition was insignificant at t1 at t2 and t3. Hypothesis H1d was supported. Organization complexity had positive
181
determinate power at t1, t2 and at t3. Hypothesis H1e was not supported. Organization scope had insignificant determinate power at t1 at t2 and at t3. TABLE 25: Results for Hypotheses H2a-c Temporal Phase Coercive Pressure will
t2 Supported Y/N
Organization Scope Access to Resources H2a: Reduce inverse Parochialism/Culture effect of resistance H2c: increase positive effect of Management Value Recognition H2b: Reduce inverse effect of Organization complexity
Beta
t3 P-value
Beta
P-value
Not supported
0.0124
0.8654
0.0243
0.7986
Not supported
0.1815
0.1914
-0.1943
0.3047
Not supported
-0.0478
0.5866
0.0969
0.2976
Not supported
0.08
0.37
-1031
0.402
Supported
0.1354
0.02
0.1699
0.0207
Hypothesis H2a was not supported. coercive pressure did not positively moderate the relationships between access to resources and EA assimilation at t2 and t3 coercive pressure did not positively moderate the relationship between parochialism & cultural resistance and EA assimilation at t2 and t3. Coercive pressure did not positively moderate the relationship between organization scope and EA assimilation at t2 and t3. Hypothesis H2b was supported. Coercive pressure positively moderated the relationship between organization complexity and EA assimilation levels at t2 and at t3. Hypothesis H2c was not supported. Coercive pressure had an insignificant impact on the relationship between management value recognition and EA assimilation at t2 and at t3. The following section reports on the regressions of the multinomial logistic regressions. Table 26 below reports analyses that tested hypotheses H3a-H3e.
182
TABLE 26: Study 3 Multinomial logistic regression results for hypotheses H3a-H3c Descriptive statistics for Multinomial logistic regressions Time period t1 EA Assimilation B Std. Error P-Val Exp(B) a 2 Intercept -38.426 6.846 0 AcctoRes -4.184 3.395 0.218 0.015 MgtPer -4.13 3.649 0.258 0.016 OrgComp 3.494 4.176 0.403 32.921 OrgScop 42.397 1.005 0 25.918 QPC 5.407 4.574 0.237 222.959 QVC 0b . . . QCP -38.802 6.873 0 3 Intercept -4.262 3.398 0.21 0.014 AcctoRes -4.266 3.653 0.243 0.014 MgtPer 3.759 4.177 0.368 42.922 OrgComp 41.887 1.014 0 15.518 OrgScop 5.623 4.576 0.219 276.772 QPC 0b . . . QVC -42.489 7.116 0 QCP -3.922 3.425 0.252 0.02 4 Intercept -4.541 3.694 0.219 0.011 AcctoRes 4.245 4.187 0.311 69.769 MgtPer 42.143 1.074 0 20.118 OrgComp 5.87 4.593 0.201 354.229 OrgScop 0b . . . QPC -49.428 8.099 0 QVC -3.469 3.542 0.327 0.031 QCP -2.672 3.797 0.482 0.069 5 Intercept 4.753 4.222 0.26 115.874 AcctoRes 41.936 0 . 16.318 MgtPer 5.563 4.636 0.23 260.677 OrgComp 0b . . . OrgScop QPC QVC QCP a. The reference category is: 1. b. This parameter is set to zero because it is redundant.
Time period t2 B
Std. Error
P-Val
B -38.426 -4.184 -4.13 3.494 42.397 5.407 0b -38.802 -4.262 -4.266 3.759 41.887 5.623 0b -42.489 -3.922 -4.541 4.245 42.143 5.87 0b -49.428 -3.469 -2.672 4.753 41.936 5.563 0b 1.356 0b 0.272
Std. Error 6.846 3.395 3.649 4.176 1.005 4.574 . 6.873 3.398 3.653 4.177 1.014 4.576 . 7.116 3.425 3.694 4.187 1.074 4.593 . 8.099 3.542 3.797 4.222 0 4.636 . 0.961 . 0.762
0.057 0.054 0.53 0.1 0.064 0.795 . 0.017 0.093 0.594 0.476 0.034 0.595 0.626 . 0.656 0.006 0.127 0.845 0.29 0.528 0.058 . 0.789 0.061 0.479 0.34 0.465 0.253 0.158 . 0.721
Time period t3 Exp(B) 2.614 0.701 1.775 2.021 1.112 . 0.433 0.798 1.385 1.882 1.222 1.167 . 0.877 2.294 1.122 1.543 1.39 2.139 . 0.905 2.081 0.214 1.829 2.803 3.879 . 1.313
B -3.787 0.961 -0.355 0.574 0.704 0.106 0b -0.836 -2.811 -0.226 0.326 0.632 0.2 0.155 0b -0.132 -6.937 0.83 0.115 0.433 0.329 0.76 0b -0.1 -8.76 0.733 -1.544 0.604 1.031 1.356 0b 0.272
Std. Error
P-Val
1.987 0.498 0.565 0.349 0.38 0.409 .
. 0.35 1.672 0.424 0.456 0.298 0.377 0.317
.
.
.
0.798 1.385 1.882 1.222 1.167 .
.
0.877 2.294 1.122 1.543 1.39 2.139 .
0.789 0.061 0.479 0.34 0.465 0.253 0.158 .
0.762
0.433
0.656 0.006 0.127 0.845 0.29 0.528 0.058
0.373 4.68 1.035 1.618 0.827 0.901 0.961 .
2.614 0.701 1.775 2.021 1.112
0.017 0.093 0.594 0.476 0.034 0.595 0.626
0.296 2.51 0.544 0.59 0.41 0.523 0.401 .
Exp(B)
0.057 0.054 0.53 0.1 0.064 0.795
0.905 2.081 0.214 1.829 2.803 3.879 .
0.721
1.313
183
Results for Hypotheses H3a – H3e Hypothesis H3a was not supported. At t2, with every unit increase in EA assimilation level from level 2 to 3, the effect of access to resources was insignificant (OR=0.853 95%CI=0.417- 1.747). At t3 with every unit increase in EA assimilation level from level 2 to 3, the effect of access to resources is equally insignificant (OR=0.798 95%CI=0.348- 1.83) Hypothesis H3b was supported. At t2, with every unit increase in EA assimilation level from level 2 to 3, the effect of Parochialism and cultural resistance increases 92.3% (OR=1.923 95%CI=0.966- 3.826) but then decreases to 27.5% (OR=1.275 95%CI=0.643- 2.529) with every unit increase in assimilation from level 3 to level 4. At t3 with every unit increase in assimilation level from level 2 to 3, the effect of parochialism and cultural resistance increases 11.2% (OR=1.112 95%CI=0.498- 2.482) and further increases to 16.7% (OR=1.167 95%CI=0.627 – 2.171) with every unit increase in assimilation level from level 3 to level 4. At t2, with every unit increase in EA assimilation level from level 2 to 3, the effect of organization complexity increases by 42% (OR=1.42 at 95%CI=0.755- 2.672) but then decreases to 26.3% (OR=1.263 at 95%CI=0.323- 4.972) with every unit increase in assimilation from level 3 to level 4. At t3, with every unit increase in assimilation level from level 2 to 3, the effect of organization complexity increases by 88.2% (OR=1.882 at 95%CI=1.05 - 3.373) but then decreases to 54.3% (OR=1.543 at 95%CI=0.691- 3.444) with every unit increase in assimilation level from level 3 to level 4. Hypothesis H3c was supported. At t2, with every unit increase in EA assimilation level from level 4 to 5, the effect of organization scope increases 30.9% 184
(OR=1.309 95%CI=0.124- 13.763). At t3 with every unit increase in EA assimilation level from level 4 to 5, the effect of organization scope increases 80.3% at level 5 (OR=2.803 95%CI=0.479- 16.396). PostHoc Analysis The results suggest that Organization complexity is more likely to be a barrier to EA assimilation level advancement from level 2 to level 3 at t3 than at t2 (OR=1.882 95%CI=1.05 - 3.373). Parochialism and cultural resistance are more likely to be a barrier to EA assimilation level advancement from level 2 to level 3 at t2 than at t3 (OR=1.923 95%CI=0.966- 3.826). The results also suggest that organization scope is more likely to be a barrier to EA assimilation level advancement from level 4 to level 5 at t3 than at t2 (OR=2.803 95%CI=0.479- 16.396). Discussion and Conclusion It is important for the IS and practitioner communities to understand what determinants influence EA assimilation. This study drew upon theoretical perspectives on the process and contexts of innovation diffusion, and developed an integrative model to examine the influence of five determinants and different environments of EA assimilation. The study’s empirical results have revealed that several determinants have differential effects at different levels and within different environments. Whereas the Makiya and Lyytinen (2011) study identified significant antecedents of EA assimilation, this study’s empirical results identified and revealed their differential effects across different levels and in different environments. This paper was an extension of previous work by emphasizing the effects of environmental circumstances on the process of assimilation. This approach has not been 185
used in the previous papers, to the best of the author’s knowledge. The evidence shows that both environmental changes and the organizational learning process have great influence on the assimilation levels and their antecedents. The regulatory pressure at t2 provides evidence administrative innovation assimilation is unsustainable without institutional intervention. The study provides further support for the Lyytinen & Damsgaard (2011) typology that assimilation analyses should be conducted separately for both adopter populations and adopter units. The results also show that adopter population behavior influences adopter units. Most importantly, the typology’s argument that the organizing vision does play a key role in defining the nature of the assimilation was supported. EA was introduced via U.S. Government mandate in 1996 and the sluggish assimilation progress necessitated institutional intervention. The evidence of the positive environmental impact at phase t2 suggests that coercive pressure may be a viable strategy for closing assimilation gaps for administrative innovations. However, the results also provide evidence that application of coercive pressure for EA assimilation has temporal limitations. The evidence shows that organization complexity and parochialism and cultural resistance regained determinacy at t3, despite the sustained presence of the sanctions. The results also show that as the organization continues to learn, the strength of the determinants fluctuates. This suggests that knowledge acquisition can be a key ingredient in taming the effects of negative influencers. However, it also shows that as organizations learn, either from their environments, their peer organizations or from industry artifacts and manuals, they become more proficient and efficient in assimilating the innovation.
186
The results showed that a majority of the determinants were significant barriers to assimilation advancement from level 2 to level 3. Organization complexity in particular is a significant barrier to assimilation progression from level 2 to 3 thus supporting Ross et al. (2006: 73) who argued that as the organization makes the transition from silo operations to a more integrated and standardized environment, complexity will be amplified. Organization scope is a barrier to EA assimilation progression only when advancing from assimilation level 2 to 3, a surprise result nonetheless based on Ross et al (ibid) claim that the painful transition from level 2 manifests itself as the organization faces challenges associated with shift in focus on organization-wide rather than business unit or functional activities and assets. It is instructive that the main activity at this level is consolidation, which threatens autonomy. It was the study’s expectation that the limited geographical scope and impact on operations associated with advancement from level 2 to 3, organization scope would not have predictive effect. The surprise result however was the plummeting of the influence of parochialism and cultural resistance from 92% when EA advances from level 2 to 3 to a low of 26% when EA progresses from level 3 to level 4. One explanation may be derived from Ross et al (2006) who wrote that some organizations at this level may also establish seamless linkages between business process modules and greater autonomy/discretion at business unit level for building or buying modules. This drive for autonomy does have the potential to escalate parochialism and cultural resistance. The other surprise result was the insignificance of Management value recognition at all the assimilation levels. The study had predicted influence of management value recognition at levels 2 through 4. According to the literature, at this level, business 187
managers begin appreciating the value of EA as it standardizes IT, reducing risk, and the cost of shared services, reliability, security and improvements in development time (Ross et al., 2006: 74). They argue that as the benefits become apparent, business unit managers begin recognizing and appreciating the value of EA.
Do the determinants of assimilation levels change as EA progresses from one assimilation phase to the next? At t1, Parochialism and cultural resistance, organization complexity and organization scope had significant influence. At t2 however, only organization complexity and organization scope had influence while at t3 parochialism and cultural resistance, organization complexity and organization scope were the significant predictors. The study’s findings on the predictive powers of the determinants showed that organization complexity and organization scope retain their determinacy at all three phases, while parochialism and cultural resistance was only significant at t1 and t3. However, the results showing the resurgence of parochialism and cultural resistance at t3 merit further discussion. The study results indicate that upon introduction of EA, parochialism and cultural resistance was quite significant. The institutional intervention via coercive pressure from the executive branch had a taming effect at t2, negating its significance. However, the results show a powerful resurgence at t3 which can be easily attributable to the study’s contextual factors. At t3 for instance, there was heavy resistance within the federal community to private sector job outsourcing. Federal trade unions, teaming with disgruntled federal executives successfully fought off competitive job sourcing. Hiring managers tilted job descriptions to favor internal applicants, 188
ensuring high internal win rates and in the process reenergizing parochialism and cultural resistance. However, the literature (e.g. Ross et al., 2006; Kappelman, 2010) argue that at the later levels of EA, the resulting organizational changes may create a new wave of resistance to change. As predicted, the results for the assimilation levels indicate that EA assimilation levels are directly related to organization complexity. The result not only shows organization complexity as a significant determinant throughout the EA assimilation lifecycle, it also shows that as EA assimilation increases, so too does the need for implementation of strategies and measures to tackle complexity. Limitations The context of the study limited its generalizability, despite the sample size. In addition to sample size, the dataset covered the period from 1999-2007, 3 years after introduction of EA within the federal government. This denied the study of the opportunity to study the factors that may have been influential at the onset of assimilation. Future Research The GAO has recently released a new 7-level assimilation framework. Future research may want to replicate this study using the new framework to establish the factors that influence assimilation progression from the current highest level -5 to level 7. Secondly, the GAO for a long time used the EA program maturity framework for assessing adopter level program portfolios. The 2004 introduction of the OMB EAAF effectively transferred portfolio level assessment to the OMB, albeit after the 2006 GAO
189
assessment. Future research may want to replicate this study utilizing the OMB data and conduct comparison between the GAO assimilation framework results and the OMB data. Future researchers may also want to replicate the longitudinal study with private sector data, since some of the study’s results were contextual.
190
CHAPTER V: DISSERTATION CONTRIBUTIONS Study 1 Contribution No published studies on the antecedents of Enterprise Architecture program assimilation have ever been conducted. EA has also never been researched as a reform and administrative innovation. Numerous studies on reform diffusion have focused on administrative reform – albeit within the political science arena (e.g. Aucoin, 1990; Barzelay, 1993; Bowornwathana, 1994a, b; Masujima, 1993; Caiden, 1991). The vast majority of the literature on IS innovation assimilation has selectively focused on technological innovation assimilation. EA program maturation within both the public and private sector organizations can be a highly complex undertaking. Likewise, innovation assimilation continues to draw multiple streams of studies, given the high volumes and rapid rate of technological innovation couple with the high cost of failure. The first study’s unit of analysis, the Federal EA programs, in addition to being an administrative innovation, were direct beneficiaries of government intervention. This study has important implications for government leaders, officials and federal program development practitioners at a time when the U.S. has promulgated manadatory reform such as the healthcare reform bill. It also has implications for the practice of EA within the public and private sector organizations, due to its transformative nature. A number of countries, including the Netherlands, Denmark, Singapore, Canada, Australia, Malaysia and India are in the process of implementing Government Enterprise Architecture. Akin to its introduction within the federal sector, EA is primarily adopted as an administrative innovation within the private sector.
191
The first study makes important extension to the literature on neoinstitutional theory. Whereas institutional theory has traditionally concerned itself with social structures and social behavior in diffusion, this study adds a new dimension by examining the role of individual behavior as a catalyst in innovation assimilation. The role of individual behavior in institutional innovation and diffusion has curiously not attracted the attention of scholars. Individual characteristics feature prominently in change management and leadership literature, but seldom from a behavioral perspective. The study also makes a contribution to the burgeoning field of Enterprise Architecture by examining constraints and inhibitors to Program assimilation within the setting of the U.S. federal government. Past emprirical and longitudinal studies in reform and Innovation assimilation have been conducted separately. This study synthesizes the twin phenomena. Numerous studies on reform diffusion and assimilation have focused on administrative reform (Aucoin, 1990; Barzelay, 1993; Bowornwathana, 1994a, b; Masujima, 1993; Caiden, 1991). Innovation assimilation studies have similarly focused on the classic two-stage diffusion model (DiMaggio & Powell, 1991; McGuire, Granoveter, & Schwartz, 1993; Zucker & Tolbert, 1983; Kennedy & Fiss, 2009) and the seven stage assimilation model (Fichman & Kemerer, 1999). Study 2 Contribution The second study has unique contributions for EA practitioners and managers. Many of this study’s findings have implications for theory, methods and practice. The study sought to test factors attributable to anecdotal evidence from case studies such as Ross et al. (2006), Kappelman (2010), Brynjolfsson and Kemerer (1997). The study’s main objective was also to explore the influence of coercive pressure (DiMaggio & 192
Powell, 1983) on EA assimilation. However, the study used cross-sectional data in order to: 1) test the new research model 2) to examine the causal factors (and confirm the direction of causality) 3) examine the interactions between the antecedents and EA assimilation levels. Whereas prior studies have extensively covered organizational level factors and their impact on assimilation of innovation, I could not locate a single study that has focused on factors that influence EA assimilation in particular. The study’s findings that sustained coercive pressure plays a significant role in assimilating EA in large and complex organizations strongly confirms earlier expectations and studies (e.g. Teo et al., 2003). The research on the role of institutional pressure, more so coercive pressure in IS assimilation remains sporadic and scarce, despite its high importance and reputation as problematic (Swanson, 2004). As organizations continue to grow globally and inorganically, so too are challenges in assimilation of IT and EA in particular. Several IS studies unanimously cite complexity, high industry regulations (targeted at individual business units) business unit autonomy and jurisdiction as significant challenges to assimilation (Fichman, 1999; Swanson, 2004; Zhu et al., 2006; Liang et al., 2007; Wang, 2008). Study 3 Contribution This study makes three specific contributions to the literature on EA assimilation. First, the study expanded our understanding of the EA assimilation process at three temporal phases (adoption, institutionalization, routinization). Second, the study theorized and tested differential effects of the determinants across the assimilation phases. There are no prior studies on the antecedents of assimilation stages. To the best of 193
my knowledge, this paper is the first longitudinal study of the differential effects in innovation assimilation. The study theorized and tested the differential effects as an important theoretical issue (Ross et al., 2006; Schekkerman, 2005). These results support the theoretical notion of “differently directioned effects,” that is, the same factors may play different roles at different assimilation stages. Although prior research has recognized such differential effects as an important theoretical issue (Tornatzky & Klein, 1982), the literature lacks empirical examination (Fichman, 2000). This study sought to validate this argument. Third, the study is the first to conduct a longitudinal study on the determinants of EA assimilation stage progression. The study’s findings that both environmental changes and the organizational learning process have great influence on the assimilation process make great contribution to IS theory and literature. The study provides further support for the Lyytinen & Damsgaard (2011) typology that assimilation analyses should be conducted separately for both adopter populations and adopter units. The results showing that adopter population behavior influences adopter units’ makes great theoretical contribution to institutional theory. Most importantly, the typology’s argument that the organizing vision does play a key role in defining the nature of the assimilation was supported. The study’s key managerial contribution was the evidence that application of coercive pressure for EA assimilation has temporal limitations. The study’s findings provided strong support for the study’s prediction that EA assimilation levels are directly related to organization complexity. In major contribution to IS research, the result not only shows organization complexity as a significant determinant throughout the EA assimilation lifecycle, it also shows that as EA 194
assimilation increases, so too does the need for implementation of strategies and measures to tackle complexity. Implications from the Three Studies for Managers and Practitioners This study has important implications for government leaders, officials and federal program development practitioners at a time when the U.S. has promulgated manadatory reform such as the healthcare reform bill. It also has implications for the practice of EA within the public and private sector organizations, due to its transformative nature. A number of countries, including the Netherlands, Denmark, Singapore, Canada, Australia, Malaysia and India are in the process of implementing Government Enterprise Architecture. Akin to its introduction within the federal sector, EA is primarily adopted as an administrative innovation within the private sector. Within the private sector, it may be introduced as a management reform initiative, usually at the instigation of external propagators such as consulting firms or IT auditors. Reform diffusion within both the public and private sector can be a highly complex undertaking. Likewise, innovation diffusion continues to draw vast amounts of research attention given the rapid rate of technological innovation. Our unit of analysis, the Federal EA programs, in addition to being an innovation, was a direct creation of a Congressional reform mandate. The study has important implications for government leaders, officials and federal program development practitioners at a time when the U.S. is on the verge of passage of the healthcare reform bill. It also has implications for the practice of enterprise architecture within the public and private sectors, due to its transformative nature. A number of countries, among them the Netherlands, Denmark, Malaysia and India are in the process of implementing Government Enterprise 195
Architecture. Akin to its introduction within the federal sector, EA is primarily adopted as an innovation within the private sector. The greatest implication for managers is that 1) in order to successfully adopt administrative and complex innovations, it is imperative that a program be set up to guide the development of capabilities to implement it 2) that a suitable framework aimed at assessing the capability and capacity development be devised 3) that innovative leadership is key to driving diffusion in the face of resistance from established interests from within and without the organization. It also has implications for senior management that a separate framework is required for measuring program portfolio level Key Result Areas (KRAs), which may or may not be linked in any way to the singular program level KRAs. Contributions to EA Assimilation Theory This study makes three specific contributions to the literature on EA assimilation. The first and most important contribution is the delineation between 1) an EA program and an EA program portfolio 2) the measures and KRAs of EA programs and adopter units. The literature has consistently referred to measures of EA programs and program portfolios as “maturity stages”, creating confusion over the unit of analysis and clouded our understanding of the differences between the two. This delineation helps clarify that 1) an EA is the primary vehicle for adoption and diffusion of EA within the organization; 2) that an EA program assimilation ranking is different from the adopter unit level (program portfolio) assimilation ranking; and 3) EA programs are charged with developing competencies and capabilities whereas EA program portfolios are charged with application of EA principles within the organization. The confusion over the use of 196
EA program level KRAs to measure adopter level assimilation is best captured in the qualitative interviews. For instance program level managers talked positively of the GAO assimilation model while top level Agency executives saw absolutely no value for it: “..We run EA based on a scorecard approach. Do you have policies in place? Do you have a team? We've run it stovepipe organization by organization.” (Agency executive) The three studies combined demonstrate that EA is a multi-level complex phenomenon that must be addressed separately at each of the constituent levels. The first study shows that 1) the singular EA programs are the foundational basis of adoption and assimilation within an adopter unit and 2) the factors that influence programs are multidimensional. The studies show that the EA program assimilation tool is primarily suited for assessing the foundational capacity building at the micro (singular program level). The dissertation also shows that EA program assimilation is not synonymous with adopter unit level assimilation levels. The study also shows that the determinants of adopter unit level differ from those at the EA program level. It is also shown that the adopter level assessment tool is aimed at assessing the adopter unit (program portfolio) application of EA principles within the organization. The application of EA principles is thus demonstrated by widespread use and institutionalization of EA. The results that should be of great interest to practitioners and managers are the ones which suggest that highly complex organizations may require higher degree of investment in EA, in addition to investment in simplification, in order to realize successful EA assimilation. There are several other practical implications for IS practitioners. One is to ensure that they focus investments in tackling complexity for successful EA assimilation. Second is that EA assimilation planning should take into 197
account jurisdictional factors (such as uneven landscape) regional infrastructure, politics, regulations as well as the impact of parochialism and cultural resistance. Thirdly, and most importantly, the results show that securing top management EA value appreciation is critical for securing vital resources necessary for EA assimilation. The results also show that investment in skilled resources may be far more critical than singular focus on funding in driving successful EA assimilation. Future Research This dissertation’s groundbreaking examination of the factors that influence EA assimilation opens up a wide range of potential research topics. This thesis presents just the first step in the broad research agenda focusing on the determinants of EA assimilation. It also represents a major milestone in establishing the influence of environmental changes on the factors that influence EA assimilation. However, the research context is the U.S. federal government. Future research may want to replicate this study in different contexts such as state and local governments or the private sector industry clusters. Future research may also want to examine the role that assessment tools play in program development and adopter unit level assimilation efforts. For instance it is now apparent from the study that from 1996-2003, the GAO used a program level EA program assessment tool to equally assess adopter unit level EA assimilation. It may be interesting to establish the impact of concurrent assessment tools from the adoption stage to simultaneously track the progress of adopter level assimilation and EA program assimilation and compare and contrast.
198
APPENDIX A: Interview Protocol 1. Tell me a little bit about yourself? • Educational background • Professional background 2. Tell me about your agency 3. I’d like to talk about your Enterprise Architecture Program. Thinking back to 200X, when the Program was initiated, can you tell me what happened that FIRST YEAR? Please tell me in as much detail as possible what it was like to start-kick your EA Program. Probes: 4. Your EA Program was first audited in 200X. At that time, it received a maturity ranking of #X. Thinking back to that particular time, what can you remember about the audit and the agency’s resultant ranking? Probes:
5.
In 2006 your EA was audited again. That time you got a ranking of #X.
Tell me about that audit and the results of it – again, in as much detail as possible, emphasizing who was involved, what was done, etc. in the period between the 200X and the 200X audits that resulted in the latest ranking. Probes: 6. Finally, I’d like you to think about the agency’s EA Program as it exists today. Probes:
199
APPENDIX B: Comparative Analysis of Tempered Radical Characteristics 1) Build relationships with people inside and outside the company who share and appreciate marginalized aspects of your identity.
In conformity with this Meyerson & Scully (1995) criterion, my respondents formed external alliances within and without, especially with auditors from the GAO and OMB as well as officials from the inspectorgenerals’ office which helped counter marginalization
2) Develop the discipline to manage heated emotions to fuel your agenda.
Whereas Meyerson’s (1995) criterion applied to “rallying the troops” primarily focusing on the emotions of the marginalized or disenfranchised, in contrast, my respondents capitalized on heated emotions of agency executives (such as reactions to appropriation cuts or budgetary freezes by OMB) to further their Programs’ agendas
3) Separate public “front stage” performances from “backstage” acts to create an appearance of conformity and credibility while acting on differences to sustain your sense of self.
My respondents, as opposed to Meyerson and Scully’s (1995) criterion of playing the internal credibility and conformity card, leveraged external conformity (such as compliance with the mandate) thus elevating the priority of their Programs and sustaining momentum of their Programs
4) Design behind-the-scenes actions and initiate conversations that create connections with other people who have similar values, beliefs, and identities.
In conformity with the Meyerson and Scully (1995) criterion, my respondents created connections such as reaching out and appealing for volunteers from other agencies and departments who shared their beliefs, identities and values leading to success of their Programs
200
APPENDIX C: Comparative Analysis of Tempered Radical Tactics 1.
Resisting quietly and staying true to oneself
Our respondents, in conforming to Meyerson & Scully (1995) did the following” • Defied hierarchy (by passed numerous layers of top agency management to get things done) • Initiated silent change without visibly being labeled as rebels. • Initiated subtle forms of resistance including creation of own job titles and descriptions and fighting stigma
2.
Turning personal threats into opportunities
In contrast to Meyerson and Scully (1995), my respondents turned organization (rather than personal) threats into opportunities for advancement of their agenda and Programs
3.
Broadening the impact through negotiation
In conformity with Meyerson & Scully (1995), my respondents broadened impact by negotiating, rather than confronting bureau and business unit leadership
4.
Leveraging small wins
In conformity with Meyerson & Scully (1995), my respondents leverage small wins by holding elaborate “victory” celebrations to inspire Program volunteers and staff
5.
Organizing collective action
In conformity to Meyerson & Scully (1995), my respondents actively organized collective action
201
APPENDIX D: Construct Definitions Construct
Definition
Organization Scope
Organization Complexity
Organization complexity is defined as the amount of differentiation among distinct elements constituting the organization including the number of missions, functional units and the number of regulations, as well as their mutual dependencies.
Organization scope is defined as the geographical extent of a firm’s operations (Zhu et al, 2006). In this paper’s context it is narrowly defined by the challenges and obstacles associated with organizations operating in heterogeneous geographies.
Sub construct and definition Administrative multiplicity: Administrative multiplicity refers to the extent to which administrative and decisionmaking authority is dispersed within an organization
Literature reviewed (Swanson, 1994, Damanpour, 1991, Fichman, 2000)
Mission heterogeneity: Mission heterogeneity refers to the degree of dissimilarity of mission areas or mission focus within an organization
Fichman, 2000; Desormeaux, 1998; Lusthaus; Datta & Nugent, 1998).
Legal and Regulatory framework: This factor is defined as the scope and number environmental and regulatory restrictions influencing the organization, including government and consumer regulations (such as the Patriot Act, HIPPA, SOX), legal (Intellectual Property & contract enforcement) and industry policies (Cyber security and nuclear data)
(Eveland & Tornatzky, 1990; Meyer & Goes, 1988;
Jurisdiction: Jurisdiction is defined as the span of geographic and operational authority including administrative, technological, political and economic influence (Lusthaus, Anderson &Murphy, 1995; Datta & Nugent, 1998).
Fichman, 2000; Massetti & Zmud, 1996)
Autonomy: Autonomy is defined as the “degree to which an organization is free to make decisions with respect to its own operations” (Molnar and Rogers, 1976:62). In general, it refers to the degree of selfgovernance in decision making
(Robertson & Gatignon, 1986; Pfeffer & Salancik, 1978)
202
Lack of or scarcity of skilled staff in areas such as strategic planning, EA tools, cyber security etc
(Orlikowski, 1993; Orlikowski & Hofman, 1997; Brynjolfsson, & Kemerer., 1997; Robertson & Gatignon, 1986; Attewell, 1992, Tornatzky & Fleischer, 1990; Gefen & Straub, 1997)
Scarcity or abundance of funding for EA. EA budgetary allocations as a percentage of the programs’ budgetary requests. The percentage allocation was then used as a measure of funding availability to meet the EA funding obligations.
(Downs & Mohr, 1976, Fichman, 2000)
Lack of understanding of what EA was and its value perception from Management
(Leonard-Barton & Deschamps, 1988b; Damanpour, 1991; Fichman & Kemerer, 1997a; Grover, et al., 1997; Kimberley & Evanisko, 1981; Swanson, 1994)
The variance of and attitude towards perceptions of changes in the environment often denoted as the “selfish pettiness or narrowness” with regard to external interests, opinions or views
(Saga & Zmud 1994; Sambamurthy and Chin, 1994)
Parochialis m and Cultural Resistance
Management Value Recognition
Access to Resources
The extent to which the organization is capable of securing adequate resources including human capital and money to successfully drive assimilation.
Construct EA assimilation
Dependent Variable Definition The depth of EA innovation integration and its pervasiveness in the organization (Ash 1996)
Size
Control Variable The number of employees in the organization
Schekkerman (2005); Bernard (2005); Kappelmann (2010); CIO Council (1999); Ross et al. (2006), GAO EAMMF (2001)
(Bretschneider & Wittmer, 1993; Damanpour, 1991; Fichman & Kemerer, 1997a; Grover, et al., 1997; Kimberley & Evanisko, 1981; Lind et al., 1989; Meyer & Goes, 1988; Swanson, 1994; Blau, 1970; Baldridge & Burnham, 1973)
203
APPENDIX E: GAO Assimilation Model Core Elements Stage 1 Creating EA Awareness
Stage 2 Building the EA Mgmt Foundation
Stage 3 Developing EA products
Stage 4 Completing EA products
If an organization has plans to develop and use an architecture yet hasn’t satisfied the criteria in stage 2, it is considered stage 1
1. Adequate resources exist
10. Written and approved organization policy exists for EA development
16. Written and approved organization policy exists for EA maintenance 17. EA products and management processes undergo independent verification and validation.
2. Committee or group representing the enterprise is responsible for directing, overseeing, and approving EA
11. EA products are under configuration management.
18. EA products describe both the “as-is” and the “to-be” environments of the enterprise, as well as a sequencing plan for transitioning from the “as-is” to the “to-be.”
26. EA is integral component of IT investment management process.
3. Program office responsible for EA development and maintenance exists.
12. EA products describe or will describe both the “as-is” and the “to-be” environments of the enterprise, as well as a sequencing plan for transitioning from the “as-is” to the “to-be.”
19. Both the “as-is” and the “to-be” environments are described in terms of business, performance, information/data, application/service, and technology.
27. EA products are periodically updated
13. Both the “as-is” and the “to-be” environments are described or will be described in terms of business, performance, information/data, application/service, and technology.
20. Business, performance, information/data, application/service, and technology descriptions address security.
28. IT investments comply with EA
14. Business, performance, information/data, application/service, and technology descriptions address or will address security.
21. Organization CIO has approved current version of EA.
29. Organization head has approved current version of EA
15. Progress against EA plans is measured and reported.
22. Committee or group representing the enterprise or the investment review board has approved current version of EA.
30. Return on EA investment is measured and reported.
23. Quality of EA products is measured and reported.
31. Compliance with EA is measured and reported.
4. Chief architect exists. 5. EA being developed using a framework, methodology, and automated tool. 6. EA plans call for describing both the “as-is” and the “to-be” environments of the enterprise, as well as a sequencing plan for transitioning from the “as-is” to the “to-be.” 7. EA plans call for describing both the “as-is” and the “to-be” environments in terms of business, performance, information/data, application/service, and technology. 8. EA plans call for business, performance, information/data, service, and technology descriptions to address security. 9. EA plans call for developing metrics for measuring EA progress, quality, compliance, and return on investment.
ELEMENT CRITERIA STATUS = SATISFIED
= PARTIALLY SATISFIED
= NOT SATISFIED
Stage 5 Leveraging the EA to Lead Change 24. Written and approved organization policy exists for IT investment compliance with EA. 25. Process exists to formally manage EA change
204
APPENDIX F: OMB and GAO Evaluation Criteria Stage Stage 2
Element Committee or group representing the enterprise is responsible for directing, overseeing, and approving EA.
EAMMF Evaluation Criteria Does the agency have a chartered committee or group composed of representatives from across the enterprise, which is responsible for directing, overseeing, and approving the EA? Do meeting minutes exist that verify that the committee is meeting periodically and executing its designated responsibilities?
Stage 2
EA being developed using a framework, methodology, and automated tool.
Stage 3
EA products are under configuration management
Stage 3
Progress against EA plans is measured and reported
Does an enterprise architecture methodology exist that defines how the enterprise is to develop, maintain, and validate enterprise architecture products? Does the methodology prescribe the standards, steps, tools, techniques, and measures to be used to provide reasonable assurance that expected product quality it attained? Does a configuration management plan exist that describes how EA configuration items are identified, controlled, and audited? Do documents exist that verify that the configuration management process is being executed according to plan? Does an EA program plan against which progress can be measured? Is progress against this plan measured and reported? Does the agency take action to address deviations from the plan?
OMB EAAF Evaluation Criteria The EA governance ensures EA compliance throughout the agency. If noncompliance is identified the committee is responsible for developing a plan to resolve the issue. Alignment to the EA standards is a common practice throughout the agency. The compliance process is reviewed and updated when deficiencies or enhancements to the process are identified. The agency’s head or a designated operations executive has approved the EA governance plan in writing. Artifacts: EA Governance plan, EA Governance committee meetings, EA governance plan approval, EA communications plan At least 90% of the DME funding amount of the entire agency exhibit 53must be aligned to completed segment architectures. The agency must demonstrate at least one Federal Transition Framework initiative within one more that one segment reported to OMB, All designated segment codes are consistent with segment architecture definitions and scope agreed upon with OMB and with primary FEA Reference model mapping. Artifacts: Segment architecture, exhibit 53, Exhibit 300 Agency can demonstrate year over year results in closing IT quality gaps identified Artifacts: IT Infrastructure segment report, Exhibit 53, IT Infrastructure Line of Business (ITI-LOB)guidance and agency 5-year plans, Target Enterprise Architecture and Enterprise Transition plan
Agency must show year over year decrease in IT spending of at least 5% or the IT steady state spending should be 5% below the federal government average adjusted for the size of the overall agency budget. Every investment in the agency Exhibit 53 includes a prior year UPI code. All IT investments in the agency Exhibit 53 have been mapped to an accurate UPI code for the previous year using definitions found in OMB circular A-11 section 53. Artifacts: EA segment report, Program improvements assessment data, Exhibit 53 and Exhibit 300
205
APPENDIX G: OMB Assessment Ranking
source: Szyszka, 2009
Assimilation
Completion Average score equal to or greater than 4 in the “completion capability area” Average score equal to or greater than 3 in the “completion capability area” Average scoreless than 3 in the “completion capability area”
Use Average score equal to or greater than 4 in the “use capability area” Average score equal to or greater than 3 in the “use capability area” Average scoreless than 3 in the “use capability area”
Results Average score equal to or greater than 4 in the “results capability area” Average score equal to or greater than 3 in the “results capability area” Average scoreless than 3 in the “results capability area”
OMB EAAF Assimilation scoring matrix
206
APPENDIX H: Covariance Table 0Theta .. Outer residual covariance MH
Reg
AM
Jur
Auto
SKD
Fund
VCog
Par
MH
0.299
Reg
-0.322
0.686
AM
-0.258
-0.137
Jur
-0.007
-0.01
0.028
0.107
Auto
0.019
0.025
-0.073
-0.276
0.715
SKD
0.005
-0.005
-0.005
-0.012
0.03
0.474
Fund
-0.029
0.037
0.018
0.005
-0.013
-0.097
0.56
VCog
-0.001
-0.004
0.008
0.007
-0.019
-0.263
0.202
0.234
Par
0.005
-0.02
0.014
0.002
-0.005
-0.011
0.402
0.25
0.678
Size
0.002
-0.003
0
-0.017
0.044
0.004
0.006
-0.002
0.001
Size
0.731
0.148
207
APPENDIX I: Correlation Matrix
208
APPENDIX J: Hypothesis Testing Results Hypotheses 2a-2d Moderation H2a Partially H2b Not supported H2c Not supported H2d Not supported H3a Supported
b
se
Z
p
Exp(B)
Wald
PoC* AccToRes
0.1558
0.0812
1.9198
0.0549
1.1686
3.6855
VCog* AccToRes
0.1635
0.0933
1.7526
0.0797
1.1776
3.0718
-0.0027
0.0144
-0.1855
0.8528
0.9973
0.0344
-0.0384
0.1238
-0.3105
0.7562
0.9623
0.0964
0.5403
0.1852
2.9178
0.0035
1.7164
8.5137
H3b Supported
OrgComp* AccToRes OrgScop* AccToRes AccToRes* Coerpress PoC* Coerpress
-0.2401
0.0734
-3.2722
0.0011
0.7865
10.7074
H3c Supported
VCog* Coerpress
-0.2697
0.0843
-3.1986
0.0014
0.7636
10.2308
H3d Not supported H3e Supported
OrgComp* Coerpress OrgScp* Coerpress
-0.0076
0.0201
-0.3792
0.7046
0.9924
0.1438
-0.3042
0.1017
-2.9911
0.0028
0.7377
8.9469
95.0% C.I. for EXP(B) -0.2754 to 0.4121 -0.5234 to 0.2313 0.0218 to 0.0697 -0.0307 to 0.4065 0.0218 to 0.0697 01038 to 0.1317 0.0988 to 0.1584 -0.0004 to 0.0001 0.0208 to 0.0812
Tolerance
VIF
.207
4.822
.235
4.258
.961
1.040
.929
1.077
.980
1.020
.207
4.822
.235
4.258
.961
1.040
.929
1.077
209
APPENDIX K: PLS Models and Interactions Model 1: PLS model with the constructs and items but No moderation.
Model used for Hypotheses H1a-H1e
Interaction terms for Hypotheses H2a-d Using the MODPROBE script by the Prof Andrew F Hayes on http://afhayes.com/spss-sasand-mplus-macros-and-code.html. Please note that MODPROBE script automatically calculated the following in coming up with the interaction terms
210
Moderating variable Access to Resources (2nd order) 1. Funding (1st order)
x
Skilled Resources (1st order)
2.
x
Access to Resources Access to Resources (2nd order) Funding (1st order)
Independent variable Organization complexity (2nd order) 1. Mission heterogeneity 2. Administrative multiplicity 3. Legal and regulatory framework 1. Mission heterogeneity 2. Administrative multiplicity 3. Legal and regulatory framework Organization Scope 1. Jurisdiction 2. Autonomy
Skilled Resources (1st order
1. 2.
Jurisdiction Autonomy
Model 1b: Moderation of Access to Resources without the control variable (size). Model used for Hypotheses H2a-H2d Interactions:
H2a - Access to resources will moderate the relationship between Parochialism and cultural resistance and EA assimilation b
se
Z
p
Exp(B)
Wald
-0.7925
0.086
-9.2101
0
0.4527
84.8264
PoC
-0.0306
0.214
-0.143
0.8863
0.9699
0.0205
AccToRes
-0.483
0.1796
-2.6889
0.0072
0.6169
7.2304
interact
0.1558
0.0812
1.9198
0.0549
1.1686
3.6855
EA Assimilation
constant
5 4.5 4 3.5 3 2.5 2 1.5 1
Low AccesToRes High AccesToRes
Low PoC
High PoC
Assimilation levels are surprisingly high with increased access to resources even when parochialism goes up. In contrast, assimilation levels decrease as t the level parochialism and cultural resistance go up when access to resources is low (as expected).
211
H2b - Access to resources will moderate the relationship between value recognition and EA assimilation B
se
Z
p
Exp(B)
Wald
constant
-0.8076
0.0859
-9.3959
0
0.4459
88.2837
VCog
-0.2499
0.2403
-1.0397
0.2985
0.7789
1.081
AccToRes
-0.2176
0.161
-1.3511
0.1767
0.8045
1.8255
0.1635
0.0933
1.7526
0.0797
1.1776
3.0718
interact
5 EA Assimilation
4 3
Low AccesToRes High AccesToRes
2 1 Low Vcog
High Vcog
Assimilation levels go up with increased access to resources and the effect of value recognition is amplified. In contrast, with low access to resources increase in value recognition leads to slightly lower assimilation. H2c - Access to resources will moderate the relationship between Organization complexity and EA assimilation b
se
Z
p
Exp(B)
Wald
constant
-1.3208
0.1749
-7.553
0
0.2669
57.0473
OrgComp
0.0475
0.0148
3.2028
0.0014
1.0486
10.2582
AccToRes
-0.0437
0.1661
-0.2628
0.7927
0.9573
0.0691
interact
-0.0027
0.0144
-0.1855
0.8528
0.9973
0.0344
5 EA Assimilation
4 3
Low AccesToRes High AccesToRes
2 1 Low OrgComp
High OrgComp 212
The interaction plot shows that the effect of organizational complexity on EA assimilation is not affected by access to resources. H2d - Access to resources will moderate the relationship between Organization scope and EA assimilation b constant OrgScp AccToRes interact
se
Z
p
Exp(B)
Wald
-1.26
0.2831
-4.4509
0
0.2837
19.8101
0.2123
0.1347
1.5758
0.1151
1.2365
2.4832
0.004
0.2662
0.015
0.9881
1.004
0.0002
-0.0384
0.1238
-0.3105
0.7562
0.9623
0.0964
EA Assimilation
5 4 3
Low AccesToRes
2
High AccesToRes
1 Low OrgScope
High OrgScope
The interaction plot shows that the effect of organizational scope on EA assimilation is not affected by access to resources.
213
Moderation of Access to Resources – with control variable (size) H2a - Access to resources will moderate the relationship between Parochialism and cultural resistance and EA assimilation b
se
Z
p
Exp(B)
Wald
constant
-0.283
0.1171
-2.4171
0.0156
0.7535
5.8422
Qsize
0.0568
0.0355
1.5999
0.1096
1.0584
2.5597
-0.1453
0.1216
-1.1947
0.2322
0.8648
1.4273
-0.76
0.1621
-4.6875
0
0.4676
21.9722
0.2697
0.0765
3.5279
0.0004
1.3096
12.4464
Parochia AcctoRes interact
The inclusion of the control variable had an insignificant effect on the moderation effect of access to resources on the relationship between parochialism and cultural resistance and EA assimilation
H2b - Access to resources will moderate the relationship between value recognition and EA assimilation b constant
se
Z
p
Exp(B)
Wald
-0.3033
0.1171
-2.5897
0.0096
0.7384
6.7065
0.0555
0.0353
1.5724
0.1159
1.0571
2.4725
Valuecog
-0.1489
0.1366
-1.0901
0.2757
0.8616
1.1884
AcctoRes
-0.5769
0.1484
-3.8881
0.0001
0.5616
15.117
0.2183
0.0875
2.4937
0.0126
1.2439
6.2185
Qsize
interact
The control variable had no significant impact on the relationship between parochialism and cultural resistance and EA assimilation. H2c - Access to resources will moderate the relationship between Organization complexity and EA assimilation b constant
se
Z
p
Exp(B)
Wald
-0.7342
0.2013
-3.6477
0.0003
0.4799
13.3057
Qsize
0.0305
0.0371
0.8217
0.4112
1.0309
0.6752
OrgComp
0.0416
0.0178
2.3376
0.0194
1.0425
5.4644
AcctoRes
-0.1635
0.2135
-0.7659
0.4437
0.8491
0.5866
interact
-0.0129
0.0189
-0.6816
0.4955
0.9872
0.4646
214
The interaction is insignificant hence the hypothesis remains unsupported despite introduction of the control variable size which also has no significant impact H2d - Access to resources will moderate the relationship between Organization scope and EA assimilation b constant
se
Z
p
Exp(B)
Wald
-0.9476
0.3414
-2.7759
0.0055
0.3877
7.7059
Qsize
0.0789
0.0367
2.1511
0.0315
1.082
4.6272
OrgScop
0.2818
0.1547
1.8218
0.0685
1.3255
3.319
AcctoRes
-0.3239
0.329
-0.9844
0.3249
0.7233
0.969
0.0095
0.1524
0.0626
0.9501
1.0096
0.0039
interact
The interaction remains insignificant hence hypothesis remains unsupported despite introduction of the control variable size which similarly has no significant impact on the interaction. Moderation of coercive pressure WITHOUT control variable (size)
H3a - Coercive pressure will moderate the relationship between Access to resources and EA assimilation b
se
Z
p
Exp(B)
Wald
constant
-1.0625
0.2872
-3.6995
0.0002
0.3456
13.6867
AccToRes
-0.2257
0.1448
-1.5581
0.1192
0.798
2.4276
Coerpress
0.1079
0.382
0.2824
0.7776
1.1139
0.0798
interact
0.5403
0.1852
2.9178
0.0035
1.7164
8.5137
5 EA Assimilation
4 3
Low Coercive High Coercive
2 1 Low AcctoRes
High AcctoRes
The interaction plot shows that higher coercive pressure leads to higher assimilation levels with both lower and higher access to resources.
215
H3b - Coercive pressure will moderate the relationship between Parochialism and cultural resistance and EA assimilation b
se
Z
p
Exp(B)
Wald
constant
-1.5888
0.1088
-14.5985
0
0.2042
213.1151
Parochia
0.1276
0.0709
1.801
0.0717
1.1361
3.2436
Coerpress
1.4147
0.1273
11.1157
0
4.1153
123.5582
-0.2401
0.0734
-3.2722
0.0011
0.7865
10.7074
interact
Coercive pressure (negatively) moderates Parochialism and cultural resistance and EA assimilation 5 EA Assimilation
4 3
Low Coercive High Coercive
2 1 Low PoC
High PoC
The interaction plot shows that higher coercive pressure leads to higher assimilation levels with lower and higher levels of parochialism. H3c - Coercive pressure will moderate the relationship between mgmt value recognition and EA assimilation b
se
Z
p
Exp(B)
Wald
constant
-1.5592
0.1076
-14.4898
0
0.2103
209.9554
Valuecog
0.0979
0.0772
1.2676
0.2049
1.1028
1.6069
Coerpress
1.4171
0.1279
11.0793
0
4.125
122.7518
-0.2697
0.0843
-3.1986
0.0014
0.7636
10.2308
interact
5 EA Assimilation
4 3
Low Coercive High Coercive
2 1 Low Vcog
High Vcog
216
The interaction plot shows that higher coercive pressure leads to higher assimilation levels with lower and higher levels of value recognition. Surprisingly with higher value recognition the effect does not increase. H3d - Coercive pressure will moderate the relationship between Organization complexity and EA assimilation
b constant
se -2.0964
Z
0.1873
p
Exp(B)
Wald
-11.1912
0
0.1229
125.2427
OrgComp
0.0582
0.015
3.88
0.0001
1.0599
15.0545
Coerpress
1.2931
0.2175
5.9453
0
3.6441
35.3471
-0.0076
0.0201
-0.3792
0.7046
0.9924
0.1438
EA Assimilation
interact
5 4.5 4 3.5 3 2.5 2 1.5 1
Low Coercive High Coercive
Low OrgComp
High OrgComp
The interaction plot shows that higher coercive pressure leads to higher assimilation levels with lower and higher levels of organizational complexity. Surprisingly with higher complexity the effect does not increase whereas with low coercive pressure assimilation levels increase as complexity goes up. H3e - Coercive pressure will moderate the relationship between Organization scope and EA assimilation b
se
Z
p
Exp(B)
Wald
constant
-1.5557
0.1152
-13.5091
0
0.211
182.4954
OrgScp
0.0956
0.0999
0.9576
0.3382
1.1003
0.9171
Coerpress
1.4282
0.1338
10.6776
0
4.1713
114.0121
-0.3042
0.1017
-2.9911
0.0028
0.7377
8.9469
interact
Coercive pressure (negatively) moderates the relationship between organization scope and EA assimilation 217
EA Assimilation
5 4 3
Low Coercive
2
High Coercive
1 Low OrgScop
High OrgScop
The interaction plot shows that higher coercive pressure leads to higher assimilation levels with lower and higher levels of organizational scope. Surprisingly with higher scope the effect does increase with higher level of coercive pressure whereas with low coercive pressure the assimilation level remains largely the same. H3a - Coercive pressure will moderate the relationship between Access to resources and EA assimilation b constant
se
Z
p
Exp(B)
Wald
-0.9966
0.1446
-6.8916
0
0.3691
47.4937
Qsize
0.0548
0.0381
1.441
0.1496
1.0564
2.0766
AcctoRes
-0.206
0.1315
-1.5661
0.1173
0.8138
2.4527
Coerpress
0.9981
0.1256
7.9442
0
2.7132
63.1106
interact
-0.093
0.1153
-0.8064
0.42
0.9112
0.6503
Introduction of control variable size leads to loss of the interaction significance H3b - Coercive pressure will moderate the relationship between Parochialism and cultural resistance and EA assimilation b constant
se
Z
p
-1.1783
0.1382
-8.5273
Qsize
0.0522
0.0381
Parochia
0.0839
0.0819
Coerpress
1.1199 -0.2044
interact
Exp(B)
Wald
0
0.3078
72.714
1.3721
0.17
1.0536
1.8826
1.0245
0.3056
1.0875
1.0496
0.1261
8.8829
0
3.0645
78.9059
0.0749
-2.7292
0.0063
0.8151
7.4485
Introduction of the control variable size increases the significance of the interaction
218
H3c - Coercive pressure will moderate the relationship between mgmt value recognition and EA assimilation b constant
se
Z
p
Exp(B)
Wald
-1.1601
0.1381
-8.4017
0
0.3134
70.5879
Qsize
0.0559
0.0382
1.466
0.1426
1.0575
2.1492
Valuecog
0.0583
0.0935
0.6231
0.5332
1.06
0.3883
1.1287
0.1269
8.8933
0
3.0916
79.0899
-0.2387
0.0877
-2.7222
0.0065
0.7877
7.4103
Coerpress interact
Introduction of the control variable size increases the significance of the interaction H3d - Coercive pressure will moderate the relationship between Organization complexity and EA assimilation b constant
se
Z
p
Exp(B)
Wald
-1.5966
0.2184
-7.3119
0
0.2026
53.4632
Qsize
0.0242
0.0402
0.6012
0.5477
1.0245
0.3614
OrgComp
0.0502
0.0174
2.8784
0.004
1.0515
8.2855
Coerpress
1.0421
0.223
4.6733
0
2.8351
21.8395
-0.0082
0.0207
-0.3974
0.6911
0.9918
0.158
interact
Introduction of the control variable size leads to insignificance of the interaction H3e - Coercive pressure will moderate the relationship between Organization scope and EA assimilation b constant
Se
Z
p
Exp(B)
Wald
-0.7494
0.3434
-2.1823
0.0291
0.4727
4.7623
0.0633
0.0383
1.6537
0.0982
1.0654
2.7347
OrgScop
-0.2067
0.1635
-1.2642
0.2062
0.8133
1.5981
Coerpress
-0.2427
0.3868
-0.6276
0.5303
0.7845
0.3939
0.5832
0.1883
3.0964
0.002
1.7917
9.5876
Qsize
interact
Introduction of the control variable size makes the interaction significant
219
APPENDIX L: Multicollinearity, Skewness and Kurtosis Variables
Multicollinearity
Singularity
Skewness
Kurtosis
Tolerance
VIF
8.309 (peaked) 7.14(peaked)
.793
1.261
.918
1.089
Mission heterogeneity Administrative multiplicity Physical infrastructure fragmentation Size
0.793(None)
1.261(None)
2.603 (right)
0.918(None)
1.089(None)
2.374(right)
0.824(None)
1.214(None)
6.605(right)
49.75(peake d)
.824
1.214
0.655(None)
1.527(None)
3.303(right)
.655
1.526
Funding
0.289(None)
3.46(None)
1.474(right)
.289
3.462
Skilled resources
0.766(None)
1.305(None)
1.716(right)
.766
1.305
Parochialism and cultural resistance or culture Value recognition Autonomy Legal and Regulatory framework Jurisdiction EA assimilation
0.328(None)
3.052(None)
1.754(right)
9.645(peake d) 0.875(peake d) 1.863(peake d) 2.227(peake d)
.328
3.053
0.328(None) 0.866(None) 0.907(None)
3.046(None) 1.155(None) 1.102(None)
1.931(right) -0.179(left) 2.92(right)
3.37(peaked) -1.727(flat) 12.095(peak ed)
.328 .867 .907
3.047 1.182 1.102
0.846(None) 0.903(None)
1.182(None) 1.107(None)
0.217(right) 0.914(right)
-1.033(flat) -1.167(flat)
.846 .654
1.182 1.670
220
APPENDIX M: GAO Survey instrument
221
APPENDIX N: Sample GAO EAMMF Scoring Charts
222
223
APPENDIX O: OMB EAAF Service domains Service domain Customer services Process automation Services Business management services Digital asset services Business analytical Services Back office services Support services
Description Interaction between the business and the customer, and customer-driven activities (directly related to the end customer) Automation of process and management activities that support managing the business Management and execution of business functions and organizational activities that maintain continuity across the business Generation, management, and distribution of intellectual capital and electronic media across the business Extraction, aggregation, and presentation of information to facilitate decision analysis and business evaluation Management of transaction-based functions Cross-functional capabilities that are independent of service domains
224
APPENDIX P: List of Agency EA Programs Agencies & Departments 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50.
Department of Agriculture Agricultural Marketing Service Agricultural Research Service Animal and Plant Health Inspection Service Cooperative State Research, Education, and Extension Service Food and Nutrition Service Food Safety and Inspection Service Foreign Agricultural Service Forest Service Risk Management Agency Service Center Modernization Initiative Farm Service Agency Natural Resources Conservation Service Rural Utilities Service Department of Commerce Bureau of the Census Economic Development Administration International Trade Administration National Oceanic and Atmospheric Administration U.S. Patent and Trademark Office Department of Defense - GRID Department of Defense BEA Ballistic Missile Defense Organization Defense Advanced Research Projects Agency Defense Commissary Agency Defense Contract Audit Agency Defense Contract Management Agency Defense Information Systems Agency Defense Intelligence Agency Defense Logistics Agency Defense Security Cooperation Agency Defense Security Service Defense Threat Reduction Agency Department of the Air Force Department of the Army Department of the Navy National Imagery and Mapping Agency Defense Legal Services Agency National Security Agency The US Marine Corps Department of Education Department of Energy Department of Health and Human Services Administration for Children and Families Agency for Healthcare Research and Quality Centers for Disease Control and Prevention Centers for Medicare and Medicaid Services Food and Drug Administration Health Resources and Services Administration Indian Health Service
225
51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104.
Program Support Center Department of Homeland Security Department of Housing and Urban Development Department of the Interior US Geological Survey Office of Surface Mining Reclamation and Enforcement National Park Service Minerals Management Service Fish and Wildlife Service Bureau of Reclamation Bureau of Land Management Bureau of Indian Affairs Department of Justice Bureau of Alcohol, Tobacco, Firearms and Explosives Drug Enforcement Administration Federal Bureau of Investigation Federal Bureau of Prisons U.S. Marshals Service Immigration and Naturalization Services Department of Labor Department of State Department of Transportation Federal Aviation Administration Federal Highway Administration Federal Motor Carrier Safety Administration Federal Railroad Administration Federal Transit Administration National Highway Traffic Safety Administration The US Coast Guard Department of the Treasury Bureau of Engraving and Printing Bureau of the Public Debt Comptroller of the Currency Financial Management Service Internal Revenue Service Office of Thrift Supervision U.S. Mint US Secret Service Federal Law Enforcement Training center U. S. Customs Service Department of Veterans Affairs Veterans Benefits Administration Veterans Health Administration Independent agencies: Agency for International Development Central Intelligence Agency Corporation for National and Community Services Environmental Protection Agency Equal Employment Opportunity Commission Executive Office of the President Export-Import Bank Federal Emergency Management Agency Federal Deposit Insurance Corporation Federal Energy Regulatory Commission
226
105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123.
Federal Reserve System Federal Retirement Thrift Investment Board General Services Administration Legal Services Corporation National Aeronautics and Space Administration National Credit Union Administration National Labor Relations Board Nuclear Regulatory Commission Office of Personnel Management Peace Corps Railroad Retirement Board Securities and Exchange Commission Small Business Administration Smithsonian Institution Social Security Administration U.S. Postal Service National Science Foundation National Telecommunications and Information Administration Bureau of Industry and Security
227
APPENDIX Q: OLS Regression Results T1 (1999-2001) Model Summary Change Statistics Adjusted Std. Error R R of the R Square Model R Square Square Estimate Change 1 .263a 0.627 0.613 0.648 0.627 a. Predictors: (Constant), Value Cognition, AcctoRes, OrgComp, OrgScop, MgtPer ANOVA(b) Sum of Squares df Mean Square F 1 Regression 7.219 15 1.444 3.436 Residual 97.068 231 0.42 Total 104.287 236 a. Predictors: (Constant), Value Cognition, AcctoRes, OrgComp, OrgScop, MgtPer b. Dependent Variable: EA Assimilation Model
Model 1
(Constant) AcctoRes MgtPer
Coefficients(a) Unstandardized Coefficients B Std. Error 1.094 0.295 0.002 0.072 0.146 0.084
OrgComp OrgScop Value Cognition Parochialism culture resistance a. Dependent Variable: EA Assimilation
F Change 2.336
df1 15
df2 231
Sig. .001
Standardized Coefficients Beta t Sig. 3.71 0 0.002 0.027 0.979 0.17 1.733 0.084
0.173 -0.072 -0.079
0.056 0.058 0.064
0.214 -0.087 -0.119
3.103 -1.242 -1.229
0.002 0.216 0.22
0.073
0.042
0.113
1.733
0.084
Sig. F Change 0.001
228
t2= (2002 - 2004) withoutcoercive pressure moderation Model Summary Change Statistics Model 1
R
R Square
Adjusted R Square
.217a
0.753
0.746
Std. Error of the Estimate
R Square Change
F Change
df1
df2
0.958
0.753
1.119
5
113
a. Predictors: (Constant), Parochialism culture resistance, Value Cognition, OrgComp, AcctoRes, OrgScop ANOVA(b) Model
Sum of Squares 1
Regression
df
Mean Square
5.141
5
1.028
Residual
103.784
113
0.918
Total
108.924
118
F
Sig. 1.119
.151
a. Predictors: (Constant), Parochialism culture resistance, Value Cognition, OrgComp, AcctoRes, OrgScop b. Dependent Variable: EA Assimilation Coefficients(a)
Model 1
Unstandardized Coefficients
Standardized Coefficients
B
Beta
Std. Error
(Constant)
1.604
0.617
AcctoRes
-0.089
0.136
OrgComp
0.153
OrgScop
-0.052
t
Sig. 2.6
0.011
-0.065
-0.651
0.516
0.105
0.148
1.456
0.148
0.127
-0.043
-0.404
0.687
Value Cognition
0.083
0.082
0.095
1.016
0.312
Parochialism culture resistance
0.107
0.086
0.121
1.243
0.216
a. Dependent Variable: EA Assimilation
Sig. F Change 0.151
229
T2 with coercive pressure moderation Model Summary Change Statistics Model
R 1
R Square
.289a
Adjusted R Square
0.756
Std. Error of the Estimate
0.747
R Square Change
F Change
0.756
1.704
0.944
a. Predictors: (Constant), Parochialism culture resistance,coercive pressure, Value Cognition, OrgComp, AcctoRes, OrgScop ANOVA(b) Model
Sum of Squares 1
Regression Residual Total
df
Mean Square
9.112
6
1.519
99.812
112
0.891
108.924
118
F
Sig. 1.704
.062
a. Predictors: (Constant), Parochialism culture resistance,coercive pressure, Value Cognition, OrgComp, AcctoRes, OrgScop b. Dependent Variable: EA Assimilation Coefficients(a)
Model 1
Unstandardized Coefficients
Standardized Coefficients
B
Beta
Std. Error
(Constant)
1.336
0.621
AcctoRes
-0.133
0.136
Institutional Pressure
0.214
OrgComp OrgScop
t
Sig. 2.152
0.034
-0.098
-0.983
0.328
0.101
0.203
2.111
0.037
0.184
0.104
0.179
1.764
0.08
-0.085
0.126
-0.071
-0.668
0.505
Value Cognition
0.053
0.082
0.061
0.65
0.517
Parochialism culture resistance
0.111
0.085
0.125
1.307
0.194
a. Dependent Variable: EA Assimilation
df1
df2 6
Sig. F Change 112
0.062
230
t3= (2005 - 2007) without coercive pressure moderation Model Summary Change Statistics Model
R 1
R Square
.354a
Adjusted R Square
0.638
Std. Error of the Estimate
0.634
R Square Change
F Change
0.638
3.239
1.144
a. Predictors: (Constant), Parochialism culture resistance, OrgComp, Value Cognition, AcctoRes, OrgScop ANOVA(b) Model
Sum of Squares 1
Regression
df
Mean Square
F
21.198
5
4.24
Residual
147.911
113
1.309
Total
169.109
118
Sig. 3.239
.005
a. Predictors: (Constant), Parochialism culture resistance, OrgComp, Value Cognition, AcctoRes, OrgScop b. Dependent Variable: EA Assimilation Coefficients(a)
Model 1
Unstandardized Coefficients
Standardized Coefficients
B
Beta
Std. Error
(Constant)
0.234
0.68
AcctoRes
0.162
0.171
OrgComp
0.232
OrgScop
0.155
Value Cognition Parochialism culture resistance a. Dependent Variable: EA Assimilation
t
Sig. 0.344
0.731
0.088
0.945
0.347
0.124
0.181
1.872
0.064
0.15
0.105
1.029
0.306
-0.002
0.097
-0.002
-0.018
0.986
0.292
0.089
0.311
3.267
0.001
df1
Sig. F Change
df2 5
113
0.005
231
t3= (2005 - 2007) with moderation ofcoercive pressure Model Summary Change Statistics Model
R 1
R Square
.354a
Adjusted R Square
0.593
Std. Error of the Estimate
0.588
R Square Change
F Change
0.593
2.675
1.149
a. Predictors: (Constant), Parochialism culture resistance, OrgComp, Value Cognition,coercive pressure, AcctoRes, OrgScop ANOVA(b) Model
Sum of Squares 1
Regression
df
Mean Square
F
21.198
6
3.533
Residual
147.911
112
1.321
Total
169.109
118
Sig. 2.675
.008a
a. Predictors: (Constant), Parochialism culture resistance, OrgComp, Value Cognition,coercive pressure, AcctoRes, OrgScop b. Dependent Variable: EA Assimilation Coefficients(a)
Model 1
Unstandardized Coefficients
Standardized Coefficients
B
Beta
Std. Error
(Constant)
0.235
0.705
AcctoRes
0.162
0.174
-0.001
OrgComp OrgScop
Institutional Pressure
Value Cognition Parochialism culture resistance a. Dependent Variable: EA Assimilation
t
Sig. 0.334
0.739
0.088
0.933
0.353
0.123
-0.001
-0.007
0.994
0.231
0.126
0.181
1.838
0.069
0.155
0.151
0.105
1.021
0.309
-0.002
0.099
-0.002
-0.016
0.987
0.292
0.09
0.311
3.252
0.002
df1
df2 6
Sig. F Change 112
0.008
232
APPENDIX R: Model Fit Summary
Intercept AcctoRes
T1 (1999 – 2001)
2002 – 2004 w/out CP
2002 – 2004 with CP
2005-2007 w/out CP
2005-2007 With CP
(p-value)
(p-value)
(p-value)
(p-value)
(p-value)
0.017 0.32
0.022
.
0.008
.
0.148
0.136
0.101
0.118
OrgComp
0.044
0.411
0.447
0.207
0.232
OrgScop
0.003
0.627
0.618
0.348
0.286
.
.
.
.
.
.
.
.
QPC QVC
. .
QCP Logit results summary with second order constructs
0.095
0.055
233
APPENDIX S: Multinomial Logistic Regressions Time period t1 Parameter Estimates EA Assimilationa B Std. Error Wald 1 Intercept -38.426 6.846 AcctoRes -4.184 3.395 MgtPer -4.13 3.649 OrgComp 3.494 4.176 OrgScop 42.397 1.005 QPC 5.407 4.574 QVC 0b . . 2 Intercept -38.802 6.873 AcctoRes -4.262 3.398 MgtPer -4.266 3.653 OrgComp 3.759 4.177 OrgScop 41.887 1.014 QPC 5.623 4.576 QVC 0b . . 3 Intercept -42.489 7.116 AcctoRes -3.922 3.425 MgtPer -4.541 3.694 OrgComp 4.245 4.187 OrgScop 42.143 1.074 QPC 5.87 4.593 QVC 0b . . 4 Intercept -49.428 8.099 AcctoRes -3.469 3.542 MgtPer -2.672 3.797 OrgComp 4.753 4.222 OrgScop 41.936 0 . QPC 5.563 4.636 QVC 0b . . a. The reference category is: 0. b. This parameter is set to zero because it is redundant.
31.506 1.518 1.281 0.7 1781.122 1.397 31.872 1.573 1.364 0.81 1707.499 1.51 35.653 1.311 1.511 1.028 1538.339 1.633 37.25 0.959 0.495 1.267 1.44
df 1 1 1 1 1 1 0 1 1 1 1 1 1 0 1 1 1 1 1 1 0 1 1 1 1 1 1 0
Sig.
Exp(B) 0 0.218 0.258 0.403 0 0.237
.
0.015 0.016 32.921 2.59E+18 222.959 .
0 0.21 0.243 0.368 0 0.219 .
.
.
0.23 .
11.005 18.055 154225.6 1.13E+19 2174386 .
2.41E-05 7.65E-06 0.019 2.44E+17 0.044 .
0.031 0.069 115.874 1.63E+18 260.677
.
.
.
.
11.834 20.526 117997.1 1.85E+19 1745595
1.80E-05 1.09E-05 0.012 2.13E+17 0.035
0.02 0.011 69.769 2.01E+18 354.229
0 0.327 0.482 0.26
.
1.96E-05 1.26E-05 0.009 3.61E+17 0.028
0.014 0.014 42.922 1.55E+18 276.772
0 0.252 0.219 0.311 0 0.201 .
95% Confidence Interval for Exp(B) Lower Bound Upper Bound
16.31 14.864 255424.5 1.65E+19 2878205 .
3.01E-05 4.05E-05 0.029 1.63E+18 0.03 .
32.221 117.954 455157.2 1.63E+18 2302778 .
234
Time period t2 (nocoercive pressure) Parameter Estimates EA Assimilationa B Std. Error Wald 2 Intercept 0.268 1.659 AcctoRes -0.159 0.366 MgtPer -0.871 0.489 OrgComp 0.311 0.293 OrgScop -0.252 0.325 QPC 0.654 0.351 QVC 0b . . QCP 0.224 0.264 3 Intercept -0.608 1.864 AcctoRes -0.594 0.406 MgtPer 0.203 0.461 OrgComp 0.351 0.322 OrgScop -0.589 0.396 QPC 0.243 0.349 QVC 0b . . QCP 0.547 0.306 4 Intercept -0.322 4.557 AcctoRes -1.446 1.165 MgtPer 0.112 1.174 OrgComp 0.234 0.695 OrgScop -0.007 0.805 QPC -0.008 0.957 QVC 0b . . QCP -0.188 0.803 5 Intercept -64.882 11.143 AcctoRes 2.421 2.121 MgtPer 1.445 1.692 OrgComp 1.404 0.838 OrgScop 0.269 1.2 QPC 0.424 0.888 QVC 0b . . QCP 16.221 0 . a. The reference category is: 1. b. This parameter is set to zero because it is redundant.
df 0.026 0.189 3.174 1.126 0.6 3.464 0.717 0.106 2.14 0.194 1.185 2.204 0.485 3.19 0.005 1.54 0.009 0.113 0 0 0.055 33.907 1.303 0.729 2.806 0.05 0.227
Sig. 1 1 1 1 1 1 0 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 0 1
Exp(B) 0.871 0.664 0.075 0.289 0.439 0.063
.
0.853 0.418 1.365 0.777 1.923 .
0.397 0.744 0.144 0.659 0.276 0.138 0.486 .
. .
.
1.747 1.091 2.427 1.471 3.826 .
0.745
2.101
0.552 1.225 1.42 0.555 1.275
0.249 0.497 0.755 0.255 0.643
1.224 3.022 2.672 1.207 2.529
.
.
1.728
0.948
3.151
0.235 1.118 1.263 0.993 0.992
0.024 0.112 0.323 0.205 0.152
2.312 11.172 4.932 4.812 6.469
. 0.815 0 0.254 0.393 0.094 0.823 0.634
0.417 0.16 0.768 0.411 0.966
1.251
. 0.074 0.944 0.215 0.924 0.737 0.993 0.993
.
95% Confidence Interval for Exp(B) Lower Bound Upper Bound
.
.
0.829
0.172
4
11.252 4.241 4.069 1.309 1.527
0.176 0.154 0.788 0.124 0.268
718.342 116.924 21.026 13.763 8.714
. 11084417
.
. 11084417
11084417
235
Time period t2 (with coercive pressure) Parameter Estimates 95% Confidence Interval for Exp(B) EA Assimilationa
B 2
3
4
236
5
Intercept AcctoRes MgtPer OrgComp OrgScop QPC QVC [QCP=1] [QCP=2] [QCP=3] Intercept AcctoRes MgtPer OrgComp OrgScop QPC QVC [QCP=1] [QCP=2] [QCP=3] Intercept AcctoRes MgtPer OrgComp OrgScop QPC QVC [QCP=1] [QCP=2] [QCP=3] Intercept
1.311 -0.261 -0.894 0.263 -0.203 0.69 0b -0.476 -1.428 0b 1.221 -0.629 0.229 0.334 -0.624 0.239 0b -1.17 -0.641 0b 0.086 -1.544 -0.065 0.087 0.02 0.13 0b 0.128 -18.722 0b -16.038
Std. Error 1.706 0.375 0.505 0.296 0.329 0.362 . 0.53 0.729 . 1.929 0.411 0.466 0.321 0.419 0.353 . 0.641 0.684 . 4.428 1.181 1.189 0.714 0.782 0.956 . 1.434 0 . 11.167
Wald
df 0.591 0.486 3.138 0.787 0.382 3.629
. 0.808 3.837 . 0.401 2.341 0.242 1.082 2.225 0.457 . 3.335 0.879 . 0 1.71 0.003 0.015 0.001 0.018 . 0.008 . . 2.063
Sig. 1 1 1 1 1 1 0 1 1 0 1 1 1 1 1 1 0 1 1 0 1 1 1 1 1 1 0 1 1 0 1
Exp(B) 0.442 0.486 0.076 0.375 0.536 0.057
.
0.77 0.409 1.3 0.816 1.994 .
0.369 0.05 .
.
.
.
. . .
1.089 2.013 .
0.021 0.091 0.269 0.22 0.175 .
1.136 7.40E-09 .
0.151
.
.
.
1.193 3.132 2.618 1.217 2.538
0.088 0.138
0.214 0.937 1.091 1.02 1.139
0.929
.
.
.
1.755 1.001
0.238 0.505 0.745 0.236 0.635
0.31 0.527
0.985 0.191 0.957 0.903 0.979 0.892
.
.
.
1.605 1.1 2.323 1.554 4.057
0.22 0.057
0.533 1.257 1.396 0.536 1.27
0.068 0.349
Upper Bound
0.369 0.152 0.728 0.428 0.98
0.621 0.24
0.527 0.126 0.623 0.298 0.136 0.499 .
Lower Bound
2.16 9.641 4.426 4.722 7.418 .
0.068 7.40E-09 .
18.888 7.40E-09 .
AcctoRes MgtPer OrgComp OrgScop QPC QVC [QCP=1] [QCP=2] [QCP=3]
2.376 1.423 1.387 0.279 0.452 0b -18.695 -18.41 0b
2.125 1.695 0.837 1.203 0.892 .
1.25 0.705 2.741 0.054 0.257
1 1 1 1 1 0 1 1 0
. 5078.435 6530.093
0 0
0.264 0.401 0.098 0.817 0.612 .
10.763 4.152 4.001 1.321 1.572 .
0.997 0.998
. . . a. The reference category is: 1. b. This parameter is set to zero because it is redundant. c. Floating point overflow occurred while computing this statistic. Its value is therefore set to system missing.
0.167 0.15 0.775 0.125 0.273 .
7.60E-09 1.01E-08 .
0 0 .
693.123 115.099 20.657 13.967 9.034 . .c .c .
Time period t3 (nocoercive pressure) Parameter Estimates
237
EA Assimilationa 2 Intercept AcctoRes MgtPer OrgComp OrgScop QPC QVC QCP 3 Intercept AcctoRes MgtPer OrgComp OrgScop QPC QVC
B -3.787 0.961 -0.355 0.574 0.704 0.106 0b -0.836 -2.811 -0.226 0.326 0.632 0.2 0.155 0b
Std. Error 1.987 0.498 0.565 0.349 0.38 0.409 . 0.35 1.672 0.424 0.456 0.298 0.377 0.317 .
Wald
df 3.634 3.728 0.395 2.705 3.435 0.067
. 5.702 2.826 0.284 0.509 4.507 0.283 0.238 .
Sig. 1 1 1 1 1 1 0 1 1 1 1 1 1 1 0
Exp(B) 0.057 0.054 0.53 0.1 0.064 0.795
.
2.614 0.701 1.775 2.021 1.112 .
0.017 0.093 0.594 0.476 0.034 0.595 0.626 .
95% Confidence Interval for Exp(B) Lower Bound Upper Bound
.
0.986 0.232 0.896 0.96 0.498 .
6.932 2.121 3.516 4.253 2.482 .
0.433
0.218
0.861
0.798 1.385 1.882 1.222 1.167
0.348 0.566 1.05 0.584 0.627
1.83 3.388 3.373 2.557 2.171
.
.
QCP -0.132 0.296 Intercept -6.937 2.51 AcctoRes 0.83 0.544 MgtPer 0.115 0.59 OrgComp 0.433 0.41 OrgScop 0.329 0.523 QPC 0.76 0.401 QVC 0b . . QCP -0.1 0.373 5 Intercept -8.76 4.68 AcctoRes 0.733 1.035 MgtPer -1.544 1.618 OrgComp 0.604 0.827 OrgScop 1.031 0.901 QPC 1.356 0.961 QVC 0b . . QCP 0.272 0.762 a. The reference category is: 1. b. This parameter is set to zero because it is redundant. 4
0.198 7.641 2.332 0.038 1.119 0.398 3.592
1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 0 1
0.072 3.504 0.501 0.911 0.533 1.308 1.99 0.128
0.656 0.006 0.127 0.845 0.29 0.528 0.058 .
0.877
0.491
1.566
2.294 1.122 1.543 1.39 2.139
0.79 0.353 0.691 0.499 0.974
6.662 3.57 3.444 3.872 4.694
. 0.789 0.061 0.479 0.34 0.465 0.253 0.158
.
. 0.436
1.88
2.081 0.214 1.829 2.803 3.879
0.274 0.009 0.362 0.479 0.59
15.826 5.09 9.254 16.396 25.5
. 0.721
.
0.905
. 1.313
. 0.295
5.842
Time period t3 (withcoercive pressure) Parameter Estimates
238
EA Assimilationa 2 Intercept AcctoRes MgtPer OrgComp OrgScop QPC QVC [QCP=1] [QCP=2] [QCP=3] 3 Intercept AcctoRes MgtPer
B -5.825 0.823 -0.246 0.503 0.736 0.058 0b 1.544 -0.533 0b -3.216 -0.241 0.336
Std. Error 2.069 0.495 0.582 0.351 0.379 0.417 . 0.67 1.199 . 1.753 0.424 0.46
Wald
df 7.928 2.758 0.179 2.054 3.766 0.019
. 5.307 0.197 . 3.366 0.322 0.533
Sig. 1 1 1 1 1 1 0 1 1 0 1 1 1
Exp(B) 0.005 0.097 0.672 0.152 0.052 0.889
.
2.277 0.782 1.654 2.087 1.06 .
0.021 0.657 .
0.862 0.25 0.831 0.993 0.468 .
4.682 0.587 .
0.067 0.57 0.465
95% Confidence Interval for Exp(B) Lower Bound Upper Bound
. 1.259 0.056
. 0.786 1.399
6.011 2.446 3.294 4.388 2.402 17.413 6.157 .
0.342 0.568
1.805 3.444
OrgComp 0.625 0.297 OrgScop 0.224 0.385 QPC 0.157 0.318 QVC 0b . . [QCP=1] 0.266 0.589 [QCP=2] -0.033 0.741 [QCP=3] 0b . . 4 Intercept -7.253 2.823 AcctoRes 0.979 0.58 MgtPer 0.014 0.626 OrgComp 0.535 0.423 OrgScop 0.042 0.603 QPC 0.783 0.42 QVC 0b . . [QCP=1] -0.059 0.856 [QCP=2] 1.539 0.853 [QCP=3] 0b . . 5 Intercept -7.294 4.533 AcctoRes 0.544 1.025 MgtPer -1.522 1.678 OrgComp 0.505 0.833 OrgScop 1.048 0.862 QPC 1.364 1.015 QVC 0b . . [QCP=1] -0.415 1.381 [QCP=2] -18.533 0 . [QCP=3] 0b . . a. The reference category is: 1. b. This parameter is set to zero because it is redundant.
4.42 0.337 0.244 0.203 0.002 6.6 2.85 0.001 1.597 0.005 3.484 0.005 3.256 2.589 0.282 0.823 0.368 1.477 1.807 0.09
1 1 1 0 1 1 0 1 1 1 1 1 1 0 1 1 0 1 1 1 1 1 1 0 1 1 0
0.036 0.562 0.621 .
1.868 1.251 1.17 .
0.652 0.964 .
.
.
.
5.047 24.781 .
0.231 0.008 0.324 0.526 0.535 .
0.661 8.94E-09 .
.
.
.
8.287 3.458 3.914 3.4 4.982
0.176 0.876
1.723 0.218 1.658 2.851 3.913
0.764
.
.
.
4.136 4.134
0.854 0.297 0.745 0.32 0.962
0.943 4.659
0.108 0.595 0.364 0.544 0.224 0.179
.
.
.
3.346 2.661 2.184
0.411 0.226
2.661 1.014 1.707 1.043 2.189
0.945 0.071
. .
. 1.304 0.968
0.01 0.091 0.982 0.206 0.945 0.062 .
1.043 0.588 0.627
12.849 5.851 8.491 15.448 28.6 .
0.044 8.94E-09 .
9.889 8.94E-09 .
239
APPENDIX T: Factor Loadings Time period t1
EA Assim 1 0.045 0.006 0.033
Skilled Res
Fund
EA Assimilation Funding 1 . Skilled Resources .133* 1 Acc to Resources .731** .774** Parochialism culture resistance 0.067 0.04 0.039 Value recognition -0.005 0.093 -0.059 Mission Heterogeneity .240** -0.116 -0.046 Administrative multiplicity 0.071 -0.055 -0.072 Regulations 0.102 -0.028 0.066 Org. Complexity .220** -0.104 -0.027 Jurisdiction -0.103 -0.099 -0.048 Autonomy -.166** -.154* -.231** Org. Scope -.177** -.179** -.221** **. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
OrgComp
240
Funding Skilled Resources Mission Heterogeneity Administrative multiplicity Regulations Jurisdiction Autonomy Value Cognition Parochialism culture resistance
Acc To Res
Paroch culture resist
Value Recog
Mission Hetero
Admin multiplicity
Regul
Org Comp
Jurisd
Auton
OrgScop
1 .113* .553** -0.024 0.044 0.025
1 .675** -.149** -.418** -.407**
1 -.107* -.300** -.293**
1 .174** .581**
1 .905**
1
. 1 0.052 0.019 -0.106
1 .159* -0.091
1 0.01
1
-0.085 0.027 -0.085 -0.096 -.259** -.266**
-0.108 -.163* -.178** -0.007 -0.003 0.002
-.147* -0.012 -0.063 0.031 0.07 0.066
0.085 .270** .748** -0.045 -.213** -.195**
Rotated Component Matrix(a) Component AcctoRes OrgScop 0.708 0.776
0.761 0.768 0.728 0.764 0.692
4
EA Assimilation Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. a. Rotation converged in 5 iterations.
0.483
Time period t2
EA Assimilation Institutional Pressure Funding Skilled Resources Acc to Resources Parochialism culture resistance Value Cognition Mission Heterogeneity Administrative multiplicity Regulations OrgComplexity Jurisdiction Autonomy
Paroch culture resistance
Value Recog
Mission Hetero
.253** 0.076 -0.086
1 0.051 -0.063
1 0.038
1
-0.14 0.02 -0.041
-0.155 0.145 -0.054
-0.121 -0.115 -0.137
-0.116 0.015 -0.021
0.002
-0.01
0.016
-.290**
-.230*
-0.007 .341** .285**
0.071
1
.589**
.848**
1
0.096 0.087 .180**
0.039 0.074 -.191*
.287** 0.045 0.019
0.104 0.002 .148** 0.056 .138**
-0.078 .243** -0.04
Fund
Skilled Res
Acc To Res
EA Assim 1 .185** 0.027 0.011 0.023
Admin multiplicity
Regula
Org Comp
.269** .249** .794**
1 .154** .647**
1 .632**
1
0.076
-.154**
-.129*
-.147**
-.206**
1
-.212*
0.129
-.252**
-0.007
-.413**
-.322**
.181**
1
-0.168
0.139
-.277**
-0.063
-.404**
-.357**
.581**
.907**
Jurisd
Auton
Org Scope
1
OrgScop -.134* -.239** -.195* **. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
1
241
T2 withcoercive pressure Rotated Component Matrix(a) Component OrgScop AcctoRes 0.811 0.945
OrgComp Funding Skilled Resources Mission Heterogeneity 0.791 Administrative multiplicity 0.702 Regulations 0.488 Jurisdiction Autonomy Value Cognition Parochialism culture resistance Institutional Pressure EA Assimilation Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. a. Rotation converged in 7 iterations.
0.595 0.671 0.761 0.492
T2 withoutcoercive pressure
Funding Skilled Resources Mission Heterogeneity Administrative multiplicity Regulations Jurisdiction Autonomy Value Cognition Parochialism culture resistance EA Assimilation a. Rotation converged in 6 iterations.
4
Rotated Component Matrix(a) Component OrgComp OrgScop AcctoRes 0.817 0.946 0.78 0.708 0.491 0.687 0.766
242
Time period t3
EA Assimil 1 0.054 -0.106 .240** 0.145
Fund
Skilled Res
EA Assimilation Institutional Pressure Funding 1 Skilled Resources 0.021 1 AcctoRes .546** .849** Parochialism culture resistance .301** 0.014 .273** Value Cognition 0.062 0.119 .186* Mission Heterogeneity .173** -0.066 0.079 Administrative multiplicity .118* -0.108 -0.082 Regulations 0.039 0.045 -0.009 OrgComp .166** -0.065 0.006 Jurisdiction -0.073 0.132 0.059 Autonomy -.135* -0.127 -0.141 OrgScop -.143** -0.049 -0.092 **. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
Acc To Res
Paroch culture resistance
Value recog
Mission Hetero
.236** .219* 0.031
1 0.045 0.064
1 0.119
1
-0.125 0.017 -0.029 0.119 -.186* -0.103
-0.095 0.021 0.005 -0.121 -.299** -.299**
-0.016 0.094 0.101 0.179 0.047 0.116
.276** .250** .798** -.160** -.256** -.281**
OrgScop
Rotated Component Matrix(a) Component OrgComp 3
5 0.433
-0.511 0.482 0.889
Regulat
Org. Comp
Jurisd
Auton
Org. Scope
1 .151** .649** -.133* -0.002 -0.058
1 .628** -.147** -.412** -.404**
1 -.210** -.321** -.356**
1 .185** .581**
* 1 .907**
1
1
T3 withcoercive pressure
243
Funding Skilled Resources Mission Heterogeneity Administrative multiplicity Regulations Jurisdiction Autonomy Value Cognition Parochialism culture resistance Institutional Pressure
Admin multiplicity
0.741 0.734 0.498 0.401 0.644 0.631 0.751
EA Assimilation Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. a. Rotation converged in 9 iterations.
0.737
T3 withoutcoercive pressure Rotated Component Matrix(a) Component OrgComp AcctoRes
OrgScop Funding Skilled Resources Mission Heterogeneity Administrative multiplicity Regulations -0.452 Jurisdiction 0.558 Autonomy 0.868 Value Cognition Parochialism culture resistance EA Assimilation Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. a. Rotation converged in 7 iterations.
0.691 0.601 0.753 0.708 0.525
0.478
244
APPENDIX U: Results for Hypothesis # 4: EA Assimilation Enhances Strategic Business Outcomes Time period t1 (1999-2001) t1= (1999 - 2001) Descriptive Statistics
Correlations
Mean Std. Deviation EA Assimilation 1.33 0.621 Strategic Business Outcomes 69.4 1.133 **. Correlation is significant at the 0.01 level (2-tailed).
Model Intercept Only Final
Model Fitting Information Model Fitting Criteria -2 Log Likelihood Chi-Square 82.47 20.858 61.613
Strategic Business Outcomes .396**
EA Assimilation 1 .396**
1
Likelihood Ratio Tests df Sig. 3
0
Pseudo R-Square Cox and Snell 0.159 Nagelkerke 0.204 McFadden 0.115 Parameter Estimates EA Assimilation 2 Intercept QSBO 3 Intercept QSBO 4 Intercept QSBO
B -53.262 0.749 -66.054 0.909 -409.775 5.731
Std. Error 8.063 0.116 16.195 0.231 0.593 0
Wald 43.635 41.996 16.635 15.527 477691.9 .
df 1 1 1 1 1 1
Sig.
Exp(B) 0 0 0 0 0
.
95% Confidence Interval for Exp(B) Lower Bound Upper Bound
2.114
1.686
2.651
2.482
1.579
3.9
308.125
308.125
308.125
245
Time period t2 (2002-2004) T2 (2002-2004) Descriptive Statistics
Correlations
Mean Std. Deviation EA Assimilation 2.12 0.91 Strategic Business Outcomes 71.07 0.786 **. Correlation is significant at the 0.01 level (2-tailed).
Model Intercept Only Final Pseudo R-Square Cox and Snell Nagelkerke McFadden
Model Fitting Information Model Fitting Criteria -2 Log Likelihood Chi-Square 85.401 55.311 30.09
Strategic Business Outcomes .269**
EA Assimilation 1 .269**
1
Likelihood Ratio Tests df Sig. 4
0
0.081 0.088 0.033 Parameter Estimates
EA Assimilation B 2 Intercept QSBO 3 Intercept QSBO 4 Intercept QSBO 5 Intercept QSBO a. The reference category is: 1.
Std. Error -15.619 0.227 -57.565 0.809 -98.246 1.357 -45.882 0.602
Wald 12.716 0.179 14.246 0.2 25.408 0.355 46.238 0.65
df 1.509 1.602 16.327 16.283 14.952 14.59 0.985 0.858
Sig. 1 1 1 1 1 1 1 1
Exp(B) 0.219 0.206 0 0 0 0 0.321 0.354
95% Confidence Interval for Exp(B) Lower Bound Upper Bound
1.255
0.883
1.783
2.245
1.516
3.325
3.886
1.936
7.797
1.825
0.511
6.518
246
Time period t3 (2005-2007) T3 (2005-2007) Descriptive Statistics Mean Std. Deviation EA Assimilation 2.12 0.91 Strategic Business Outcomes 71.07 0.786 **. Correlation is significant at the 0.01 level (2-tailed).
Model Intercept Only Final
Model Fitting Information Model Fitting Criteria -2 Log Likelihood Chi-Square 101.154 56.291 44.863
Correlations EA Assimilation
SBO -.214**
1 -.214**
1
Likelihood Ratio Tests df Sig. 4
0
Pseudo R-Square Cox and Snell 0.118 Nagelkerke 0.125 McFadden 0.043 Parameter Estimates EA Assimilation B 2 Intercept QSBO 3 Intercept QSBO 4 Intercept QSBO 5 Intercept QSBO a. The reference category is: 1.
Std. Error 35.889 -0.507 21.472 -0.303 32.46 -0.466 39.251 -0.586
Wald 6.49 0.091 6.697 0.094 7.167 0.101 11.568 0.165
df 30.58 30.706 10.279 10.346 20.515 21.193 11.512 12.536
Sig. 1 1 1 1 1 1 1 1
Exp(B) 0 0 0.001 0.001 0 0 0.001 0
95% Confidence Interval for Exp(B) Lower Bound Upper Bound
0.602
0.503
0.721
0.739
0.614
0.888
0.628
0.515
0.765
0.557
0.403
0.77
247
APPENDIX V: OLS Correlation Matrix Descriptive Statistics
Mean 1.33 1.855 2.0399 1.8249 2.2745
Std. Deviation 0.621 0.61709 0.77506 0.81156 0.80831
t1= (1999 - 2001) Correlations EA Assimilation 1 0.014 0.04 .215** -.152**
EA Assimilation AcctoRes MgtPer OrgComp OrgScop Parochialism culture resistance 2.03 1.031 0.072 Mgmt Value Recognition 2.05 1.003 -0.011 **. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
Descriptive Statistics Mean 2.12 2.00 2.105 2.2059 1.9739 2.2731
Std. Deviation 0.91 0.958 0.70372 0.79301 0.92574 0.80748
EA Assimilation Institutional Pressure AcctoRes MgtPer OrgComp OrgScop Parochialism culture resistance 2.37 1.088 Value Cognition 2.04 1.1 **. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
AcctoRes
MgtPer
OrgComp
OrgScop
1 0.045 -0.085 -.266**
1 -.160* 0.046
1 -.293**
1
0.052 0.019
.770** .753**
-.178** -0.063
0.002 0.066
Time period t2 (2002-2004) Correlations EA Institutional Assimilation Pressure AcctoRes 1 .185** 1 -0.023 0.139 1 0.127 0.158 .227* .144** -.203** -0.054 -.128* .168** -.285** 0.096 0.087
0.026 .203*
.253** 0.076
Parochialism culture resistance
Value Cognition
1 .159*
1
MgtPer
OrgComp
OrgScop
1 -0.108 -0.019
1 -.357**
1
.721** .728**
-0.137 -0.021
-0.168 0.139
Parochialism culture resistance
1 0.051
248
Descriptive Statistics Mean 2.41 2 2.1429 2.3277 1.9944 2.2731
Std. Deviation 1.142 0.909 0.65141 0.87439 0.93481 0.80748
EA Assimilation Institutional Pressure AcctoRes MgtPer OrgComp OrgScop Parochialism culture resistance 2.52 1.275 Value Cognition 2.13 1.142 **. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
T3 (2005-2007) Correlations EA Institutional Assimilation Pressure 1 0.054 1 0.145 0.164 .260** 0.117 .166** -.193** -.143** .172** .301** 0.062
-0.015 .195*
AcctoRes
MgtPer
OrgComp
OrgScop
1 .315** -0.029 -0.103
1 0.07 -0.142
1 -.356**
1
.236** .219*
.758** .685**
0.005 0.101
-.299** 0.116
Parochialism culture resistance
1 0.045
249
APPENDIX W: Moderation Testing Results H2a:coercive pressure will significantly influence EA assimilation at phase t3 Complete Model Regression Summary R-sq F df1 df2 p .0760 9.6452 3.0000 352.0000
n .0000 356.0000
=================================================================== b se t p constant 70.3973 0.2636 267.0572 0 QEA 0.362 0.1198 3.0208 0.0027 QCP 0.0764 0.1194 0.6398 0.5227 interact -0.057 0.0512 -1.1141 0.266 Interact is defined as: QEA X QCP ===================================================================== Conditional Effect of Focal Predictor at Values of the Moderator Variable QCP b se t p LLCI(b) ULCI(b) 1.0943 .2995 .0712 4.2081 .0000 .1595 .4395 2.0028 .2477 .0461 5.3770 .0000 .1571 .3383 2.9113 .1959 .0592 3.3096 .0010 .0795 .3123 Alpha level used for confidence intervals: 0.05 Moderator values are the sample mean and plus/minus one SD from mean ------ END MATRIX -----
250
H2b:coercive pressure will positively moderate management value recognition and EA assimilation at t2 and t3 Time period: t2 Complete Model Regression Summary R-sq F df1 df2 p .0611 1.8395 4.0000 113.0000
n .1261 118.0000
=================================================================== b se t p constant 1.5663 0.5254 2.9811 QSZ 0.0772 0.0386 2.002 QVC -0.0535 0.2416 -0.2216 QCP 0.1055 0.2247 0.4695 interact 0.0455 0.1016 0.4475 Interact is defined as: QVC X QCP ===================================================================== Conditional Effect of Focal Predictor at Values of the Moderator Variable QCP b se t p LLCI(b) ULCI(b) 1.0810 -.0044 .1436 -.0304 .9758 -.2888 .2801 1.9915 .0370 .0869 .4261 .6708 -.1351 .2092 2.9020 .0784 .1078 .7278 .4682 -.1351 .2919 Alpha level used for confidence intervals: 0.05 Moderator values are the sample mean and plus/minus one SD from mean ------ END MATRIX -----
0.0035 0.0477 0.8251 0.6396 0.6554
251
Time period: t3 Complete Model Regression Summary R-sq F df1 df2 p .0139 .3981 4.0000 113.0000
n .8097 118.0000
=================================================================== b se t p constant 1.4362 0.6509 2.2064 0.0294 QSZ 0.0108 0.0484 0.2225 0.8243 QVC 0.3628 0.2999 1.2098 0.2289 QCP 0.2614 0.2821 0.9265 0.3562 interact -0.1448 0.1247 -1.1611 0.248 Interact is defined as: QVC X QCP ===================================================================== Conditional Effect of Focal Predictor at Values of the Moderator Variable QCP b se t p LLCI(b) ULCI(b) 1.0810 .2063 .1790 1.1527 .2515 -.1483 .5610 1.9915 .0745 .1070 .6964 .4876 -.1375 .2866 2.9020 -.0573 .1290 -.4440 .6579 -.3128 .1983 Alpha level used for confidence intervals: 0.05 Moderator values are the sample mean and plus/minus one SD from mean ------ END MATRIX -----
252
H2c:coercive pressure will negatively moderate organization complexity and EA assimilation at t2 and t3 Time period: t2 Complete Model Regression Summary R-sq F df1 df2 p .1002 9.6905 4.0000 348.0000
n .0000 353.0000
=================================================================== b se t p constant 1.7979 0.2684 6.6991 QSZ 0.0569 0.0221 2.5758 OrgComp -0.127 0.1184 -1.0723 QCP -0.0505 0.1262 -0.3997 interact 0.1505 0.0587 2.5621 Interact is defined as: OrgComp X QCP ===================================================================== Conditional Effect of Focal Predictor at Values of the Moderator Variable QCP b se t p LLCI(b) ULCI(b) 1.0867 .0365 .0688 .5305 .5961 -.0988 .1718 1.9943 .1731 .0567 3.0525 .0024 .0616 .2846 2.9020 .3096 .0859 3.6050 .0004 .1407 .4786 Alpha level used for confidence intervals: 0.05 Moderator values are the sample mean and plus/minus one SD from mean ------ END MATRIX -----
0 0.0104 0.2843 0.6896 0.0108
253
Time period: t3 Complete Model Regression Summary R-sq F df1 df2 p .0421 3.8385 4.0000 349.0000
n .0046 354.0000
=================================================================== b se t p constant 2.3537 0.3483 6.7568 0 QSZ -0.0015 0.0285 -0.0526 0.9581 OrgComp -0.0613 0.1523 -0.4023 0.6877 QCP -0.2106 0.1638 -1.2857 0.1994 interact 0.156 0.0753 2.0719 0.039 Interact is defined as: OrgComp X QCP ===================================================================== Conditional Effect of Focal Predictor at Values of the Moderator Variable QCP b se t p LLCI(b) ULCI(b) 1.0836 .1078 .0889 1.2116 .2265 -.0672 .2827 1.9915 .2494 .0731 3.4115 .0007 .1056 .3932 2.8994 .3910 .1101 3.5513 .0004 .1745 .6076 Alpha level used for confidence intervals: 0.05 Moderator values are the sample mean and plus/minus one SD from mean ------ END MATRIX -----
254
APPENDIX X: Interactions H2b (i):coercive pressure will positively moderate management value recognition and EA assimilation
Name of moderator:
VoC Institutional Pressure
Unstandardized Regression Coefficients: Independent variable: Moderator: Interaction:
0.093 0.088 -0.013
Intercept / Constant:
1.763
Means / SDs of variables: Mean of independent variable: SD of independent variable: Mean of moderator: SD of moderator:
5 4.5 EA Assimilation
Variable names: Name of independent variable:
4
3.5 Low Coercive Pressure
3
2.5
High Coercive Pressure
2 1.5 2.09 1.12 2.08 1.084
1 Low VoC
High VoC
The interaction at t2 shows thatcoercive pressure (both high and low) leads to higher assimilation levels with high management value recognition. However, highercoercive pressure seems to have a higher impact in driving assimilation levels with both low and high management value recognition.
255
H2b (ii):coercive pressure will negatively moderate organization complexity and EA assimilation
Name of moderator:
OrgComp Institutional Pressure
Unstandardised Regression Coefficients: Independent variable: Moderator: Interaction:
-0.04 -0.129 0.151
Intercept / Constant:
2.119
Means / SDs of variables: Mean of independent variable: SD of independent variable: Mean of moderator: SD of moderator:
5 4.5 EA Assimilation
Variable names: Name of independent variable:
4 3.5 Low Coercive Pressure
3 2.5
High Coercive Pressure
2 1.5 1.984 0.92939 2 0.908
1 Low OrgComp
High OrgComp
The interaction shows that highcoercive pressure leads to higher assimilation, with both low and high organization complexity
256
REFERENCES Aiken, M., & Hage, J. 1971. The organic organization and innovation. Sociology, 5: 63– 82. Alchian, A. A., & Demsetz, H. 1972. Production, Information costs, and economic organization. American Economic Review, 46(5): 777–795. Aldrich, J. H., & Nelson, F. D. 1984. Linear probability, logit, and probit models. Newbury Park, CA: Sage Publications. Aldrich, H. E., & Fiol, C. M. 1994. Fools rush in? The institutional context of industry creation. Academy of Management Review, 19(4): 645–670 Allison, P. D. 1999. Logistic regression using the SAS system: Theory and application. SAS Publishing. Andrues, W. 2006. The Clinger-Cohen Act, 10 years later: Becoming enterprise architects. Government Executive [online], July 25, 2006. Available at http://www.govexec.com/technology/2006/07/the-clinger-cohen-act-10-yearslater-becoming-enterprise-architects/22329/ Ansari, S., Bell, J., & Lundblad, H.. 2002. TransEuro corporation: A management practices case. Strategic Finance (July): 52–57. (Student case competition analyze this company's cost using ABM). Armstrong, C. P., & Sambamurthy, V. 1999. Information technology assimilation in firms. Information Systems Research, 10(4): 304-327. Ash, J. 1997. Organizational factors that influence information technology diffusion in academic health sciences centers. Journal of the American Medical Informatics Association, 4: 102–111. Ashworth, R., Boyne, G., & Delbridge, R. 2005. Institutional pressures on public organizations: An empirical test of isomorphism. Paper presented at the Public Management Research Association Conference, Los Angeles, CA. Attewell, P. 1992. Technology diffusion and organizational learning: The case of business computing. Organization Science, 3(1): 1–19. Aucoin, P. 1990. Administrative reform in public management: Paradigms, principles, paradoxes and pendulums. Governance, 3: 115–137.
257
Bainbridge, W. S. 2003. Religious opposition to cloning. Journal of Evolution and Technology, 13, October 2003. Available at http://jetpress.org/volume13/bainbridge.htm Bajwa, D. S., Garcia, J. E., & Mooney T. 2004. An integrative framework for the assimilation of enterprise resource planning systems: Phases, determinants, and outcomes. Journal of Computer Information Systems, 44(3): 81–90. Baron, R. M., & Kenny, D. A. 1986. The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6): 1173–1182. Barua, A., Konana, P., Whinston, A. B., Yin F. 2004. Assessing Internet enabled business value: An exploratory investigation. MIS Quarterly, 28(4): 585–620. Barzelay, M. 1995. Public management: The state of the art. B. Bozeman (Ed.). San Fracisco: Jossey-Bass. Beamer, R., Henning, P., & Cullen, R. 2004. The USNORTHCOM integrated architecture: Developing and managing a capabilities-based architecture as a program to enhance the homeland defense and military assistance to civil authorities mission areas. MITRE. Available at http://www.dtic.mil/cgibin/GetTRDoc?AD=ADA456326 Bellman, B., & Rausch, F. 2004. Enterprise architecture for e-government. Lecture Notes in Computer Science, 3183: 48–56. Bernard, S. 2005. An introduction to enterprise architecture (2nd ed.). Bloomington, IL: AuthorHouse. Bhalla, A. S., & James, D. (Eds.). 1988. New technologies and development: Experiences in technology blending. Boulder, CO: Lynner Reinner. Bittler, R. S., & Kreizman, G. 2005. Gartner enterprise architecture process: Evolution 2005. Stamford, CT: Gartner Inc. Blau, P. M. 1970. A formal theory of differentiation in organization. American Sociological Review, 35: 201–218. Boddy, D., Boonstra, A., & Kennedy, G. 2003. Managing information systems: Strategy and organisation. Pearson Education Limited (United Kingdom). Bolgherini, S. 2007. The technology trap and the role of political and cultural variables: A critical analysis of the e-government policies. Review of Policy Research, 24(3): 259–275. 258
Bowornwathana, B. 1994a. Democratic reform visions and the reinvention of Thai public officials. Paper presented to the seminar on Comparative Public Sector Reform hosted by EROPA, the Australian Public Service Commission, and the University of Canberra, 28-30 November, Canberra. Bowornwathana, B. 1994b. Administrative reform and regime shifts: Reflections on the Thai Polity. Asian Journal of Public Administration, 16(2): 152. Boonstra, A., & de Vries, J. 2008. Managing stakeholders around inter-organisational systems: a diagnostic approach. Journal of Strategic Information Systems, 17(3): 190–201. Bozeman, B., & Loveless, S. 1987. Sector context and performance: a comparison of industrial and government research units. Administration and Society. 19: 290– 322. Buchanan, R., & Soley, R. M. 2002. Aligning enterprise architecture and IT investments with corporate goals. An OMG Whitepaper. OMG and Meta Group. Burgess, D., & Fried, J. 2002. The foreign property rule: A cost-benefit analysis. Unpublished manuscript (Canada: University of Western Ontario). Burk, R. 2005. Updates on the Federal Enterprise Architecture Program. ISPAB Quarterly Meeting. December 7, 2005. Available at http://csrc.nist.gov/groups/SMA/ispab/documents/minutes/2005-12/D_BurkDec2005-ISPAB.pdf Bryman, A. 1988. Quantity and quality in social research. London: Routledge Brynjolfsson, E., & Kemerer, C. 1996. Network externalities in microcomputer software: An econometric analysis of the spreadsheet market. Management Science, 42(12): 1627–1647. Caiden, G. E. 1991. What really is public maladministration? Public Administration Review, 51(6): 486–493. Blackwell Publishing. Chatterjee, D., Grewal, R., & Sambamurthy, V. 2002. Shaping up for e-commerce: Institutional enablers of the organizational assimilation of web technologies. MIS Quarterly, 26(2): 65–89. Chau, P. Y. K., & Tam, K. Y. 1997. Factors affecting the adoption of open systems: An exploratory study. MIS Quarterly, 21(1): 1–21. Chin, W. W. 1998. The partial least squares approach to structural equation modeling. In G. A. Marcoulides (Ed.), Modern methods for business research: 295–336. Hillsdale, NJ: Lawrence Erlbaum Associates. 259
Chin, W. W., & Newsted, P. R. 1999. Structural equation modeling analysis with small samples using partial least squares. In Rick Hoyle (Ed.), Statistical strategies for small sample research: 307–341. Sage Publications. Cohen, W. M., & Levinthal, D. A. 1990. Absorptive capacity: A new perspective on learning and innovation. Administrative Sciences Quarterly, 35: 569–596 Cooper, R. B., & Zmud, R. W. 1990. Information technology implementation research: A technological diffusion approach. Management Science, 36(2): 123-139. Cooprider, J. G., & Victor, K. M. 1993. The contribution of shared knowledge to IS group performance. J. I. DeGross, R. P. Bostrom & D. Robey (Eds.), Proceedings of the 14th International Conference on Information Systems, Orlando, FL, December 5–8, 1993, pp. 285–299. Creswell J. W. 2006. Qualitative inquiry & research design: Choosing among five approaches. SAGE Publications. Crotty, M. 1998. The foundations of social research: Meaning and perspective in the research process. London: Sage Publications. Crow, M., & Bozeman. B. 1998. Limited by design—R&D laboratories in the U.S. national innovation system. New York: Columbia University Press. Damsgaard, J., & Lyytinen, K. 1996. Government strategies to promote the diffusion of electronic data interchange (EDI): What we know and what we don’t know. Information Infrastructure and Policy, 5(3): 169–190. Damanpour, F. 1988. Innovation type, radicalness, and the adoptive process. Communication Research, 15(5): 545–567. Damanpour F. 1991 Organizational innovation: A meta-analysis of effects of determinants and moderators. Academy of Management Journal, 34: 555–590. Damanpour, F. 1991. Organizational innovation: A meta-analysis of effects of determinants and moderators. Academy of Management Journal, 34(3): 555– 590. Dasgupta, S., & Agarwal, D., Ioannidis, A., & Gopalakrishnan S. 1999. Determinants of information technology adoption: An extension of existing models to firms in a developing country. Journal of Global Information Management, 7(3): 41–49. Datta, S. K., & Nugent, J. B. 1986. Adversary activities and per capita income growth. World Development, 14(12): 1457–1461. 260
Davenport, T. H. 1998. Putting the enterprise into the enterprise system. Harvard Business Review, 76(4): 121-131. Davis, F. 1989. Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13(3): 319–340. DeLone, W. H., & McLean, E. R. 2003. The DeLone and McLean model of information systems success: A ten-year update. Journal of Management Information Systems, 19(4): 9–30. DeLone, W. H., & McLean, E. R. 1992. Information systems success: The quest for the dependent variable. Informations Systems Research, 3(1): 60–95. DeLuca, J., & McDowell, R. 1992. Managing diversity: A strategic ‘grass roots’ approach. In S. E. Jackson and others (cdi), Diversity in the workplace: Human resources motives (Vol. I). New York: Guilford Press. Dess, G. G., & Beard, D. W. 1984. Dimensions of organizational task environments. Administrative Science Quarterly, 29: 52–73. Diamantopoulos, A., & Winklhofer, H. M. 2001. Index construction with formative indicators: An alternative to scale development. Journal of Marketing Research, 38(2): 259–277. DiMaggio, P. J., & Powell, W. W. 1983. The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48(2): 147–160. DiMaggio, P. 1992. Cultural boundaries and structural change. In M. Lamont & M. Fournier (Eds.), Cultivating differences: 21–57. University of Chicago Press. DiMaggio, P. J., & Powell, W. W. 1991. Introduction. In W. W. Powell & P. J. DiMaggio (Eds.), The new institutionalism in organizational analysis: 1-38. The University of Chicago Press. Dooley, K. 2002. Organizational complexity. In M. Warner (Ed.), International encyclopedia of business and management: 5013–5022. London: Thompson Learning. Downs, G., & Mohr, L. 1976. Conceptual issues in the study of innovation. Administrative Science Quarterly, 21: 700–714. Eby, D. W., Trombley, D. A., Molnar, L. J., & Shope, J. T. 1998. The assessment of older driver's capabilities: A review of the literature. Ann Arbor: The University of Michigan. 261
Efron, B., & Tibshirani, R. J. 1993. An introduction to the bootstrap. New York: Chapman & Hall. Ereaut, G. 2007. What is qualitative research? QSR International. Eveland, J. D., & Tornatzky, L. G. 1990. The deployment of technology. In L. G. Tornatzky & M. Fleischer (Ed.), The processes of technological innovation: 117–148. Lexington, MA: Lexington Books. Fichman, R. G. 2000. The diffusion and assimilation of information technology innovations. R. W. Zmud (Ed.), Framing the domains of IT management: Projecting the future through the past: 105–127. Pinnaflex Publishing. Fichman, R. G., & Kemerer, C. F. 2002. Activity based costing for component-based software development. Information Technology and Management, 3(1-2): 137– 60. Fichman, R. G. 2001. The role of aggregation in the measurement of it-related organizational innovation. MIS Quarterly, 25(4): 427–455. Fichman, R. G., & Kemerer, C. F. 1999. The illusory diffusion of innovation: An examination of assimilation gaps. Information Systems Research, 10(3): 255275. Fichman, R. G., & Kemerer, C. F. 1997. The assimilation of software process innovations: An organizational learning perspective. Management Science, 43(1): 1345–1363. Fichman, R. G. 2004. Going beyond the dominant paradigm for IT innovation research: emerging concepts and methods. Journal of the Association for Information Systems, 5(8): 314–355. Fioretti, G., & Visser, B. 2004. A cognitive approach to organizational complexity. In G. Minati, E. Pessa, & M. R. Abram (Eds.), Systemics of emergence: research and development: 495-514. Springer. Fiss, P. C., & Kennedy, M. T. 2009. Frames and framing in financial markets. Academy of Management Meetings, Anaheim, CA. Ford, J., & Slocum, J. 1977. Size, technology, environment and the structure of organizations. Academy of Management Review, 2: 561–575. Fornell, C. 1983. Issues in the application of covariance structure analysis: A comment. Journal of Consumer Research, 9(March): 443–448. 262
Garfinkel, H. 1967. Studies in ethnomethodology. Englewood Cliffs, NJ: Prentice-Hall. Gelfand, M. 1975. A national compendium of cities. New York: Oxford University Press. Gentile, M. (Ed.). 1994. Differences that work: Organizational excellence through diversity. Boston: Harvard Business School Press. Glaser, B., & Strauss, A. 1967. The discovery of grounded theory. Chicago: Aldine Publishing Company. Goffman, E. 1974. Frame analysis: An essay on the organization of experience. New York, NY: Harper & Row. Granlund, M., & Lukka, K. 1998. Towards increasing business orientation: Finnish management accountants in a changing culture. Management Accounting Research, 9: 185–211. Grasso, A. 2009. Information technology acquisition: A common-sense approach. The Mitre Report. Greenleaf, K. 1977. Servant leadership. Mahwah, NJ: Paulist Press Spears. Greve, H. R. 2000. Market niche entry decisions: Competition, learning, and strategy in Tokyo banking, 1894-1936. Academy of Management Journal, 43(4): 816–836. Griffith, E. 1974. A history of American city government: The progressive years and their aftermath, 1900-1920. New York: Praeger. Grover, V., & Goslar, M. D. 1993. The initiation, adoption, and implementation of telecommunications technologies in U.S. organizations. Journal of Management Information Systems, 10(1): 141–163. Gurbaxani, V., & Whang, S. 1991. The impact of information systems on organizations and markets. Communications of the ACM, 32(1): 59–73. Hair, J. R., Black, W. C., Babin, B. J., & Anderson, R. E. 2010. Multivariate data analysis: A global perspective (7th ed.). Pearson Prentice Hall. Hargadon, A. B., & Douglas, J. Y. 2001. When innovations meet institutions: Edison and the design of the electric light. Administrative Science Quarterly, 46: 476–501. Hart, P. J., & Saunders, C. S. 1998. Emerging electronic partnerships: Antecedents and dimensions of EDI use from the supplier's perspective. Journal of Management Information Systems, 14(4): 87–111. 263
Hayes, A. F., & Matthes, J. 2009. Computational procedures for probing interactions in OLS and logistic regression: SPSS and SAS implementations. Behavior Research Methods, 41(3): 924–936. Hellriegel, D., & Slocum, J. 2007. Organizational behavior (11th ed.). New York, NY. Higgins, P. 2005. Performance and User satisfaction indicators in local government: Lessons from a case study. Public Management Review, 7(3): 445–446. Hite, R. 2005. The state of enterprise architecture maturity in the US federal government. Presentation to CIO Conference. Hobbs, A. 2006. Establishing a successful enterprise architecture program office. Retrieved from www.pmibaltimore.org – accessed 11/09/09. Hofstede, G. 1980. Cultures consequences: International differences in work-related values. Beverly Hills, CA: Sage. Hofstede, G. 1983. The cultural relativity of organizational practices and theories. Journal of International Business Studies, Fall: 75–92. Hoque, Z. & Hopper, T. 1994. Rationality, accounting and politics: A case study of management control in a Bangladeshi jute mill. Management Accounting Research, 5(5): 5–30. Hughes, E. C. 1939. Institutions. In R. E. Park (Ed.), An outline of the principles of sociology: 283–346. New York. Jick, T. D. 1979. Mixing qualitative and quantitative methods: Triangulation in action. Administrative Science Quarterly, 24(4): 602–611. Kappelman L. 2010. The SIM guide to enterprise architecture. CRC Press. Kappelman, L. 2009. The SIM guide to enterprise architecture: Creating the information age enterprise. New York: Auerbach. Kappelman, L., McGinnis, T., Petit, A., Salmans, B., & Sidorova A. 2008. Enterprise architecture: Charting the territory for academic research. AMCIS 2008 Proceedings. Kanter, R. M. 1979. Power failure in management circuits. Harvard Business Review, July-August: 65–75. Katz, M. L., & Shapiro, C. 1986. Technology adoption in the presence of network externalities. Journal of Political Economy, 94(4): 822–841. 264
Kaul, I. 2006. The new public finance: Responding to global challenges. Oxford University Press. Kaul, M. 1987. National policies and intervention process. In M. Kaul, N. R. Patel, & K. Shams (Eds.), Searching for a paddle: Trends in IT applications in Asian government systems, Vol. 1: 1–23. Asia/Pacific Development Center, Kuala Lumpur, Malaysia. Keenan, B., & Mauch, M. 1986. Questionnaires & structured interviews. GAO Training institute. Kennedy, M. T., & Fiss, P. C. 2009. Institutionalization, framing, and diffusion: The logic of TQM adoption and implementation decisions among U. S. Hospitals. Academy of Management Journal, 52(5): 897–918. Kim, Y. 2004. Organizational structures by types of federal agencies: a policy approach to bureaucracy. Paper presented at the annual meeting of The Midwest Political Science Association, Palmer House Hilton, Chicago, IL. Accessed on 2009-05-26 from http://www.allacademic.com/meta/p83870_index.html Kimberly, A. E. 1998. The sanctions glass: Half full or completely empty? International Security, 23: 50–65. Kimberly, J. R., & Evanisko, M. 1981. Organizational innovation: The influence of individual, organizational, and contextual factors on hospital adoption of technological and administrative innovation. Academy of Management Journal, 24(4): 689–713. King, J. L., Gurbaxani, V., Kraemer, K. L., McFarlan, F. W., Raman, K. S., & Yap, C. S. 1994. Institutional factors in information technology innovation. Information Systems Research, 5(2): 139–169. Knight, K. 1967. A descriptive model ofthe intra-firm innovation process. Journal of Business, 40: 485–495. Kotter, J. P., & Heskett, J. L. 1992. Corporate culture and performance. New York: Free Press. Kotter, J. 1996. Leading change. Harvard Business Press. Kraemer, K. L., Dedrick, J., Melville, N., & Zhu, K. 2006. Global e-commerce: Impacts of national environments and policy. Cambridge, UK: Cambridge University Press.
265
Kraemer, K. L., S. Dickhoven, J. L. King, & S. F. Tiemey, 1987. Datawars: The politics of computer modeling in federal policy making. New York: Columbia University Press. Kroeber, A. & Kluckhohn, C. 1952. Culture. New York: Meridian Books. Kwon, T. K., & Zmud, R. W. 1987. Unifying the fragmented models. In J. Boland & R. A. Hirschheim (Eds.), Critical issues in information systems research: 227–251. New York: Chichester [Sussex]. Lawrence, P., & Lorsch, J. 1967. Differentiation and integration in complex organizations. Administrative Science Quarterly, 12: 1–30. Leonard-Barton, D., & Deschamps, I. 1988b. Managerial influence in the implementation of new technology. Management Science, 34(10): 1252–1265. Levitt, B., & March, J. G. 1988. Organizational learning. Annual Review of Sociology, 14: 319–340. Liang, H., Saraf, N., Hu, Q., & Xue, Y. 2007. Assimilation of enterprise systems: The effect of institutional compliances and the mediating role of top management. MIS Quarterly, 31(1): 59–87. Lincoln, Y., & Guba, E. 1985. Naturalistic inquiry. New York: Sage. Litwinenko, A., & Cooper, C. L. 1994. The impact of trust status on corporate culture. Journal of Management in Medicine, 8(4): 8–17. Lohmöller, J.-B. 1989. Latent variable path modeling with partial least squares. Heidelberg, Germany: Physica-Verlag. Long, J. S. 1997. Regression models for categorical and limited dependent variables. Los Angeles: Sage Publications. Luo, A., Huang, L., & Luo, X. 2006. Research on activity-centric architecture methodology. Systems Engineering and Electronics, 30(3): 449. Lusthaus, C., Anderson, G., & Murphy, E. 1995. Institutional assessment: A framework for strengthening organizational capacity for IDRC’s research. Canada: International Development Research Centre. Lyytinen K., & Newman M. 2008. Explaining information system change: A punctuated socio-technical change model. European Journal of Information Systems, 17(6): 589–613. 266
Lyytinen, K., & Damsgaard, J. 2011. Inter-organizational information systems adoption a configuration analysis approach. EJIS, 20(5): 496–509. Lyytinen, K., & Robey, D. 1999. Learning failure in information system development. Information Systems Journal, 9(2): 85–101. Makiya G. K., & Lyytinen, K. 2011. The antecedents of enterprise architecture assimilation in organizations. Unpublished Quantitative Research Report, Weatherhead School of Management, Case Western Reserve University. Massetti, B., & Zmud, R.W. 1996. Measuring the extent of EDI usage in complex organizations: Strategies and illustrative examples. MIS Quarterly, 20(3): 331– 345. Maxwell, J. A. 2005. Qualitative research design: An interactive approach. Thousand Oaks, CA: Sage. McGuire, P., Granovetter, M., & Schwartz, M. 1993. Thomas Edison and the social construction of the early electricity industry in America. In R. Swedberg (Ed.), Explorations in economic sociology. New York: Russell Sage Foundation. McKay, G. 1996. Senseless acts of beauty: Cultures of resistance since the sixties. London: Verso. McKinley, W., Mone, M., & Moon, G. 1999. Determinants and development of schools in organization theory. Academy of Management Review, 24: 634–648. Meyer, J. W., & Rowan, B. 1977. Institutionalized organizations: Formal structure as myth and ceremony. American Journal of Sociology, 83(2): 340–363. Meyer, J. W., & Rowan, B. 1975. Notes on the structure of educational organizations. Paper presented at annual meeting of the American Sociological Association, San Francisco. Meyer, J. W., & Scott, W. R. 1983. Organizational environments: Ritual and rationality. Beverly Hills, CA: Sage. Meyer, A. D., & Goes, J. B. 1988. Organizational assimilation of innovations: A multilevel contextual analysis. Academy of Management Journal, 31(4): 897– 923. Meyerson, D. E., & Scully, M. A. 1995. Tempered radicalism and the politics of ambivalence and change. Organization Science, 6(5): 585–600. Meyerson, D. 2003. Tempered radicals: How everyday leaders inspire change at work. Harvard Business School Press. 267
Mezias, S. 1990. Regulatory capture, interest group theory, and institutional mediation: The regulatory politics of financial reporting rules, 1973–1987. Available at http://w4.stern.nyu.edu/emplibrary/Mezias.Stephen.pdf Miller, J. P. 2000. Grainger says it will miss estimates after installing complex ERP software. Wall Street Journal, January 10: C.12. Mizruchi, M. S., & Stearns, L. B. 1994a. A longitudinal study of borrowing by large American corporations. Administrative Science Quarterly, 39: 118–140. Molnar, J. J., & Rogers, D. L. 1976. Organizational effectiveness: An empirical comparision of the goal and system resource approaches. The Sociological Quarterly, 17(3): 401–413. Moon, M. J., & Bretschneider, S. I. 2002. Does the perception of red tape constrain IT innovativeness in organizations? Unexpected results from a simultaneous equation model and implications. Journal of Public Administration Research and Theory, 12: 273–291. Mustonen-Ollila, E., & Lyytinen, K. 2004. How organisations adopt IS process innovations: A longitudinal analysis. European Journal of Information Systems, 13: 35–51. Nelson, E. 1989. The British counter-culture, 1966-73: A study of the underground press. London: Macmillan. Newfield, Jack. Nickles, M., & Weiss, G. 2004. A framework for the social description of resources in open environments. Lecture Notes in Computer Science, 2782: 206–221. Nilakanta, S., & Scamell, R. W. 1990. The effect of information sources and Communication channels on the diffusion of innovation in a data base development environment. Management Science, 36(1): 24–40. Nunn, S. 2007. The open group and enterprise architecture. Admag. Orlikowski, W. J. 1993. CASE tools as organizational change: Investigating incremental and radical changes in systems development. MIS Quarterly, 17(3): 309–340. Orlikowski, W. J., & Hofman, J. D. 1997. An improvisational model for change management: The case of groupware technologies. Sloan Management Review, 38(2): 11–21. Park, S. H., & Luo, Y. 2001. Guanxi and organizational dynamics: Organizational networking in Chinese firms. Strategic Management Journal, 22: 455–477. 268
Parsons, T. 1951. The social system. New York: Free Press. Petter, S., Straub, D. W., & Rai, A. 2007. Specifying formative constructs in information systems research. MIS Quarterly, 31(4): 623–656. Penttinen, E., & Tuunainen, V. 2009. Assessing the effect of external pressure in interorganizational IS adoption - Case electronic invoicing. Lecture Notes in Business Information Processing, 52(2): 269–278. Pfeffer, J., & Salancik, G. R. 1978. The external control of organizations: A resource dependence perspective. New York: Harper and Row. Powell, W., & DiMaggio P. W. (Eds.). 1991. The new institutionalism in organizational analysis. Chicago: The University of Chicago Press Prahalad, C. K. 2004. Fortune at the bottom of the pyramid: Eradicating Poverty through profits. Pearson Prentice Hall. Purvis, R. L., Sambamurthy, V., & Zmud, R. W. 2001. The assimilation of knowledge platforms in organizations: An empirical investigation. Organization Science, 12(2): 117–135. Rico, D. F. 2006. A framework for measuring the ROI of enterprise architecture. Journal of Organizational and End User Computing, 18(2): i–xii. Robbins, S. P. 2001. Organizational behavior. Prentice Hall. Robertson, T. S., & Gatignon, H. 1986. Competitive effects on technology diffusion. Journal of Marketing, 50(3): 1–12. Robey D., Bourdreau, M., & Rose, G. M. 2000. Information technology and organizational learning: a review and assessment of research. Accounting, Management and Information Technologies, 10(2): 125–155. Rogers, E. M. 1995. Diffusion of innovations. New York: The Free Press. Rogers, E. M. 1983. Diffusion of innovations (3rd ed.). New York: Free Press. Ross, J., Weill, P., & Robertson, D. C. 2006. Enterprise architecture as strategy. Harvard Business School Press. Roszak, T. 1968. The dissenting academy. New York: Pantheon. Roszak, T. 1968. The making of a counter culture. University of California Press. 269
Rowan, B. 1982. Organizational structure and the institutional environment: The case of public schools. Administrative Science Quarterly, 27: 259–279. Russell, R. F., & Stone, A. G. 2002. A review of servant leadership attributes: Developing a practical model. Leadership & Organization Development Journal, 23: 14 Sackmann, S. 1991. Cultural knowledge in organizations: Exploring the collective mind. London: Sage. Saga, V. L., & Zmud, R. 1994. The nature and determinants of IT acceptance, routinization, and infusion. Diffusion, Transfer and Implementation of Information Technology, 45(A-45): 67–86. Saha, P. 2004. Analyzing the open group architecture framework (TOGAF) from the GERAM Perspective. from The Open Group Website: http://www.opengroup.org/architecture/wp Saha, P. 2006. A real options perspective to enterprise architecture as an investment activity. Journal of Enterprise Architecture, November. Salmans, B., Kappelman, L., & Pavur, R. 2009. Organization size, IT capabilities, and EA perceptions: Dark clouds on the ERP horizon? AMCIS 2009 Proceedings. Paper 747. Sambamurthy, V., & Zmud, R. W. 1996. Information technology and innovation: Strategies for success. Morristown, NJ: Financial Executives Research Foundation. Sambamurthy, V., & Chin, W. W. 1994. The effects of group attitudes toward alternative GDSS designs on the decision-making performance of computer-supported groups. Decision Sciences, 25(2): 215–241 Schein, E. H. 1992. Organizational culture and leadership. San Francisco: Jossey-Bass. Schekkerman, J. 2006. Extended enterprise architecture maturity model (E2AMM). Institute for Public Enterprise Architecture Development, http://www.enterprisearchitecture.info. Schekkerman, J. 2005. How to survive the jungle of enterprise architecture: Creating or choosing an enterprise architecture framework. IFEAD. Schekkerman, J. 2005. Trends in enterprise architecture. Menedeley.
270
Schuman, H., & Presser, S. 1981. Questions and answers in attitude surveys. New York: Academic Press. Scott, W. R. 1992. Organizations: Rational, natural, and open systems (3rd ed.). Englewood Cliffs: Prentice Hall. Sedera, D., & Zakaria, N. I. 2009. Managing knowledge for enterprise systems in Indian organizations: Case insights. In proceedings of the Pacific Asian Conference on Information Systems (PACIS '09), July 06-09 India. Senior, B., & Fleming, J. 2006. Organization change. Prentice Hall. Sessions, R. 2007. A comparison of the top four enterprise-architecture methodologies. Architecture, 466232: 1–28. Sethi, V., & King, W. R. 1994. Development of measures to assess the extent to which an information technology application provides competitive advantage. Management Science, 40(12): 1601–1627. Snow, C. C., & Hrebiniak, L. G. 1980. Strategy, distinctive competence, and organizational performance. Administrative Science Quarterly, 25: 317–335. Strassmann, P. 2005. The structure of IT spending as measure of organizational disorder, presentation at Stevens Institute of Technology. The Structure of IT Spending as a Measure of Organizational Disorder. Lecture, Stevens Institute of Technology (slides), May 19, 2005. Straub, D., Boudreau, M.-C., & Gefen, D. 2004. Validation guidelines for IS positivist research. Communications of the AIS, 13(24): 380–427. Straub, D. W., & Burton-Jones, A. 2007. Veni, vidi, vici: Breaking the TAM logjam. Journal of the AIS, 8(4): 223–229. Strauss, A., & Corbin, J. 1990. Basics of qualitative research: Grounded theory procedures and techniques. Sage. Strauss, A. L., & Corbin, J. M. 1998. Basics of qualitative research: Techniques and procedures for developing grounded theory. Sage Publications Inc. Staw, B. M., & Epstein, L.D. 2000. What bandwagons bring: Effects of popular management techniques on corporate performance, reputation, and CEO pay. Administrative Science Quarterly, 45(3): 523–556. Suchman, M. C. 1995. Managing legitimacy: Strategic and institutional approaches. Academy of Management Review, 20(3): 571–610. 271
Swanson, E. B., & Ramiller, N. C. 2003. Organizing visions for information technology and the information systems executive response. Journal of Management Information Systems, 20(1): 13–50. Swanson, E. B., & Ramiller, N. C. 2004. Innovating mindfully with information technology. MIS Quarterly, 28(4): 553–583. Swanson, E. B., & Ramiller, N. C. 1997. The organizing vision in information systems innovation. Organization Science, 8(5): 458–474. Tallon, P. P., Kraemer, K. L., & Gurbaxani, V. 2001. Executives’ perceptions of the business value of information technology: a process-oriented approach. Working Paper #ITR-148. University of California, Irvine, CA, USA. Teece, D. 1980. Vertical Integration and Technological Innovation” (with Henry Armour). Review of Economics and Statistics, 62(3): 470–474. Teo, H. H., Wei, K. K., & Benbasat, I. 2003. Predicting intention to adopt interorganizational linkages: An institutional perspective. MIS Quarterly, 27(1): 19–49. The Chief Information Officers Council. 1999. Federal Enterprise Architecture Framework Version 1.1. September 1999. Thelen, D. P. 1972. The new citizenship origins of progressivism in Wisconsin, 18851900. Columbia, MO: University of Missouri Press. The Open Group. 1995. The Open Group Architecture Framework. An Open Group Standard. Tolbert P. S., & Zucker, L. G. 1983. Institutional sources of change in the formal structure of organizations: The diffusion of civil service reform, 1880-1935. ASQ, 28: 22–39. Tolbert, P. S. 1985. Institutional environments and resource dependence: Sources of administrative structure in institutions of higher education. Administrative Science Quarterly, 30(1): 1–13. Tornatzky, L.G., & Fleischer, M. 1990. The processes of technological innovation. Lexington. Tornatzky, L. G., & Klein, K. J. 1982. Innovation characteristics and innovation adoption-implementation: A meta-analysis of findings. IEEE Transactions on Engineering Management (EM-29:1): 28–45. 272
Tversky, A., & Kahneman, D. 1981. The framing of decisions and the psychology of choice. Science, 211: 453–458. Vinzi, V., Chin, W. W., Henseler, J., & Wang, H. (Eds.). 2010. Handbook of partial least squares: Concepts, methods and applications. Springer. Wang, P. 2008. Assimilating IT innovation: The longitudinal effects of institutionalization and resource dependence. ICIS 2008 Proceedings. Webb, B., & Young, S. 2007. Enterprise architecture. TBTLA Quarterly Event November 2007. Powerpoint presentation. Weber, M. 1922. Bureaucracy. In J.M Shafritz, A.C. Hyde & S.J. Pa (Eds.), Classics of public administration: 50–55. Weick, K. 1969/1979. The social psychology of organizing. Reading, MA: AddisonWesley Weick, K. E. 1990. Technology as an equivoque: Sensemaking in new technologies. In P. Goodman & L. Sproull (Eds.), Technology and organizations. San Fransisco: Josey Bass. Weill, P., & Ross, J. W. 2003. IT governance – How top performers manage IT decision rights for superior results. Boston: Harvard Business School Press. Weiss, R. S. 1994. Learning from strangers: The art and method of qualitative interview studies. New York: The Free Press Wiebe, R. 1967. The search for order, 1877-1920. New York: Hill. Wisnosky, D. E., & Vogel, J. 2004. DoDAF Wizdom: A practical guide to planning, managing, and executing projects to build enterprise architectures using the department of defense architecture framework (DoDAF). Wold, H. 1982. Soft modeling: the basic design and some extensions. In K. G. Jöreskog & H. Wold (Eds.), Systems under indirect observations, Vol. 2: 1–54. Amsterdam: North-Holland. Wold, H. 1988. Specification, predictor. In S. Kotz & N. L. Johnson (Eds.), Encyclopedia of statistical sciences, vol. 8: 587–599. New York, NY: Wiley. Wong, S. 2001. Electronic health records implementation: Lessons we can learn from the pioneers. School of Health Information Science, August 16.
273
Wright, R. E. 1995. Logistic regression. In L. G. Grimm, & P. R. Yarnold (Eds.), Reading and understanding multivariate statistics: 217–244. Washington, DC: American Psychological Association. Zhu, K., Kraemer, K. L., & Xu, S. 2006. The process of innovation assimilation by firms in different countries: A technology diffusion perspective on e-business. Management Science, 52(10): 1557–1576. Xue, Y., Liang, H., Boulton, W. R., & Snyder, C. A. 2005. ERP implementation failures in China: Case studies with implications for ERP vendors. International Journal of Production Economics, 97(3): 279–295. Zachman, J. A. 1997. Concepts of the framework for enterprise architecture: Background, description and utility. Zachman International. Accessed 19 Jan 2009. Zhu, K. 2004. The complementarity of information technology infrastructure and ecommerce capability: A resource-based assessment of their business value. Journal of Management Information Systems, 21(1): 167–202. Zhu, K., & Kraemer, K. L. 2002. e-Commerce Metrics for net-enhanced organizations: Assessing the value of e-commerce to firm performance in the manufacturing sector. Information Systems Research, 13(3): 275–295. Zhu, K., & Kraemer, K. L. 2005. Post-adoption variations in usage and value of ebusiness by organizations: Cross-country evidence from the retail industry. Information Systems Research, 16(1): 61–84. Zhu, K., Kraemer, K. L., Xu, S. & Dedrick, J. 2004. Information technology payoff in ebusiness environments: An international perspective on value creation of ebusiness in the financial services industry. Journal of Management Information Systems, 21(1): 17–54. Zmud, R. W., & Apple, L. E. 1992. Measuring technology incorporation/infusion. Journal of Product Innovation Management, 9(2): 148–155. Zmud, R. W. 1982. Diffusion of modern software practices: Influence of centralization and formalization. Management Science, 28: 1421–1431.
274
Government Studies and Documents Cohen, W. 1994. Computer chaos: Billions wasted buying federal computer systems. Investigative Report of Senator William Cohen, Ranking Minority Member, Subcommittee on Oversight of Government Management, Senate Government Affairs Committee, Washington D.C. Chief Information Officers Council. 2001. Federal Enterprise Architecture Framework, Version 1.1 (September 1999) and Chief Information Officers Council, A Practical Guide to Federal Enterprise Architecture, Version 1.0 (February 2001). U.S Government Accountability Office. 2009. GAO, Information Technology: Leadership Remains Key to Agencies Making Progress on Enterprise Architecture Efforts, GAO-04-40 (Washington, D.C.: Nov. 17, 2003) accessed Jan 20, 2009 U.S Government Accountability Office. 2006. GAO, Information Technology: Leadership Remains Key to Establishing and Leveraging Architectures for Organizational Transformation, GAO-06-831 (Washington, D.C.: August 20, 2006) accessed Jan 20, 2009 US Government Accountability Office. 2004. Information Technology Investment Management: A Framework for Assessing and Improving Process Maturity. GAO-04-394G. (Washington, D.C.: Mar. 2004). U.S. General Accounting Office. 2003. Information Technology: A Framework for Assessing and Improving Enterprise Architecture Management (Version 1.1), GAO-03-584G (Washington, D.C.: April 2003). U.S. General Accounting Office. 2002. Information Technology: Enterprise Architecture Use across the Federal Government Can Be Improved, GAO-02-6 (Washington, D.C.: Feb. 19, 2002). U.S. Congress - Clinger-Cohen Act of 1996. Public Law 104-106, section 5125, 110 Stat. 684 (1996). 3A Practical Guide to Federal Enterprise Architecture, Version 1.0, Chief Information Officers Council (February 2001). U.S. Office of Management and Budget, Information Technology Architectures, Memorandum M-97-16 (June 18, 1997); rescinded with the update of OMB Circular A-130 (Nov. 28, 2000). 37 OMB, Management of Federal Information Resources, Circular No. A-130 (Nov. 28, 2000).
275
U.S. Office of Management and Budget. Management of Federal Information Resources, Circular A-130 (Nov. 30, 2000), which implements the Clinger-Cohen Act of 1996, Public Law 104-106, section 5125, 110 Stat. 684 (1996), 40 U.S.C. 11315. Accessed Feb. 1, 2009. Comptroller General of the United States. Executive Guide: Improving Mission Performance Through Strategic Information Management and Technology, Learning From Leading Organizations. GAO/AIMD-94-115. U.S. General Accounting Office P.O. Box 6015 Gaithersburg, MD 20877, 1994. U.S General Accounting Office. Information Technology Assessment: A Governmentwide Overview. GAO/AIMD-95-208. Washington, D.C.: GAO, July 1995
276
