organization (low PO fit) are less motivated to lead, values homogeneity in leadership ... The interaction of PO fit and relativism was also examined. An.
Assessing the Relationships Between Person-Organization Fit, Moral Philosophy, and the Motivation to Lead
Dissertation
Submitted to Northcentral University
Graduate Faculty of the School of Psychology in Partial Fulfillment of the Requirements for the Degree of
DOCTOR OF PHILOSOPHY
by ELENA MARIE PAPAVERO Prescott Valley, Arizona January 2009
Copyright Notice
© Copyright 2009 Elena Marie Papavero
Approval
Assessing the Relationships Between Person-Organization Fit, Moral Philosophy, and the Motivation to Lead by Elena Marie Papavero
Approved by:
Chair: William G. Shriner, PhD
Date
Member: Robert Haussmann, PhD
Member: Nadira Tidwell Pardo, PhD
Certified by:
School Chair: Heather Frederick, PhD
Date
Abstract Assessing the Relationships Between Person-Organization Fit, Moral Philosophy, and the Motivation to Lead by Elena Marie Papavero Northcentral University, January 2009
When individuals who perceive their values as different from those of their organization (low PO fit) are less motivated to lead, values homogeneity in leadership may occur, resulting in ethical dysfunction. Likewise, if idealists are less attracted to leading, this may influence homogeneity towards pragmatism. The primary goal of this research was to explore the prediction of three dimensions of motivation to lead (MTL) from PO fit and idealism. The interaction of PO fit and relativism was also examined. An online survey, including Cable and DeRue’s fit measure, Forsyth’s EPQ, and Chan’s MTL scale, was completed by 1,024 working adults. Lower fit predicted lower MTL on all dimensions, and higher idealism predicted lower MTL on all dimensions (with socialnormative MTL receiving limited support). No support was found for relativism as a moderator of the fit to MTL relationship. These results suggest that low fit individuals are self-selecting away from leadership positions. Practical recommendations include considering fit in advancement processes and using fit as a gap-analysis diagnostic for organizational values misalignment. Future research on a situational model of MTL should consider situations that promote involvement or identification with organizations
iv
and objectives, and those that create a lack of alternatives or a sense of obligation due to a psychological contract.
v
Acknowledgments
I would like to express great appreciation to my chairperson, Dr. William Shriner, for his steady guidance and enthusiastic support of my ideas, and to my committee members, Dr. Robert Haussmann and Dr. Nadira Pardo, for their patience and wisdom. I would also like to thank my external reviewer, Dr. Jon Billsberry, for providing inspiration, and for freely sharing his knowledge and insights. The support and camaraderie of my school colleagues, especially Judy Kelly and Brian Cesario, made a world of difference, not only instrumentally, but also in the inspiration provided by their demonstrations of intellectual curiosity, will, and spirit. Finally, I would like to acknowledge all family, friends, and work colleagues who supported this effort, with special thanks to Marc Saxton for his interest and faith in my work, and for his special talent for listening.
vi
Table of Contents Copyright Notice ................................................................................................................. ii Approval ............................................................................................................................ iii Abstract .............................................................................................................................. iv Acknowledgments.............................................................................................................. vi List of Tables ...................................................................................................................... x List of Figures ................................................................................................................... xii Chapter 1: Introduction ....................................................................................................... 1 Background ..................................................................................................................... 3 Problem Statement .......................................................................................................... 4 Purpose of the Study ....................................................................................................... 5 Conceptual Framework ................................................................................................... 5 Research Questions ......................................................................................................... 8 Hypotheses ...................................................................................................................... 9 Hypotheses for PO fit and motivation to lead. ........................................................ 9 Hypotheses for PO fit, relativism, and motivation to lead. ................................... 12 Hypotheses for idealism and motivation to lead. .................................................. 14 Definition of Terms....................................................................................................... 16 Limitations .................................................................................................................... 18 Summary and Conclusions ........................................................................................... 19 Chapter 2: Review of the Literature.................................................................................. 21 Introduction ................................................................................................................... 21 Conceptualizing Person-Organization Fit ..................................................................... 21 The person-organization fit focus. ........................................................................ 21 Choosing a PO fit interaction type. ....................................................................... 22 Operationalizing PO fit with values. ..................................................................... 27 Choosing a view of PO fit. .................................................................................... 28 PO misfit. .............................................................................................................. 32 Conceptualizing PO fit in the present study. ........................................................ 35 PO Fit and the Motivation to Lead ............................................................................... 37 vii
Individual outcomes of PO fit. .............................................................................. 37 Organizational outcomes of PO fit. ...................................................................... 39 How PO fit changes. ............................................................................................. 42 PO fit, individual characteristics, and situation. ................................................... 44 Ethical fit. .............................................................................................................. 45 Moral Philosophy and the Motivation to Lead ............................................................. 48 PO fit and moral philosophy. ................................................................................ 48 Moral philosophy overview. ................................................................................. 48 PO fit, ethical conflict, and relativism. ................................................................. 52 Ethical conflict, relativism, and the motivation to lead. ....................................... 54 Idealism and the motivation to lead. ..................................................................... 61 Conceptualizing and Studying Motivation to Lead ...................................................... 63 Theoretical model of the motivation to lead. ........................................................ 64 The motivation to lead construct........................................................................... 65 Relevant studies using Chan’s motivation to lead construct. ............................... 68 Studies on motivation to lead and situation. ......................................................... 70 Motivation to Lead Antecedents ................................................................................... 73 Personality trait antecedents and situation. ........................................................... 73 Values antecedents and situation. ......................................................................... 76 Leadership antecedents and situation.................................................................... 78 Summary of motivation to lead antecedents and situation. .................................. 78 Summary of Literature Review ..................................................................................... 78 Chapter 3: Methodology ................................................................................................... 80 Overview ....................................................................................................................... 80 Restatement of Hypotheses ........................................................................................... 80 Research Design............................................................................................................ 82 Operational Definition of Variables.............................................................................. 82 Instrumentation ............................................................................................................. 83 Sampling ....................................................................................................................... 85 A priori power calculations. .................................................................................. 85 Selection of participants. ....................................................................................... 89 Procedures ..................................................................................................................... 90 Data Analysis ................................................................................................................ 91 Methodological Assumptions, Limitations, and Delimitations .................................... 92 Ethical Assurances ........................................................................................................ 93 Summary ....................................................................................................................... 94 Chapter 4: Findings ........................................................................................................... 95
viii
Overview ....................................................................................................................... 95 Data Preparation............................................................................................................ 95 Sample Description ....................................................................................................... 96 Common Method Variance ........................................................................................... 97 Nonresponse Bias.......................................................................................................... 97 Descriptive Statistics ................................................................................................... 100 Tests of Statistical Assumptions ................................................................................. 101 Hypothesis Testing Procedure .................................................................................... 103 Hypothesis Testing...................................................................................................... 105 Hypothesis testing for PO fit and motivation to lead. ......................................... 105 Hypothesis testing for PO fit, relativism, and motivation to lead. ...................... 108 Hypothesis testing for idealism and motivation to lead. ..................................... 110 Summary of Findings .................................................................................................. 112 Supplemental Analysis................................................................................................ 114 Chapter 5: Discussion ..................................................................................................... 117 Conclusions for PO Fit and Motivation to Lead ......................................................... 118 Conclusions for PO Fit, Relativism, and Motivation to Lead..................................... 120 Conclusions for Idealism and Motivation to Lead ...................................................... 121 Practical Implications.................................................................................................. 123 Study Limitations ........................................................................................................ 126 Recommendations for Future Motivation to Lead Research ...................................... 127 Recommendations for Future PO Fit Research .......................................................... 128 Recommendations for Future Moral Philosophy Research ........................................ 131 Recommendations for Future Group Differences Research ....................................... 133 Epilogue ...................................................................................................................... 134 References ....................................................................................................................... 138 Appendix A: Scale Items ................................................................................................ 158 Appendix B: Request for Participation ........................................................................... 161 Appendix C: Informed Consent, Survey, and Debriefing............................................... 162 Appendix D: Additional Statistical Tables and Figures ................................................. 171
ix
List of Tables Table 1. Hierarchical Multiple Regression: A priori Power Calculation ........................ 86 Table 2. Moderated Multiple Regression: A priori Power Calculation Not Considering Coefficient Differences .............................................................................................. 87 Table 3. Moderated Multiple Regression: A priori Power Calculation Considering Coefficient Differences .............................................................................................. 88 Table 4. Personal Characteristics .................................................................................... 98 Table 5. Job and Organization Characteristics................................................................ 99 Table 6. Coefficient Alphas, Correlations, Means, and Standard Deviations for Study Variables ................................................................................................................. 100 Table 7. Correlations of Characteristics and Study Variables ....................................... 115 Table 8. One-Way Analyses of Variance for Ethnicity on Study Variables .................... 116 Table 9. Means and Standard Deviations for Study Variables and Age ........................ 171 Table 10. Means and Standard Deviations for Study Variables and Gender ................. 171 Table 11. Means and Standard Deviations for Study Variables and Educational Level 172 Table 12. Means and Standard Deviations for Study Variables and Work Experience . 172 Table 13. Means and Standard Deviations for Study Variables and Leadership Experience............................................................................................................... 173 Table 14. Means and Standard Deviations for Study Variables and Ethnicity .............. 173 Table 15. Means and Standard Deviations for Study Variables and Job Tenure........... 174 Table 16. Means and Standard Deviations for Study Variables and Job Level ............. 174 Table 17. Means and Standard Deviations for Study Variables and Employment Status ................................................................................................................................. 175 Table 18. Means and Standard Deviations for Study Variables and Organization Size 175 Table 19. Means and Standard Deviations for Study Variables and Organization Tenure ................................................................................................................................. 176 Table 20. Summary of Hierarchical Regression Analysis for PO Fit Predicting General Motivation to Lead .................................................................................................. 184 Table 21. Summary of Hierarchical Regression Analysis for PO Fit Predicting AffectiveIdentity Motivation to Lead ..................................................................................... 185 Table 22. Summary of Hierarchical Regression Analysis for PO Fit Predicting NonCalculative Motivation to Lead .............................................................................. 186 Table 23. Summary of Hierarchical Regression Analysis for PO Fit Predicting SocialNormative Motivation to Lead ................................................................................ 188
x
Table 24. Summary of Moderated Regression Analysis for PO Fit and Relativism Interacting to Predict General Motivation to Lead ................................................ 189 Table 25. Summary of Moderated Regression Analysis for PO Fit and Relativism Interacting to Predict Affective-Identity Motivation to Lead .................................. 190 Table 26. Summary of Moderated Regression Analysis for PO Fit and Relativism Interacting to Predict Non-Calculative Motivation to Lead ................................... 191 Table 27. Summary of Moderated Regression Analysis for PO Fit and Relativism Interacting to Predict Social-Normative Motivation to Lead ................................. 192 Table 28. Summary of Hierarchical Regression Analysis for Idealism Predicting General Motivation to Lead .................................................................................................. 193 Table 29. Summary of Moderated Regression Analysis for Idealism Predicting AffectiveIdentity Motivation to Lead ..................................................................................... 194 Table 30. Summary of Hierarchical Regression Analysis for Idealism Predicting NonCalculative Motivation to Lead .............................................................................. 195 Table 31. Summary of Hierarchical Regression Analysis for Idealism Predicting SocialNormative Motivation to Lead ................................................................................ 195
xi
List of Figures Figure 1. Initial model: Person-organization fit predicting motivation to lead, and idealism predicting motivation to lead. .................................................................... 10 Figure 2. Initial Model: Person-organization fit, as moderated by relativism, predicting motivation to lead. .................................................................................................... 10 Figure 3. Revised model: Person-organization fit predicting motivation to lead, and idealism predicting motivation to lead. .................................................................. 113 Figure 4. Histogram of participants' reported person-organization fit scores with normality curve superimposed. ............................................................................... 177 Figure 5. Normal probability of participants’ reported person-organization fit scores. 177 Figure 6. Histogram of participants' reported idealism scores with normality curve superimposed. ......................................................................................................... 178 Figure 7. Normal probability of participants’ reported idealism scores. ...................... 178 Figure 8. Histogram of participants' reported relativism scores with normality curve superimposed. ......................................................................................................... 179 Figure 9. Normal probability of participants’ reported relativism scores. .................... 179 Figure 10. Histogram of participants' reported general motivation to lead scores with normality curve superimposed. ............................................................................... 180 Figure 11. Normal probability of participants’ reported general motivation to lead scores. ..................................................................................................................... 180 Figure 12. Histogram of participants' reported affective-identity motivation to lead scores with normality curve superimposed. ............................................................ 181 Figure 13. Normal probability of participants’ reported affective-identity motivation to lead scores. ............................................................................................................. 181 Figure 14. Histogram of participants' reported non-calculative motivation to lead scores with normality curve superimposed. ....................................................................... 182 Figure 15. Normal probability of participants’ reported non-calculative motivation to lead scores. ............................................................................................................. 182 Figure 16. Histogram of participants' reported social-normative motivation to lead scores with normality curve superimposed. ............................................................ 183 Figure 17. Normal probability of participants’ reported social-normative motivation to lead scores. ............................................................................................................. 183
xii
1 Chapter 1: Introduction Organizations are thought to become more homogenous over time through attraction, selection, attrition, and individual socialization (Schaubroeck, Ganster, & Jones, 1998; Schneider, 1987; Schneider, Smith, Taylor, & Fleenor, 1998). Individuals who fit the organization are more likely to be attracted to it, and more likely to be selected. Those who do not fit tend to leave, although some who do not fit experience a socialization process that increases fit. In this way, the organization is a function of the persons behaving in it, rather than the person and environment producing behavior (Schneider). The definition of fit in this context initially focused on the similarity of values between person and organization, also known as values congruence (Chatman, 1989, 1991; O’Reilly, Chatman, & Caldwell, 1991), but has since been extended to encompass types of fit that consider how the person and organization complement one another on a variety of characteristics. The idea of fit has also been expanded to include interactions between persons and their jobs, supervisors, peers, and groups, in addition to the original concept of fit between the person and organization. Person-organization fit (PO fit) at the individual level is seen as positive, producing higher commitment, job satisfaction, and lower intention to leave (Davis, 2006; Westerman & Cyr, 2004). However, the desirability of organizational homogeneity has been questioned (Atwater & Dionne, 2007; Boone, Olffen, Witteloostuijn, & Brabander, 2004; Giberson, Resick, & Dickson, 2005). In general, it is thought that homogenous organizations have more difficulty changing in response to increasingly dynamic external environments. Further, Judge (2008) recently questioned the ethicality
2 of homogenizing around fit, as selecting and socializing for attitudes could be considered an invasion of individual privacy and an abuse of power. One area where homogeneity may not be desirable is organizational values. Deep implicit socialization can lead to values homogeneity in an organization’s leadership through the promotion process, the opportunity structure, and implicit leader emergence schemas (Hackman & Wageman, 2007; Rayburn & Rayburn, 1996; Snell, 2000). Snell proposed that this in turn leads to ethical dysfunction. Snell further suggested that ethical dysfunction might be necessary for organizational survival. However this has not been proven and may not always be the case. Ethics is becoming an increasingly important business issue as business problems, and the moral dilemmas they produce, become more complex (Bennis, 2007; Nicholson, 1994). Although it is not yet known, it is possible that an organization that is heterogeneous on values might be better equipped to address complex moral dilemmas and shape the organization’s ethical standards to reduce ethical dysfunction. Recent scandals at large corporations give evidence that this question is worthy of exploration (Ghoshal, 2005). It is possible that low PO fit at the individual level increases values homogeneity at the organizational level by diminishing motivation to lead. Therefore, the present study explored the prediction of motivation to lead from PO fit. The motivation to lead construct used in the present study is based on commitment (Chan, 1999). Further, an individual difference, known as moral philosophy, predicts commitment (Peterson, 2003; Shaub, Finn, & Munter, 1993). An individual’s moral philosophy provides guidelines used to solve ethical dilemmas. Paralleling previous findings for commitment, moral
3 philosophy was explored as an individual difference that could predict motivation to lead, and also as a moderator of the relationship between PO fit and motivation to lead. Background As mentioned previously, several theories propose that, over time, employees with low PO fit leave the organization (for example, Schneider’s [1987] attractionselection-attrition or ASA theory, and Ponemon’s [1992] studies of the effect of selection-socialization on the ethics of auditors). However, it is possible that employees with low PO fit join organizations despite their lack of fit (Chatman, Wong, & Joyce, 2008) and remain with organizations for a variety of reasons related to embeddedness (Harman, Lee, Mitchell, Felps, & Owens, 2007) or a perceived, or actual, tight marketplace (Stern, 2003). Further, they may remain with the organization and function in informal and possibly less influential leadership roles, such as internal networkers (Senge, 1996) or tempered radicals (Meyerson & Scully, 1995). In practice, identifying these employees and encouraging their participation makes a contribution to the organization that includes increased diversity of values. For theory extension, identifying situational factors that predict motivation to lead is a first step in building a situational model. This study also makes other contributions to leadership theory. Another researcher has proposed an evolutionary psychology theory that those with poor PO fit might self-select away from leading (Nicholson, 2005). This self-selection away from leading may be explained, in part, by low motivation to lead. In addition, Ashforth and Anand (2003) proposed that those with low PO fit might be systematically excluded from leadership positions. If this proposition were to be tested, the present study suggests a
4 way to determine if self-selection away from leading is occurring rather than, or in addition to, exclusion. The study of individuals who avoid leadership roles contributes to a neglected area in organizational theory. In general, low PO fit on values means the organizational system is not aligned. Vital information about an organization’s operating values versus espoused values could be tapped in a positive way by identifying and studying individuals who self-select away from leading (Papavero, 1999). This could result in better integration of diverse values within the organization. Problem Statement Powell (1998) suggests that diverse values at higher levels of the organization are more important to extending existing organizational values versus reinforcing them. Further, Bretz, Ash, and Dreher (1989) suggest that values homogeneity increases with organizational level. Therefore, one way to increase the values heterogeneity of an organization would be to diversify the values of those in leadership roles, allowing the ethics of the current culture to be tested (Thorne & Saunders, 2002). This could be done by attracting, developing, supporting, and rewarding leaders whose values differ from the existing organizational culture. However, it is possible that those whose values do not fit the existing culture are less motivated to move into leadership roles (Papavero, 1999). To determine if PO fit predicts motivation to lead, a motivation to lead theory and construct are necessary. Chan’s (1999) motivation to lead framework offers both. Chan’s motivation to lead theory seeks to explain why individuals choose to lead. His model uses individual differences to predict motivation to lead. However, the model does not account for situational aspects, such as PO fit, which may influence motivation to lead. Chan has
5 called for an exploration of situational variables that may influence motivation to lead to give direction in creating a situational model. Other individual attributes may contribute to motivation to lead. Past research has found that moral philosophy predicts commitment (Peterson, 2003; Shaub et al., 1993), which is an attitude on which Chan’s (1999) motivation to lead construct is based. In addition, Papavero (1999) found that individuals who rejected promotions exhibited qualities associated with idealism. These include commitment to profession, conscientiousness, and intrinsic motivation (Bierly, Kolodinsky, & Charette, in press; Forsyth, 1992; Shafer, Park, & Liao, 2002b). As the motivation to lead construct is relatively new, the present research considered whether these findings also apply to motivation to lead. Purpose of the Study In a previous qualitative study of six software engineers, Papavero (1999) uncovered various factors that contributed to the rejection of advancement offers, most of which were value laden. The present quantitative study was designed to corroborate and generalize these findings by examining the relationships between measured perceived values similarity (also known as perceived supplementary person-organization fit) and motivation to lead with a larger and more diverse sample. Conceptual Framework Chan’s (1999) motivation to lead construct, which is based on Meyer and Allen’s (1991) model of organizational commitment, has three dimensions: (a) affective-identity motivation to lead (liking to lead), (b) non-calculative motivation to lead (making a rational decision to lead), and (c) social-normative motivation to lead (feeling a duty to
6 lead). A person low in non-calculative motivation to lead would lead only if they see a net benefit. They consider all types of costs, including non-economic ones. Someone higher in the non-calculative dimension would lead even if there were no net benefit, and they disregard the costs (although it should be noted that they may not be aware of the costs). Chan (1999) identified four major factors that predict motivation to lead: (a) personality traits, (b) values, (c) leadership self-efficacy, and (d) previous leadership experiences, all of which interact with the environment (Amit, Lisak, Popper, & Gal, 2007). Chan’s premise is that even though the potential to lead is present, it will not manifest without motivation to lead. In other words, motivation to lead is essential for leadership behavior emergence (Popper & Mayseless, 2002). A first-person perception of how one’s values fit with an organization may be at least as important as actual fit, especially when the organization does a poor job of advertising and promoting its values. As Schein (2004) put it so aptly, “if the founders or leaders are trying to ensure that their values and assumptions will be learned, they must create a reward, promotion, and status system that is consistent with those assumptions” (p. 260). Intuitively, it seems that an individual’s motivation to lead could vary depending on how similar they perceive their values to be to those of the organization. PO fit has been shown to predict organizational commitment in a number of studies (Cable & Judge, 1996; Chatman, 1991; McConnell, 2003; O’Reilly et al., 1991; Silverthorne, 2004; van Vianen, 2000; Westerman & Cyr, 2004). As Chan’s (1999) motivation to lead construct is modeled after a theory of commitment, one might expect to also find that PO fit predicts motivation to lead. Therefore, PO fit was assessed as a predictor of general
7 motivation to lead, and each of the three correlated dimensions that make up general motivation to lead: (a) affective-identify motivation to lead, (b) non-calculative motivation to lead, and (c) social-normative motivation to lead. Individual differences could change the relationship between PO fit and motivation to lead. Moral philosophy as defined by Forsyth (1980) is conceptualized using two orthogonal continuous dimensions: (a) relativism (rejecting universal moral rules) and (b) idealism (preferring solutions to ethical problems that cause no harm to others). It is possible that if conflict arises due to poor PO fit, an individual with a higher relative moral philosophy will handle this conflict differently than someone with a lower relative moral philosophy, creating a moderating effect on the relationship. This idea is based on Peterson’s (2003) findings that individuals who did not believe ethics were relative were less committed to their organization when pressured to engage in unethical behavior. Accordingly, this study assessed individual relativism as a possible moderator of the relationship between PO fit and each of general, affective-identity, non-calculative, and social-normative motivation to lead. Although not a situational variable, idealism has also been considered in relation to organizational commitment. Shaub et al. (1993) found that idealism was not related to organizational commitment. However, they did find that idealists were committed to their professions. Given that taking a leadership position may change an employee’s ability to participate in their profession, idealism may predict general motivation to lead, and each of the three dimensions of general motivation to lead.
8 Research Questions Because PO fit has previously been found to predict commitment, and motivation to lead is based on a model of commitment, studying the prediction of motivation to lead from PO fit is a reasonable choice. Relativism has been studied as a moderator of the relationship between ethical conflict and commitment. Similarly, this research considered relativism as a moderator of any relationship found between PO fit and motivation to lead. Another dimension of moral philosophy, idealism, could predict motivation to lead. Previous research found that idealists were committed to their professions, rather than to their organizations. Therefore, idealism was studied as a predictor of motivation to lead. The three dimensions of motivation to lead were also looked at separately as predicted by PO fit, PO fit with relativism as a moderator, and idealism. Studies have shown that a number of personal, job, and organization factors influence PO fit, moral philosophy, commitment, and motivation to lead. For example, age, organization tenure, work experience, previous leadership experience, and current job level were all found to affect motivation to lead (Chan & Drasgow, 2001). Gender and job tenure are known to influence PO fit outcomes (Ostroff & Rothausen, 1997; Young & Hurlic, 2007). Regarding organizational commitment, Sommer, Bae, and Luthens (1996) showed that employees at larger organizations are less committed, and employees with a part-time employment status have been found to exhibit lower levels of job involvement and inclusion (Clinebell & Clinebell, 2007). Furthermore, ethnicity and education were found to influence idealism and relativism (Singhapakdi, Vitell, & Franke, 1999; Swaidan, Rawwas, & Vitell, 2008). As such, these factors were controlled, and all predictions of motivation to lead were made over and above personal (age,
9 gender, educational level, work experience, leadership experience, ethnicity), job (tenure, level, employment status), and organization (size, tenure) characteristics. Specifically, these questions were asked. 1. To what extent does PO fit predict general, affective-identity, non-calculative, and social-normative motivation to lead among employed individuals, over and above personal, job, and organization characteristics? 2. To what extent does relativism moderate PO fit’s prediction of general, affective-identity, non-calculative, and social-normative motivation to lead among employed individuals, over and above personal, job, and organization characteristics? 3. To what extent does idealism predict general, affective-identity, noncalculative, and social-normative motivation to lead among employed individuals, over and above personal, job, and organization characteristics? Hypotheses Hypotheses for PO fit and motivation to lead. The initial model guiding the questions regarding PO fit and motivation to lead, and idealism and motivation to lead is shown in Figure 1. The initial model guiding the questions regarding PO fit, as moderated by relativism, predicting motivation to lead is shown in Figure 2.
10 General Motivation to Lead + 0/+ +
PersonOrganization Fit
Affective-Identity Motivation to Lead
+
Idealism
Non-Calculative Motivation to Lead
0/-
Social-Normative Motivation to Lead
Figure 1. Initial model: Person-organization fit predicting motivation to lead, and idealism predicting motivation to lead.
Low Relativism
PersonOrganization Fit
High Relativism
General Motivation to Lead
+
0
+
Affective -Identity 0 Motivation to Lead
+ +
0
Non-Calculative Motivation to Lead
0
Social-Normative Motivation to Lead
Figure 2. Initial Model: Person-organization fit, as moderated by relativism, predicting motivation to lead.
11 Previous studies have shown that PO fit predicts organizational commitment (Cable & Judge, 1996; Chatman, 1991; McConnell, 2003; O’Reilly et al., 1991; Silverthorne, 2004; van Vianen, 2000; Westerman & Cyr, 2004). Therefore, it is possible that PO fit also predicts general motivation to lead. The hypothesis is that, as in the prediction of commitment from PO fit, lower PO fit will predict lower general motivation to lead. H1: Lower levels of PO fit will predict lower levels of general motivation to lead, over and above personal, job, and organization characteristics.
Because the antecedents of affective-identity motivation to lead are a unique pattern of personality traits that are relatively stable (Chan & Drasgow, 2001), it is not expected that a relationship will be found between PO fit and affective-identity motivation to lead. However, PO fit may affect two antecedents of affective-identity motivation to lead (openness to experience and leadership self-efficacy). Therefore it is also possible that PO fit will predict affective-identity motivation to lead. Because the relationship between PO fit and affective-identity motivation to lead is not clear, the following competing hypotheses are proposed. H2a: PO fit will not be associated with affective-identity motivation to lead. H2b: Lower levels of PO fit will predict lower levels of affective-identity motivation to lead, over and above personal, job, and organization characteristics.
12 Because the compromises that come as a result of low PO fit can be seen as a serious detriment and cost, it is expected that lower PO fit will predict lower noncalculative motivation to lead. Further, because the major antecedent of non-calculative motivation to lead is a collectivist value system, and lower PO fit may decrease a feeling of membership in the collective, those lower in PO fit may be more likely to be aware of and account for this cost when deciding to lead. Further, Chan (2001) found that participants in his original 1999 study had an increased level of non-calculative motivation to lead after they took leadership positions. Lower non-calculative motivation to lead due to PO fit could limit leadership experience. This could reduce the feedback effect from leadership experience to higher non-calculative motivation to lead. H3: Lower levels of PO fit will predict lower levels of non-calculative motivation to lead, over and above personal, job, and organization characteristics.
Because PO fit predicts prosocial behaviors such as teamwork (Posner, 1992) and organizational citizenship behaviors (O’Reilly & Chatman, 1986), it is possible that lower PO fit will predict lower social-normative motivation to lead, as it may be less likely that a feeling of connection and duty toward the organization is present. H4: Lower levels of PO fit will predict lower levels of social-normative motivation to lead, over and above personal, job, and organization characteristics.
Hypotheses for PO fit, relativism, and motivation to lead. When PO fit is low, individuals who believe ethics are based on a universal moral code may be less likely to be motivated to lead, compared to those who believe ethics are
13 relative and based on situation. This parallels a previous study where, when ethical conflict was present, a low relativistic moral philosophy, or belief in a universal moral code, was found to predict lower organizational commitment (Peterson, 2003). However, when ethical conflict was present, there was no effect on organizational commitment for those with highly relative moral philosophies. A similar result is predicted for this study. H5: Lower levels of PO fit will predict lower levels of general motivation to lead when relativism is low, over and above personal, job, and organization characteristics. However, lower levels of PO fit will not predict lower levels of general motivation to lead when relativism is high.
As relativism may cause the employee to see low PO fit as less salient because they are more willing to rationalize and adjust their values (thus reducing the notice, cost, and impact of low PO fit), it is suspected that relativism will moderate the relationship between PO fit and all three dimensions of motivation to lead. H6: Lower levels of PO fit will predict lower levels of affective-identity motivation to lead when relativism is low, over and above personal, job, and organization characteristics. However, lower levels of PO fit will not predict lower levels of affective-identity motivation to lead when relativism is high.
H7: Lower levels of PO fit will predict lower levels of non-calculative motivation to lead when relativism is low, over and above personal, job, and organization characteristics. However, lower levels of PO fit will not predict lower levels of non-calculative motivation to lead when relativism is high.
14 H8: Lower levels of PO fit will predict lower levels of social-normative motivation to lead when relativism is low, over and above personal, job, and organization characteristics. However, lower levels of PO fit will not predict lower levels of social-normative motivation to lead when relativism is high.
Hypotheses for idealism and motivation to lead. Idealism indicates the extent to which an individual feels that “harming others is always avoidable, and they would rather not choose between the lesser of two evils which will lead to negative consequences for other people” (Forsyth, 1992, p. 462). Idealists believe that a moral solution is always possible where no harm comes to another. In contrast, those low in idealism believe that negative consequences are acceptable for some in order to attain positive consequences for others. It seems plausible that a highly idealistic employee will foresee situations where, as a leader with formal influence and power, they would have to make uncomfortable decisions that could lead to negative consequences for others. Therefore, their desire to lead may be lower. This indicates that higher idealism may predict lower general motivation to lead. Others studies (Shaub et al., 1993) found that higher idealism predicted higher professional commitment, but idealism was not related to organizational commitment. Taking a leadership position may affect an employee’s ability to continue in their profession. Changing occupation within an organization is a difficult decision; this decision has been found to be more difficult than leaving the organization and remaining in the same occupation (Blau, 2000). As idealists are committed to their professions, they may find the decision to change to a leadership role especially difficult. Therefore, although no relationship was found previously between idealism and
15 organizational commitment, it is possible that higher idealism will predict lower general motivation to lead. H9: Higher levels of idealism will predict lower levels general motivation to lead, over and above personal, job, and organization characteristics.
Because the antecedents of affective-identity motivation to lead are a unique pattern of personality traits that are relatively stable (Chan & Drasgow, 2001), it is not expected that a relationship will be found between idealism and affective-identity motivation to lead. However, idealism may be related to one antecedent of affectiveidentity motivation to lead (conscientiousness). Further, higher levels of dedication to professional ideals may produce conflict for professionals and lower organizational commitment (Shafer et al., 2002b). Therefore it is also possible that idealism will predict affective-identity motivation to lead. Because the relationship between idealism and affective-identify motivation to lead is not clear, the following competing hypotheses are proposed. H10a: Idealism will not be associated with affective-identity motivation to lead. H10b: Higher levels of idealism will predict lower levels of affective-identity motivation to lead, over and above personal, job, and organization characteristics.
Similar to PO fit, because an idealist might see potential conflicts as a nonnegotiable cost of leadership, it is expected that higher idealism will predict lower
16 non-calculative motivation to lead. In summary, the inordinate costs to the idealist of leading may bring an individual to give these costs more consideration. H11: Higher levels of idealism will predict lower levels of non-calculative motivation to lead, over and above personal, job, and organization characteristics.
Idealism, with its preference for avoiding decisions that may affect another individual negatively, may exert a force greater than a feeling of duty to the group. In addition, Shafer, Lowe, and Fogarty (2002a) found that idealists become desensitized and defensive, also indicating a lower feeling of obligation towards the group. Therefore, it is suspected that higher idealism will predict lower social-normative motivation to lead. H12: Higher levels of idealism will predict lower levels of social-normative motivation to lead, over and above personal, job, and organization characteristics. Definition of Terms Affective-identity motivation to lead is a first-order factor representing motivation to lead based on liking to lead and seeing oneself as having leadership qualities based on past leadership experience. Ethics is a theory or system of moral values. General motivation to lead is a second-order construct of motivation to lead that accounts for the common variance among the three first-order factors of affectiveidentify motivation to lead, non-calculative motivation to lead, and social-normative motivation to lead (Chan & Drasgow, 2001).
17 Idealism is a personal belief that ideal consequences that cause harm to no one can always be attained when making a moral judgment. Morals provide motivation based on ideas of right and wrong. Moral philosophy is a personal theory of right and wrong. An individual’s moral philosophy gives guidelines for moral judgments and suggests actions in ethical dilemmas. Moral philosophy is also known as ethical ideology. Idealism and relativism are two types of individual moral philosophy that were identified by Schlenker and Forsyth (1977). Motivation to lead affects the decision to assume leadership training, roles, and responsibilities, and the amount of effort given to leading and persistence as a leader (Chan & Drasgow, 2001). Non-calculative motivation to lead is a first-order factor representing motivation to lead based on not requiring rewards for leading and being generally agreeable to leading, even without prior leadership experience or feelings of leadership self-efficacy (Chan & Drasgow, 2001). Perceived supplementary person-organization fit is the extent to which the values of an individual are similar to those of their organization as reported directly by that person. Relativism is a personal moral philosophy, where the correctness of a moral judgment is not considered absolute. Rather, moral judgment is correct relative to the convictions and practices of a culture. Further, universal moral rules are not considered possible. Relativists consider situation and personal values over ethical principles when making a decision.
18 Social-normative motivation to lead is a first-order factor representing the motivation to lead based on a sense of social duty, being accepting of social hierarchies, and rejecting of social equality (Chan & Drasgow, 2001). Universalism is a personal moral philosophy, where the correctness of a moral judgment is based on absolute and universal principles. Values are enduring beliefs that a type of conduct or end-state mode of existence is preferable, on a personal or social level, to opposing types of conduct or end-states of existence (Rokeach, 1973). Values are considered when making ethical decisions. Limitations The present study required participants with a variety of attributes and experiences, across several different organizations, and at various job levels. As it was not feasible to meet these sample requirements by partnering with multiple organizations, a convenience sampling method was used. Although non-random sampling limits the external validity of the study results, this method enabled the sample to represent a diversity of participants and organizational contexts, giving an increase in external validity over most organizational research, which usually involves participants from a single organization (Eaton & Struthers, 2002). The sample consisted of 1,024 adults employed by organizations in the United States. Because a U.S. sample was used, the study results cannot be generalized to employees in other countries. Also, as the majority of participants were adult learners at an online university, and the other participants were friends and work colleagues who were mainly professionals, the high educational and job levels of the sample may have affected generalizability.
19 The study was conducted using an online survey. The participants remained anonymous and the survey was hosted by a third party to ensure the confidentiality, reliability, and safety of the data. However, online surveys can suffer from sampling biases due to data introduced when an uninvited respondent completes the survey. Even though an uninvited respondent could complete a mailed survey, online surveys may be more susceptible to fraudulent data, as the survey is universally available to anyone who happens upon it. Therefore, to increase data validity, the survey was completed by invitation only and a password was required to enter the survey. As self-report measures were used, and predictor and criterion variables were reported by the same individual at the same time, common method bias may have occurred (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Summary and Conclusions We need to motivate the types of individuals who will create healthy, resilient, and productive workplaces, which may not necessarily be the type of person already leading in the organization. If those with a values mismatch (and especially those who are not willing to compromise their values) are not motivated to lead, the leadership pool may become homogenous enough to create a negative effect. For example, it has been proposed that leadership homogeneity can decrease the organization’s ability to react and change with its external environment (Bowen, Ledford, & Nathan, 1991; Giberson et al., 2005; Powell, 1998). A first step to determining if a “fit” filter is creating values homogeneity is to determine if PO fit predicts the will to pursue (or avoid) leading. Chan (1999) has developed a measure of motivation to lead. However, situational variables, such as PO
20 fit, are not included. This dissertation explored the predictive strength of one situational variable, PO fit, for Chan’s motivation to lead construct, directly and as moderated by a relative moral philosophy. In addition, idealism may predict motivation to lead, especially for professionals. Therefore, the predictive strength of idealism for motivation to lead was also assessed.
21 Chapter 2: Review of the Literature Introduction This literature review gives analyses of PO fit, moral philosophy, and motivation to lead. An understanding of previous research on PO fit and its various forms helps to identify and justify the conceptualization, operationalization, and measurement of PO fit used to predict motivation to lead. Special attention is given to how PO fit predicts commitment, an attitude on which motivation to lead is based. Further, links among moral philosophy dimensions (relativism and idealism), commitment, and motivation to lead are explored. A treatment of the current literature on motivation to lead, especially the area of motivation to lead and situation, follows. The review ends with a short summary of the literature and gives a description of how the present study was conceptualized from it. Conceptualizing Person-Organization Fit The person-organization fit focus. Person-environment fit (PE fit) represents a broad array of interactions that can occur between a person and environment. PE fit can occur between a person and another individual (e.g., a supervisor, as in PS fit), a job (PJ fit), a group (PG fit), a vocation (PV fit), or an organization. These fit types are currently being explored to create a multidimensional theory of PE fit (Jansen & Kristof-Brown, 2006). In addition, the attribute used to determine the content of the interaction can vary, as can the outcome of the interaction, which can be attitudinal or behavioral. PE fit has been conceptualized and operationalized in a variety of ways. In fact, in a recent meta-analysis of 172 studies, Kristof-Brown, Zimmerman, and Johnson (2005) found that there is no agreed upon way
22 to define, measure, or account for the impact of PE fit. There is also some disagreement on how best to proceed in utilizing and extending research in the area. For example, Judge (2007) calls for an ideational consolidation of terms and continuing research in current directions, whereas Harrison (2007) suggested that the actual scope and boundaries of PE fit should be revisited and possibly revised. Papavero (1999) conducted a qualitative study of six software engineers that found that perceived values dissimilarity between the individual and the organization diminished the desire to advance. Based on that finding, the present study is concerned with measuring the relationship between an individual’s perception of the similarity of their values and those of their organization, and how that perception relates to their decision to take a leadership role. This goal identifies the environment of interest to be the organization, and the type of fit to be studied as PO fit. The individual may exercise several options when understanding and processing the term “organization.” It has been found that individuals consider both the organization and other employees in their conceptualization of PO fit (Billsberry, Marsh, & Moss-Jones, 2004). Further, individuals do see the organization as a different entity than the employees that comprise it (Piasentin, 2007). Although not often mentioned, another area of concern is whether it is best to assess fit in a single organization, or across organizations. Ostroff (2007a) suggested that it is preferable to measure fit across organizations, as is done in the present study. Choosing a PO fit interaction type. An interaction between person and organization can occur when these two entities are similar (supplementary PO fit), or when they differ but also mesh and interlock
23 (complementary PO fit). Further distinctions on the nature of complementary fit interactions between person and organization have been made (Kristof-Brown et al., 2005). Needs-supplies PO fit occurs when individual needs are met by environmental supplies. Demands-abilities PO fit occurs when individual abilities meet environmental demands. There is still some question as to whether complementary PO fit reflects needssupplies PO fit or demands-abilities PO fit (or both), or whether complementary fit is a type of PO fit independent of these. Edwards and Shipp (2007) treat needs-supplies PO fit and demands-abilities PO fit as types of complementary PO fit, as do Cable and DeRue (2002). However, Piasentin (2007) found support for complementary PO fit as a construct distinct from each of needs-supplies fit and demands-abilities fit that represents “perceptions of being different in the organization combined with the belief that these differences are valued by the organization” (p. 121). Harrison (2007) pointed out that it is not tenable to study all possible person and environment interactions. To bound its definition, he classified fit as the intersection of both entities on a shared commensurate attribute. Supplementary fit represents affinity, where there is similarity in the magnitudes of compatibility (or intersection), which comes from person and environment sameness. By commensurate, Harrison was clear that attributes must be defined by the same or similar content. As complementary fit represents joining rather than intersection, it is nonviable under Harrison’s definition of fit. However, Judge (2007) found Harrison’s definition of fit as intersection, and his requirement for strictly commensurate attributes, too restrictive. Judge gives the study of high achievers seeking organizations that pay well as an obvious example of fit that would fail Harrison’s characterization.
24 There is some debate as to whether supplementary or needs-supplies fit is better suited to predicting attitudinal outcomes. Kristof-Brown et al. (2005) suggested that needs-supplies fit has a stronger and more direct link to outcomes than supplementary fit. However, this has not been proven. Cable and Edwards (2004) found that both complementary and supplementary fit affected attitudes. However, although these were interrelated, they also contributed independently to outcomes. So, as proposed by Kristof (1996), supplementary fit may represent the similarity between individual values (the needs side of complementary fit) and the cultural values of the system of the organization (the supplies side of complementary fit), making supplementary fit one possible instantiation of needs-supplies fit. Further supporting Cable and Edwards’ model, Greguras and Diefendorff (in press) found that supplementary PO fit had indirect effects on employee outcomes through psychological need satisfaction, as the need for autonomy mediated the relationship between PO fit and affective organizational commitment. Edwards and Shipp (2007) further explored the relationship between supplementary and needs-supplies fit. Their results showed that supplementary fit influenced affective (affiliation needed), continuance (benefits needed), and normative (shared values needed) commitment through needs-supplies fit (which they view as a type of complementary fit). Edwards and Shipp found that supplementary fit predicted attitudes indirectly by affecting the needs of people and the organization’s ability to meet these needs. However, Karakurum (2005) saw a more direct relationship between perceived supplementary PO fit and affective commitment. Karakurum used multiple hierarchical regression analysis to study the relative abilities of PO fit measurement types (perceived supplementary PO fit and four indirect measures) to predict organizational
25 commitment. Surveying 180 employees from various departments of a Turkish public company, Karakurum found that directly measured PO fit was the best predictor of commitment. Karakurum suggested that perceived supplementary PO fit predicts affective commitment because “employees perceive an emotional attachment to the organization or identify with the organization. Congruence between personal and organizational values can be cited as one of the most important factors underlying such an emotional attachment or identification” (p. 86). Considering motivation may give direction for exploring the relationship between supplementary fit on values, and complementary fit as needs fulfillment in the context of values. Latham and Pinder (2005) discussed needs-based theories in terms of motivation, noting that these theories explain why a person must act, but they do not explain why specific actions are selected in specific situations. They propose that situational knowledge, assessments, and intentions are driven by an individual’s values. Values may be rooted in needs, but they provide the principled basis for goals, and guide the behavior of an organization’s members (Chatman, 1989). Values are similar to needs, as they arouse, direct, and sustain behavior, i.e., motivate. However, Latham and Pinder view needs as inborn, whereas values can be acquired through cognitions and experience (i.e., needs are more stable than values). In their view, values are closer to action than needs. Needs are common among most individuals, but values differentiate the actions taken to satisfy them (Pinder, 2008). For example, gaining the esteem of others is a common need. However, a person who values material acquisition might wear expensive jewelry to gain esteem, while someone who values service might volunteer in the community (Pinder). An example more applicable to a work setting also considers
26 environmental supplies of the need for esteem. If individual values of material attainment are met with public recognition for mentoring, fit may not be present. If met with a bonus, fit might be realized. Conversely, an individual who values service would be more likely to fit an organization offering recognition, rather than one offering bonuses. The nature of the environmental supply must match the individual’s values, in addition to meeting the more universal need. That is, the need for esteem must be met with a supply that satisfies the values of the individual. Therefore, it appears that supplementary PO fit on values could be considered a context of motivation where, as individual and organizational values similarity changes when situations arise, motivation increases and diminishes. This view seems related to Edwards and Shipp’s (2007) assertion that fit on values predicts attitudes through needs; however, it describes the converse situation where needs predict attitudes through fit on values. It seems clear that further work is required to determine possible interactions and precedence of needs and values when determining behavioral and attitudinal outcomes of PO fit. The issue of the relationship between needs-supplies fit and supplementary fit, and the underlying meaning (and appropriate uses) of supplementary and complementary fit remains unresolved, and research is in preliminary stages. However, in their metaanalysis, Kristof-Brown et al. (2005) found supplementary fit had a larger effect for commitment than needs-supplies fit. More recently, Piasentin and Chapman’s (2007) results showed that both supplementary and complementary fit contributed to commitment, with supplementary fit being a much stronger predictor. Their results also showed that, although similarity is central to subjective fit, complementary fit also plays a role when similarity is low.
27 Operationalizing PO fit with values. The content used to compare PO fit is usually determined by the fit interaction type. Abilities-demands fit research has focused on knowledge, skills, and abilities (KSAs), and organizational demands, while needs-supplies fit research measured general individual need preferences (Kristof-Brown et al., 2005). However, personality traits and values have been used with needs-supplies fit studies, and values have been included in abilities-demands fit studies (Kristof-Brown et al.). More recently, Piasentin (2007) included values and goals in her assessment of complementary fit as a unique construct. For supplementary PO fit, the same content must be measured for both the person and organization. Values, goals, personality traits, and attitudes are the most common content types. There is disagreement as to which content dimension is best at predicting particular outcomes of supplementary PO fit (Kristof-Brown et al., 2005). In early research, the meaning of PO fit was interwoven with values content. Chatman defined PO fit as "the congruence between the norms and values of organizations and the values of individuals” (1989, p. 339). More stable content types are thought to have a larger effect on attitudes and behavior (Kristof-Brown et al.). For example, if personality were considered more stable than values, then personality should offer a better result than values (Kristof-Brown et al.). However, as Kristof-Brown et al. noted, if similarity on all personality traits is not related to the outcome, then content consisting of all personality traits would have a smaller effect on the outcome than values. Further, measuring values alone produced only slightly weaker relationships than a combination of values, personality, needs, and KSAs (Kristof-Brown et al.). This supports Chatman’s emphasis
28 on values as fit content, and gives further support for operationalizing fit using values in the present study. Choosing a view of PO fit. Subjective PO fit considers PO fit from the viewpoint of the individual. Subjective PO fit can be measured indirectly, where individuals evaluate themselves and the environment, with these evaluations compared afterwards. Alternatively, PO fit can be measured directly, where individuals report their perception of the compatibility of themselves and the environment. Subjective PO fit that is measured directly is often termed perceived PO fit. PO fit can also be viewed objectively, by assessing fit indirectly using different sources. The content to be compared is assessed for the individual, by the individual. The same content is then assessed separately for the environment, by another source. The two assessments are then compared to measure objective PO fit, which is also known as actual PO fit. There is some disagreement as to the value of subjective PO fit. Objective and subjective PO fit are not equivalent (Ravlin & Ritchie, 2006; van Vuuren, Veldkamp, de Jong, & Seydel, 2007). Subjectively measured PO fit has been found to have the strongest relationship with outcomes, especially attitudes. However, subjective measures can suffer from common method bias (Kristof-Brown & Jansen, 2007). There is also some debate specific to the value of directly assessed (i.e., perceived) PO fit. Perceived PO fit does not give the direction of variance (Edwards & Shipp, 2007). Therefore, perceived PO fit operates at a global level and represents similarity in a general sense with no dimensions of comparison. This limits analysis of the functional form of the relationship between PO fit and the outcome (Ostroff, 2007a).
29 For example, finding higher subjective PO fit on values reveals nothing about how the values of the person and organization differ, only that they do differ. If the values of the organization were highly ethical and the employee’s were not, or vice versa, the difference in outcome variance would be identical. Perceived PO fit has been categorized as a viable construct that captures the overall affective reaction to the contextual environment. It has been suggested that perceived fit should be used only in certain restricted cases (Harrison, 2007; Ostroff, 2007b). In fact, Harrison called for perceived PO fit to be reclassified out of the area of fit. Perceived PO fit may be an affective measure of the extent to which a person perceives or feels they fit, but it is still a construct that can be studied in its own right as an implicit theory of fit (Cable & DeRue, 2002; Cable & Edwards, 2004; Finegan; 2000; Kristof-Brown & Jansen, 2007; Ostroff, 2007a). Perceived PO fit is especially suited to examining feelings of fit at an individual level of analysis, where fit is unique to each individual (Autry & Wheeler, 2005; Kristof-Brown & Jansen; Ostroff, 2007b). Efforts are being made to define and differentiate subjective PO fit approaches so they can be examined in more detail. Edwards et al. (2006) studied the cognitive process of comparison used to determine subjective needs-supplies PE fit (note that these results may not generalize to others types of fit, such as supplementary fit). Three approaches to studying subjective PE fit were defined. Molar fit signifies affect and gives a holistic assessment of similarity, rather than a judged match of perceived personal needs and environmental supplies. Molar fit appears to be equivalent to perceived fit. Subjective fit can also be molecular, where the individual compares themselves to others, but not with a focus on similarity. Finally, subjective fit can be atomistic, a reductionist approach,
30 where the individual describes their own values and those of the organization, which are then compared. These three approaches were found to be distinct, but issues were found with each. With the molar approach, fit varied with the direction of the relationship between the person and environment. When environmental supplies exceeded the person’s needs, fit increased on dimensions such as pay and vacation time, but when the person exceeded the environment, fit decreased on dimensions such as span of control and supervision. The molecular approach produced different results depending on whether the person was the target (my needs exceed the organization’s supply) or the environment was the target (the organization’s supply does not meet my needs), suggesting that an unequally weighted comparison was taking place. Edwards et al. explored these results further in an unplanned post hoc analysis and found that both the molar and molecular approaches had high correlations with satisfaction, and may actually represent satisfaction rather than fit. They also found that molar and molecular judgments were not made by combining atomistic impressions of the environment. Further, they observed that atomistic perceptions are probably subjective because individuals could use some other standard of comparison, such as other experiences or referent others, when making judgments. Edwards et al. concluded that it is not clear as to exactly what subjective PE fit represents, and that more research is needed to find its true meaning. Others continue to support perceived PO fit as a viable construct for which objective measurement is not appropriate (Autry & Wheeler, 2005; Caldwell, 2003). According to Kristof-Brown et al. (2005), perceived PO fit allows the greatest level of cognitive manipulation, as the individual applies their own weighting, and it continues to be the best predictor of attitudes because it gives a holistic assessment of fit. Supporting
31 this view, Cable (personal communication, as cited in Caldwell) gave a description of perceived PO fit as molar and representing “idiosyncratic processing of demands, needs, and attributes of the environment relative to themselves in ways that cannot be constructed objectively by researcher's formulae” (p. 47). Giving some clarification as to which values are considered by individuals when reporting perceived PO fit, human relations values predicted perceived fit best (van Vuuren et al., 2007). van Vuuren et al. suggested that, when reporting perceived PO fit, respondents are thinking of “typically human values, ethics, or morale, and neglect values like stability and innovation, for which ‘values’ is a less obvious connotation” (p. 1743). Although the status of subjective PO fit is still being considered, Judge (2007) calls for consensus, and cautions that free intellectual thought should continue to be exercised to explore the issues further. In an example of continuing efforts to elicit the meaning of subjective fit, Billsberry, Ambrosini, Marsh, Moss-Jones, and van Meurs (2005) have used causal mapping and storytelling as methods of uncovering and organizing individual experiences of fit. A direct measure of subjective PO fit is the most widely used because it has consistently been found to obtain results for attitudes (Kristof-Brown & Jansen, 2007). Recently, Piasentin and Chapman (2007) identified perceived PO fit as a unique construct that predicts attitudes. Although perceived PO fit has been criticized as increasing common method bias, it continues to reflect reality, as interactional psychology theory suggests that people can only be influenced by fit with their environment as they perceive it (Kristof-Brown et al., 2005). Perceived PO fit gives no information regarding the direction of fit variance. However, as the goal of the present study is to determine if any
32 outcome variance occurs, it is not concerned with the direction of the variance (nor will it make judgment on whether one direction of variance is qualitatively different from the other, i.e., better or worse). PO misfit. The topic of misfit continues to receive increased attention. In Harrison’s (2007) view, examining fit produces too much information, and may not be as productive as examining misfit, which is more focused and manageable. Employees may not start as misfits; they either fit or are neutral (Billsberry et al., 2005). Billsberry et al. proposed that a misfit is something one becomes, alluding to a process of moving from fit to misfit. Fit cannot be assumed to be positive, as those with higher fit can become complacent (Harrison, 2007). However misfit can also be detrimental. Individuals may not be able to choose where they work, which may increase the level of detrimental misfit in an organization (Schneider, 2007). When misfits think there are no viable alternatives to their current position, they stay. With a sample of 205 employees from two regions in the U.S., Wheeler, Gallagher, Brouer, and Sablynski (2007) used an online survey, and hierarchical mediated and moderated regressions, and found that low PO fit was more likely to result in intent to turnover when perceived job mobility was high. Wheeler et al. did note a limitation, in that actual turnover was not measured. Staying in a job when PO fit is low may result in cynicism, which may have a negative impact on the individual and organization (Naus, 2004). Billsberry (2007) has proposed four fit meta-categories (that appear to fold in the impact of consequences of misfit) that may be useful in determining when misfit is damaging. When fit is advantageous for both the individual and the
33 organization, misfit is detrimental for both. While self-serving misfit is detrimental for the organization, organization-serving misfit is negative for the individual. Detrimental misfit can stem from a variety sources. For example, Billsberry et al. (2005) found that while senior employees did not value work-life balance, lower-level employees did. It should be noted that higher-level employees might have better access to resources that can help them cope with work-life conflicts, which may reduce the salience of misfit. Regardless of the reasons for differences in employee levels, Billsberry et al. note that the importance of work-life balance may have been underplayed in the past, given that most studies of fit involve managers. If management requires increased commitment and a devaluing of work-life balance, then valuing work-life balance may contribute to lower motivation to lead (Papavero, 1999; Stern, 2003). However, addressing this type of misfit may not be straightforward. With a stratified random sample of 460 employees at a large university in the U.S., Rothbard, Phillips, and Dumas (2005) used hierarchical multiple regression and found that individuals who preferred work-life segmentation and had access to integration policies (e.g., onsite childcare) were less committed than those with less access. Further, individuals who wanted work-life segmentation and had access to segmentation policies (e.g., flextime) were more committed. These results were found over and above age, gender, domestic partnership status, income, number of children, and age of children. However, Rothbard et al. did note that their study might have been limited by common method bias (although Harman’s one-factor test found common method variance was not present), and the fact that the sample was sourced with one university.
34 Billsberry et al. (2005) found that a sense of misfit most often resulted from poor fit with direct management and organizational values. Talbot, Billsberry, and Marsh (2007) later found different root causes for fit (e.g., job, environment, and colleagues), suggesting misfit as a unique construct. Further, the same root cause resulted in fit or misfit depending on subsequent managerial actions. The line between fit and misfit, which is perhaps the point at which low fit becomes detrimental, is not clearly defined. However, fit seems to be emerging as a categorical construct. Those employees who fit well may also have elements of misfit, which may not be entirely detrimental. As Talbot et al. state, “it is possible that this is desirable in employees; perhaps the people who are able to look critically at organisational behaviours, policies, procedures and others’ behavior are the best fit” (p. 12). Much like heterogeneity at the organizational level, a state of misfit for the individual could be a useful state that brings necessary change and increased resilience at the individual and, perhaps, organizational levels. Of course, that change may entail leaving the organization. Arthur, Bell, Villado, and Doverspike (2006) pointed out that subgroup differences in PO fit are “understudied or not reported in the extant literature” (p. 797). Those who see themselves as misfits, which may include a disproportionate number of women and minority group members, may leave (Hoobler, 2005). However, this might not produce the best outcome for the organization, especially if diversity is a priority. Welsh and Dehler (2001) described conditions that may lead to misfit, and different reactions to misfit in the form of resistance versus leaving the organization. Even with efforts to change and modernize organizations, control processes have not
35 inherently changed. A colonization process is used to obscure the contradictions between demands of a high level of commitment from the employee, and the low level of commitment actually received by the employee (Welsh & Dehler). Colonization works best when employees have higher PO fit, but as commitment to employees wanes, lower PO fit may result, and resistance may increase. Resistance can create heretics, who are low in PO fit, but apply critical reason to identify values incongruity and invite debate. Membership in the organization is essential for heretics in maintaining separateness and context for resistance. Eventually, heretics may surrender their separate selves, or they may become tempered radicals who seek to maintain both personal and organizational identities simultaneously in order to advance in the organization. However, tempered radicals spend more energy balancing opposing forces than transforming the organization. It is likely that when heretics or tempered radicals express dissent, their advancement is prevented. Constructive deviants, whose personal identity is separate from the organization, occur at all organizational levels. Constructive deviants take action on discrepancies identified by the heretics and tolerated by the tempered radicals (Welsh & Dehler). They identify organizational limits and use action to break through for their own personal transformation and emancipating change (Welsh & Dehler). Conceptualizing PO fit in the present study. No matter what terminology is used Harrison (2007) suggests that any study concerning fit make all assumptions clear. For the present study, a molar, or direct, comparison is made on content consisting of values. Fit here is a subjective perceived construction, not a cognitive representation. Maximum and minimum fit have no meaning in this study, as no strict correlation of the person and environment is being
36 measured independently. However, if necessary, minimum and maximum fit could be defined as the lowest and highest values on the perceived fit scale. No attempt is made in the present study to categorize fit or misfit. It should be noted that although it remains an open question as to whether misfit is a continuum moving from fit to misfit, recent research suggested that misfit is a distinct construct (Talbot et al., 2007). In any case, the need to characterize participants as misfits is not anticipated. The present study considers PO fit as fit to the organization itself as an entity, rather than to other members. Ostroff and Schulte (2007) described this mode as personsituation fit that characterizes features of a situational context (values), which is differentiated from person-person fit. Strictly speaking, the present study is concerned with the social person-situation subtype of PO fit in terms of values congruence. As the present study examines the similarity of the entities being compared for purposes of exploring individual outcomes, the relationship can also be characterized as supplementary, rather than complementary. Further, as the latest findings on supplementary, needs-supplies, and complementary fit have shown that supplementary fit continues to be a good predictor of an attitude with similarities to motivation to lead (i.e., commitment), supplementary fit will be used in the present study. Finally, the present study is mainly concerned with how an individual’s perception of fit motivates them to lead or not lead. The goal is not to assess how other environmental factors contribute to fit, but to determine if fit perception contributes to motivation to lead. In summary, the present study is concerned with the relationship between an individual’s perception of fit and their decision to take a leadership role (or not), not all causes for this attitudinal outcome. Further, directly perceived PO fit gives
37 the strongest correlation with commitment (Verquer, Beehr, & Wagner, 2003), and may do the same with motivation to lead. Therefore, perceived supplementary PO fit on values is appropriate and is used in the present study. For brevity, the term PO fit will be used to identify perceived supplemental PO fit on values in subsequent references. PO Fit and the Motivation to Lead Individual outcomes of PO fit. Billsberry (2004) noted that a good number of studies have shown that low PO fit can predict attrition. However, there is a scarcity of studies on the role that low PO fit plays in incumbent attitudes and job performance. In their meta-analysis of 25 studies, Kristof-Brown et al. (2005) confirmed the relevance of PO fit for post-entry attitudes, with commitment being the most strongly predicted. Several studies found links between PO fit and outcomes related to motivation to lead, including commitment, and some researchers have made proposals in the same area. In a meta-analysis of over 100 studies regarding PO fit, job performance, and attitudes, Arthur et al. (2006) found the relationship between PO fit and organizational commitment to be strong and generalizable. Further, PO fit may impact motivation to lead indirectly through commitment. In a study of 103 white-collar employees, Fowke (1998) found that layoffs lowered affective commitment, which in turned decreased career motivation. Because Arthur et al. found a much weaker relationship for PO fit and job performance they suggest that, compared to using PO fit in entry-level selection, PO fit is more useful in “employment related decision making, including promotions, appointments to leadership positions, transfers, terminations, and even the formation of work teams” (p. 797).
38 Billsberry et al. (2004) used causal mapping and storytelling with 63 individuals and found that fit on opportunities for growth and development was important enough that employees with poor fit in this area would leave the organization. However, no mention was made of the likely outcome when poor fit occurs because growth and development at work is not preferred. This situation could occur when the organization demands growth and development that benefits its goals, while the individual prefers growth and development outside of work that benefits individual goals. In other words, the result of low fit on growth and development opportunities could vary depending on whether strong work-life segmentation is desired by the individual. Finegan (2000) surveyed employees at a large petroleum company and obtained a sample of 121 mostly male (83%) employees. Using hierarchical multiple regression, Finegan found that PO fit predicted affective commitment, but not normative or continuance commitment. However, he notes that the results are limited as all participants were from the same organization, and most had long tenure with the organization. In a study of 783 graduates from two industrial relations programs, Bretz and Judge (1992) used regression analysis and LISREL, and found that individuals with higher PO fit achieved higher levels of intrinsic and extrinsic success. They did note that this was not a longitudinal study and, therefore, fit could not be measured before assessing success to determine a causal relationship. Bretz and Judge suggest that once a promotion decision is made based on fit, false-positive selection errors will not be undone by subsequent decisions because the selected individual competes in a smaller and more homogeneous group at each level. If an individual is passed over at lower levels because of perceived low fit, it is unlikely that they will be given the opportunity to compete at
39 higher levels. These false-negative selection errors are particularly damaging since those who might have been highly successful at higher levels are less likely to be considered. These inaccurate decisions all occur under conditions of sponsored mobility. This implies that low PO fit could reduce sponsored mobility (Bretz & Judge), which could in turn affect an individual’s motivation to lead. Haley and Sidanius (2005) proposed that PO fit and promotion may be linked based on socio-political attitudes. Socio-political attitudes reflect a preference for hierarchy attenuating or hierarchy enhancing institutions as defined by social dominance theory. Previous studies have found that egalitarian individuals are attracted to hierarchy attenuating organizations, whereas those who endorse social hierarchies are attracted to hierarchy enhancing organizations (Haley & Sidanius). Socio-political homogenization could result due to self-selection, organizational selection, organizational socialization, differential rewards, and differential attrition rates (Haley & Sidanius). As Chan (1996) stated, “over time, individuals in cognitive misfit are likely to be less motivated, less committed, and experience more work-related stress and job dissatisfaction than those in fit” (p. 199). Haley and Sidanius see this as “a process that should affect not just an individual’s turnover intentions, but also the likelihood that an individual will suffer when it comes to salary, promotion, and layoff decisions” (p. 196). In essence, those who do not fit may have less opportunity to gain the leadership experiences that contribute to leadership self-efficacy, which in turn decreases motivation to lead. Organizational outcomes of PO fit. Argyris (1954) studied the organization of a bank through observations and interviews and found a caste-like system between employees and officers that had
40 adaptive value in that it helped maintain peace by minimizing interactions. People became increasingly agreeable to avoid confrontations. However, 89% of the employees said inflexibility and rigidity were detrimental because everyone continued doing things in the same way, simply because it was the way it was done before. It appeared that the bank perpetuated the personality that came to the organization through feedback and fusion (Argyris). This observation, although offered over fifty years ago, seems to reflect the same process of homogenization in organizations, and the accompanying complacency, that is prompting researchers to turn their attention to the positive aspects of fit diversity for organizations today. The dynamic processes underlying PE fit are described by Schneider’s (1987) attraction-selection-attrition theory (ASA). ASA focuses on both individual and organizational outcomes. Because this theory argues that individual fit may lead to organizational homogeneity over time, the consequences of individual fit could transcend individual outcomes. Although not the focus of this study, the consequences of lowered motivation to lead as a result of poor PO fit may very well create long-lasting effects for the organization as a whole. For example, Scott (2000) found that individuals were more likely to interview with organizations that fit their moral values, and suggested that as a result the range of values represented in the organization would become smaller. Current thought continues to focus on the causes of homogeneity and its consequences. Schneider (2007) noted that organizational development change practices would not work because people in the organization are similar and comfortable. He suggested that it is impractical to change the homogeneity of personalities of organizational incumbents, as homogeneity is a cause for resistance. Boone et al. (2004)
41 offer a concrete example of this, where top management teams experienced homosocial reproduction and closed ranks during crisis. Boone et al. suggested it would be better to diversify the team to help their membership cope. van Vianen and Stoelhorst (2007) proposed that bottom-up fit produces homogeneity in the organization through the behavioral homogeneity of conformity. Individuals prefer to copy similar others and those whose behavior has brought the highest payoff. People may especially imitate prestigious models in the organizational hierarchy, staying as long as the benefits outweigh the costs of adaptation to the organizational culture. Those with low fit might try to create a niche to adapt and fit in, and leave if they cannot. This may explain why homogeneity develops more readily in stronger cultures, where niches are discouraged. The power of imitation may relate to low non-calculative motivation to lead, where a leadership role is not taken if the perceived cost is too high. It is possible that the cost of leadership figures highly in the decision to identify with and imitate current leaders, and that what is considered acceptable cost and payoff varies widely. For example, the definition of success and prestige may not be equivalent for someone who values home, family, and community first, versus someone who values material wealth first. Favorable models of behavior, worthy of imitation, probably vary by individual. Imitation of current leadership would produce leaders who share values with incumbent leaders. Low propensity to imitate current leaders by those who do not share their definition of success would also feed reproduction of current leadership values. This could also work in a top-down fashion, as Giberson et al. (2005) found that organizational values are congruent with the values of top leadership because leaders surround themselves with similar others.
42 Zhang, Dolan, Straub, and Kusyk (2007) observed that both life and work values are important to fit and wondered if female values in executive boards would decrease scandals and fraud. Diversity of values in teams can test the ethics of the existing organizational culture. Dukerich, Nichols, Elm, and Vollrath (1990) found that the moral reasoning of the group was at a higher level after discussion, but some individuals moved higher than their original level and some moved lower. Moon and Woolliams (2000) later found that ethical debate changed the individual values and norms of groups. Finally, Nelson and Billsberry (2007) point out that it is not clear that organizational homogeneity is advantageous or detrimental, no matter what the situation. No study to date has proven this either way because the effect of fit on organization-level performance has not been shown successfully. The positive and negative effects of organizational homogeneity continue to be considered in the literature. However, it is still unknown as to whether the values homogeneity that may result from PO fit is a detriment. This question warrants consideration. In fact, fit with an environment is probably not beneficial when the environment is unethical. How PO fit changes. PO fit can change over time as the individual and organization interact. A central proposal in the present study is that motivation to lead is not only associated with PO fit, but that changes in PO fit, whether sourced with the individual or organization, may increase or diminish the motivation to lead. The study of the idea of temporal fit, where estimates of the environment change over time, is in its early stages (Kristof-Brown & Jansen, 2007). PO fit can change with time, as individuals may change, or the individual may change the organization. For example, goals and values can change in importance as
43 groups evolve, which may impact the needs of the group. PO fit can change when the organization changes (Caldwell, Herold, & Fedor, 2004) and when career stage changes (Powell & Meyer, 2004; Shafer et al., 2002b). One example of an individual changing the organization would be when a cooperative organization hires a competitive person, who changes the organization over time (Chatman, 1989). An example of an organization changing an individual would be that of a professional, whose values shift from those of their profession to those of the organization as they advance. Age and other individual attributes may be related to the interplay of values and needs, and how these impact on PO fit change. For example, PO fit was found to be unaffected for older workers, but changed for younger workers, at an organization experiencing turmoil (Shafer et al., 2002b). PO fit can be consciously extended or reinforced by the organization (Powell, 1998). Which is preferable depends on personal and job attributes, and situational characteristics. In general, reinforcement is needed for values central to the organization, especially for lower levels of the organization, and at early stages of the organization. However, according to Powell, values should be extended for those at higher levels of the organization who have decision-making responsibilities. This seems to imply that, as organizational needs change due to context, values can be extended or reinforced to meet these needs. Extending values at higher levels of the organization may be difficult to accomplish. West (2007) found that conflict within top management teams (TMTs) was detrimental, especially when values diverged. This led to task and relationship conflict, and lower organizational commitment. On the other hand, Yokota and Mitsuhashi (2008) found that long-term reproduction of demographically similar executive teams caused
44 inertia and an inability to meet changing external environments. Like Powell, West suggested that optimal diversity (or optimal homogeneity) is needed. Although the riddle of exactly how changes in needs and values occur, there is probably a difference in the result of these changes based on other factors, such as the consciousness and transparency with which the changes are approached. One might wonder, for example, how consciously and explicitly introducing conflict, as suggested by Powell, and Yokota and Mitsuhashi, might produce different results than allowing mismatch to occur by chance, as described by West. Although not addressed directly by the present study, conscious change of the values mix of leadership teams could be accomplished by encouraging the full participation of qualified individuals who avoid leadership roles due to low PO fit. PO fit, individual characteristics, and situation. PO fit outcomes have been shown to be affected by gender (Young & Hurlic, 2007). Women with low PO fit sought promotions more often when the organization accepted gender differences in behavior (a tolerant macro culture) than if gender differences were not accepted. Both women and men viewed CEO, vice presidential, and mid-level management roles within organizations as positive and possible when the organization had a feminine image, such as clothing manufacturing (Killeen, LópezZafra, & Eagly, 2006). However, for organizations with a more masculine image (e.g., auto manufacturing), women saw these roles as positive, but not possible; whereas men saw the roles as both positive and possible, regardless of the organizational image. A contextualization of the aspirations of women appears to be occurring. This indicates that differences may be found for gender when predicting attitudes, such as motivation to lead, from PO fit.
45 Nwadei (2003) found relationships between values congruence and organizational commitment based on different values for socio-cultural groups. Bottom-line values congruence, on values such as health and safety, predicted commitment for Africans, whereas change values congruence, on values such as openness, growth, innovation, and flexibility, predicted commitment for Americans. For Europeans, people-based values congruence made a difference, whereas in the Middle East, ethical congruence predicted commitment. This indicates that differences may be found for socio-cultural groups when predicting attitudes from PO fit, such as motivation to lead. Numerous other situational and individual characteristics that affect PO fit have been explored. These include self-efficacy, personal control, past work experience, openness to influence, ethnicity (e.g., PO fit had a smaller effect for African Americans and fit was lower), organizational culture strength (i.e., tightness-looseness), organizational support, leader-member exchange quality, and burnout (Erdogan, Kraimer, & Liden, 2004; Gelfand, Nishii, & Raver, 2006; Kristof-Brown et al., 2005; Siegall & McDonald, 2004). It appears that individual characteristics, such as gender and sociocultural group, serve to increase or decrease the salience of PO fit to magnify or diminish its effect. Others factors, such as organizational support and tightness-looseness, change PO fit based on the situation. It is possible that the relationship between needs and values is circular, or at least multi-directional. Situation may reprioritize needs, which then necessitates a change in individual values priorities. Ethical fit. Pierce and Snyder (in press) term ethical fit as compatibility in ethical values and behavior. Ethical fit can be based on organizational norms that are ethical or unethical.
46 Pierce and Snyder point out that when ethical fit is defined by organizational norms that include illegal behavior, the outcome can be very serious. Using behavioral data from vehicle inspection stations to determine ethical fit, they found that ethical diversity mitigated attrition due to ethical misfit. By measuring directional misfit they found that unethical employees left ethical organizations, and ethical employees left unethical organizations. Pierce and Snyder sensed that this effect would be stronger for ethical employees, but they did not predict or test this. Their evidence suggests that vehicle emissions testing is a market where ethics is unprofitable, meaning that some organizations may suffer financially by hiring ethical employees when competitors do not. This situation may be analogous to an arms race. Organizations may race to the bottom to match the unethical behavior of their competition as a matter of survival. Pierce and Snyder suggest monitoring and fining for unethical behavior in the marketplace to counterbalance this impulse. Ethical fit has been found to predict affective and continuance commitment (Sims & Kroeck, 1994). Another more recent study (Ambrose, Arnaud, & Schminke, 2008) also found that ethical fit (how well ethical climate matched individual moral development) predicted higher levels of organizational commitment. In a study of 314 employees at 128 organizations, Sims and Keon (1997) used multiple regression analysis and found that fit on individual business ethics and the organization’s ethical climate predicted lower intent to leave. However, they did note that commitment or satisfaction could have also contributed to lower intent to leave. Later, Valentine, Godkin, and Lucero (2002) found that corporate ethical climate itself predicted fit on values, as well as predicting commitment. Coldwell, Billsberry, van Meurs, and Marsh (2008) give an explanation for
47 these results. Corporate social responsibility (CSR) has become increasingly important to the public and to employees. CSR and corporate reputation are linked, so ethically oriented employees may be attracted and retained due to CSR. In fact, they note that it has been shown that people would rather work for an ethical company for less pay, and when employees observe ethical behavior by management, they are more satisfied. Coldwell et al. also suggest that ethical fit could be an issue for retention when the public face of the corporation does not match internal reality. Coldwell et al. (2008) proposed that when misfit occurs between the organization’s moral stage and the individual’s, negative attitudes and behavior can result. This refers to Kohlberg’s stages of moral development: (a) post-conventional - a level never attained by most adults, with social mutuality and genuine interest in welfare of others, respect for universal principles, and the demands of individual conscience; (b) conventional - where approval of others is paramount; and (c) pre-conventional - a level reached by most at primary school, where obedience and punishment guide morality centered on law and order. Ambivalence may also occur instead if the degree of misfit is minor. The actual impact of ethical fit on organizational performance is still unknown. Peterson (2004) posited that corporate social performance could influence stakeholder groups, in addition to financial performance. Based on social identity theory, he found that, when employees believed social responsibility to be important, economic, legal, and ethical corporate citizenship predicted commitment, with ethical corporate citizenship being the best predictor. It is possible that lower ethical fit relates to lower motivation to
48 lead, which in itself is not a negative attitude, but may be an attitude that negatively impacts the organization by limiting leadership resources. Moral Philosophy and the Motivation to Lead PO fit and moral philosophy. PO fit is known to predict commitment, an attitude on which the motivation to lead construct is based. In the present study PO fit is treated as an affective construct that reflects perceptions of fit on values, or how what is important to the individual is perceived as similar to what is important to the organization. However, as discussed previously, PO fit can change for an individual. This could occur due to changes in the organization. Additionally, the individual values on which PO fit is based can change (Kristof-Brown & Jansen, 2007; Puente, 2004). It is possible that some individuals are more willing to adjust their values or value priorities to meet the conditions of the organization in which they are embedded. An individual’s ethical processing system, also known as a moral philosophy, could be an individual difference that influences values change. Further, openness to values adjustment may influence the relationship between level of PO fit and an attitude such as motivation to lead, even if the need for values change is anticipatory rather than immediate. Moral philosophy overview. While morality is a set of beliefs about what is right or wrong, ethics is a conscious reflection on the adequacy of these beliefs (Dodig-Crnkovic, 2007). A moral philosophy describes the process of how ethicality is decided, rather than morals themselves (Dodig-Crnkovic). Schlenker and Forsyth (1977) developed a widely used
49 model of moral philosophy that is explored in the present study in relation to motivation to lead. There are a variety of ethical bases for moral philosophies. Schlenker and Forsyth (1977) chose questions around teleology (also known as utilitarianism), deontology, and skepticism to explore factors that could be used to measure moral philosophy. Teleology minimizes self-interest to maximize utility using a cost to risk ratio. In teleology, intrinsic values (pleasure, happiness, ideals, preferences, self-realization, and fulfillment) are considered most important (Schlenker & Forsyth). Teleology has been criticized because a person cannot be responsible for all consequences, as they cannot be foreseen. Further, putting aside self-interest could include putting aside personal integrity. It is also unclear as to who should be in the domain of concern (Dodig-Crnkovic, 2007). Finally, luck contributes to consequences, making it even more difficult to predict outcomes. Deontology rejects the consequences of rules or actions as a basis for moral evaluation (Schlenker & Forsyth, 1977). Deontology grounds decisions on rules and universal laws of humanity (Dodig-Crnkovic, 2007). This philosophy appeals to natural law and rationality to determine ethical judgments. Acts are judged as moral by comparing them with universal moral rules, and there are no exceptions, regardless of consequences. For example, it would be immoral to lie, even for benign motives. Like teleology, deontology has difficulties with calculating and balancing rewards and risks because future consequences are unknown (Schlenker & Forsyth). Contrary to deontology, ethical skepticism, with many moral points of view (e.g., emotivism, cultural relativism, and ethical egoism), holds that inviolate moral codes cannot be formulated (Schlenker & Forsyth, 1977). For example, emotivism says that a
50 person cannot decide what is moral unless they can see, touch, hear, or otherwise sense its meaning. Cultural relativism ties morality to society. Egoism holds that there are no moral standards, except in reference to what one feels is right, and further, everyone acts to promote their own self-interest. Egoism, like teleology, considers consequences, but only for the self. The initial version of Schlenker and Forsyth’s (1977) Ethics Position Questionnaire (EPQ) used 50 questions that tapped the common major dimensions of ethical concern for teleology, deontology, and skepticism: (a) importance of consequences, (b) consideration of consequences, and (c) feasibility of universal moral codes. Two major distinctions were found among the moral philosophies: (a) relativism, which is the extent to which one is willing to accept the existence of a universal moral code, and (b) idealism, which is an endorsement of idealistic versus pragmatic beliefs and actions. These two orthogonal dimensions represent individual differences that influence actions, judgments, and emotions when dealing with moral issues (Forsyth, O’Boyle, & McDaniel, 2008; Park, 2005). Orienting the two dimensions around the original ethical bases helps to explain their meaning. On the high end of the continuum of relativism, skeptics deny the existence of universal ethical rules. On the low end, a deontologist would condemn an act that fails to meet a rule, regardless of the amount of harm or benefit. Somewhere between skeptics and deontologists on the relativism continuum, teleologists tolerate negative consequences to the degree that positive consequences outweigh them, so they are more pragmatic than idealistic. Skeptics are guided by consequences, but some may judge consequences as idealists, where others would be more pragmatic. However, Schlenker
51 and Forsyth (1977) found that most skeptics are pragmatic. In either case, relativists differ from both teleologists and deontologists by denying the applicability of universal moral rules under any circumstances. Along the continuum of idealism, an underlying calculation process is used to weigh decisions. However, at highest end, the idealist is far more concerned with costs than benefits, whereas the pragmatist considers both. Finally, deontology is most closely related to universalism (i.e., low relativism) and idealism in Schlenker and Forsyth’s model. Schlenker and Forsyth (1977) noted that science can provide answers to questions concerned with the means used to obtain or implement particular values and goals, and the consequences and affect of their implementation. The question of whether a value or goal is moral is not a scientific question, but rather, morality is determined by moral philosophy. Although Schlenker and Forsyth’s initial work grew out of an effort to analyze ethics codes used in social science research, the resulting theory has been used extensively in business research, especially in the area of business ethics (Forsyth et al., 2008). The EPQ can be viewed as a four-way classification of relativism and idealism, or as two orthogonal dimensions. Davis, Andersen, and Curtis (2001) found discriminant validity for idealism and relativism. They also found that idealism is stable for age and gender, whereas relativism is not. In the present study, the role that universalism might play in moderating the relationship between PO fit and motivation to lead is explored. It is possible that individuals who make decisions based on strict moral rules might find low PO fit to be more salient in this situation, whether they are pragmatic or idealistic. It seems plausible that an intrapersonal conflict concerning immutable rules would override
52 any influence of pragmatism. In addition, universalists have been shown to experience lower self-esteem when they succeed, whether the goal is selfish or selfless (Forsyth, 1992). This may depress the motivation to lead. As idealism is more stable than relativism, it is considered a direct predictor of motivation to lead in the present study. Whether based on universal rules or not, the tendency for some idealists to almost exclusively calculate costs, with less consideration for benefits, drives the view in the present study that idealism is a potential negative correlate of motivation to lead. PO fit, ethical conflict, and relativism. In the present study, PO fit describes the extent to which individual and organizational values are perceived by the individual to be similar. An individual’s decision-making process involves value judgments (Liedtka, 1989). These judgments are produced by assessing the fit between the course of action proposed by the organization (organizational values) and the individual’s self-image (personal values). Liedtka observed that conflict occurred when individuals were unsure as to whether organizational expectations were consistent with their personal values. Conflict between personal values and the values held by the organization produces ethical conflict (Toffler, 1986). Perceptions of ethical conflict have been shown to be based on comparisons between personal values and the perceived values of direct management (Schwepker, Ferrell, & Ingram, 1997; Soutar, McNeil, & Molster, 1994). However, the behavior of top management is also considered by individuals (Soutar et al.). The influence of direct management is especially strong when the ethical code of the organization is unclear. Values are not always explicitly stated to employees. In fact, Kristof-Brown et al. (2005)
53 proposed that actual and perceived fit on ethical values might be distally related for this reason. It appears that when the beliefs of top management are unclear, the values of direct management have the greatest influence on the individual, and individual perceptions of differences produce ethical conflict. As values influence the process of determining what is ethical, PO fit could also be said to describe how well the individual perceives that, when making ethical decisions, consulting their values will produce a result similar to relying on the organization’s values. When personal and organizational values are incongruent due to low PO fit, conflict may occur. However, relativism may determine how this conflict is handled, or whether conflict is experienced at all. Ethical conflict can occur for employees at any hierarchical level (Peterson, 2003). Further, employees who do not agree with the organization’s values, and who feel pressured to compromise their own, may experience cognitive dissonance. This scenario may be common, as employees almost always see themselves as more ethical than their co-workers, supervisors, and top management (Brenner & Molander, 1977). Ethical conflict occurs when employees feel pressured by their peers and management to compromise their personal values in order to achieve organizational goals (Leicht & Fennell, 1997). Employees have also been found to experience pressure to go against formal organizational standards that they see as ethical (Goodell, 1994). This probably reflects the influence of the informal organizational standards described by Quinn, Reed, Browne, and Wesley (1997). In addition, a large majority (70%) of managers at all levels were found to feel pressured to conform to ethical norms of their organizations with which they disagreed (Posner & Schmidt, 1984). Upper managers and entrepreneurs were also found to feel pressure to make business decisions that conflicted with their personal
54 moral values (Longenecker, McKinney, & Moore, 1988). Values compromise is not limited to the private sector, as it was found to occur in the public sector by Bowman (1976). Treviño, Weaver, and Reynolds (2006) proposed that professionals might become morally compromised gradually over time. Individuals carve out private “identity spaces” (or niches) and situationally defined organizational identifies (Weaver, 2006). As their organizational identities become incorporated with the organization, the moral content of this niche may become different from their individual identity. Differences in these identities would probably vary given the norms of the organization. In addition, the organization could be normless. Anomie (defined as a lack of purpose, identity, or values in a person or society) can lead to a breakdown of the norms that rule the conduct of people and assure the social order (Kuczmarski & Kuczmarski, 1995). This results in a loss of meaning and a sense of injustice, and can affect moral thinking. Tsahuridu’s (2006) findings showed that individuals view the work context as more normless than the world outside of work. The process of values incorporation may be more apparent to low relativists, as the content of the organizational niche becomes more different from the inviolate rules that are part of their personal identity, and this may increase the salience of low PO fit. For example, Tsai and Shih (2005) suggested that relativists are more likely to excuse an unethical decision, and therefore experience less role conflict. Ethical conflict, relativism, and the motivation to lead. A review of the literature did not reveal any theories or empirical research directly relating relativism and motivation to lead. However, it was found that when organizational values were considered ethical, higher organizational commitment resulted
55 (Herndon, Fraedrich, & Yeh, 2001). Further, Schwepker (1999) found that when a personal and organizational values mismatch was experienced by an individual, personal ethical conflict resulted. This ethical conflict produced lower organizational commitment. It has been shown that managers believe their jobs require them to compromise their ethics (Moser, 1988). Moser points out that if this were not so, a code of ethical conduct would be unnecessary. The tension created by the incongruence of what an individual acting alone would do, versus actions as an agent of the organization, is the source of ethical conflict (Fasching, 1981). As a coping mechanism, complete detachment from ethical concerns and personal responsibility may result (Moser). To eliminate or reduce ethical conflict, individuals may withdraw or resign. More subtle effects of ethical conflict include whistle-blowing, poor morale, disloyalty, strained personal relationships, uncooperativeness, reduced quality, and absenteeism, all of which lead to lower productivity (Moser). Prior research concerning the relationship between ethical conflict and outcomes has produced mixed results (Peterson, 2003). Peterson sought to uncover the cause of these inconclusive findings by examining possible moderators. Using regression analysis with 161 responses, Peterson found that lower commitment and higher intention to leave were each predicted from ethical conflict, over and above age, gender, and educational level. Peterson then used moderated regression analysis to examine relativism as a moderator of the relationship between ethical conflict and commitment, and as a moderator of the relationship between ethical conflict and intention to leave. He chose relativism because it influences the ethical decision-making process, specifically when formulating an intention to act. As relativists would be more likely to consider the
56 situation when faced with an ethical dilemma, Peterson hypothesized that they would be better able to cope with the pressure to engage in unethical behavior in an organization. Peterson found a strong negative relationship between ethical conflict and organizational commitment. However, when relativism was high, this relationship was no longer present. When relativism was low and ethical conflict was high, organizational commitment decreased. The effect on intention to leave was small, and Peterson proposed any number of reasons for this. For example, even if someone is conflicted, they may not have the option to leave their job due to monetary concerns. As the relationship with commitment was especially strong for moral universalists, Peterson suggested that they might experience much more stress when ethical conflict arises. No interaction was found between ethical conflict and relativism for intention to leave. Peterson notes that interaction detection in the field is known to be difficult, and that this limitation might explain the absence of this interaction in his findings. Relativists have been shown to be less ethically sensitive (Chan & Leung, 2006; Sparks & Hunt, 1998). For example, in a study of 151 buying professionals at 52 companies, Park (2005) used hierarchical multiple regression and found that relativists were less likely to consider socially responsible behavior, which could affect their intentions when making an ethical decision. Park suggested that although the study was limited due to a low response rate of 18.4%, this rate is comparable to other surveys on business ethics and social responsibility. Jackall (1988) studied the nature of moral behavior in organizations. He found that the rules for success in an organization form a bureaucratic ethic that necessitates separating personal morality from that of the organization. He further argued that
57 personal ethics might be sublimated to get ahead in the organization. As Senge (1990) pointed out, “Only [when the organization fosters values in alignment with peoples' own core] will it be possible for managers to stop living by two codes of behavior, and start being one person" (p. 312). The sublimation of personal values seems related to Goodpaster’s (2004) description of how teleopathy, or the unbalanced pursuit of organizational purpose, is characterized by fixation, rationalization, and detachment. And as commitment to those higher in the organization increases, the need to sublimate personal moral codes increases. Ashforth and Vaidyanath (2002) likened this process to experiencing faith in the organization as a secular religion, with normative controls instilling a shared moral code. Jackall further describes how managers' moral compromises preserve the organizational culture: As it happens, given their pivotal institutional role in our epoch, they help create and re-create, as one unintended consequence of their personal striving, a society where morality becomes indistinguishable from the quest for one's own survival and advantage. (p. 204) To advance in an organization, an individual must assimilate its rules (Quinn et al., 1997). As Jackall (1988) observed, as well as Ford and Richardson (1994), when one shifts into the management structure of a large organization more is involved than a simple change in job description. The management context is a social and cultural environment, with rules of behavior that differ from society at large (Quinn et al.). These rules of behavior are generally unwritten and sometimes communicated using oral tradition, and the new manager must be able to determine and assimilate these using observation and discussion (Quinn et al.). This assimilation is required for advancement in the bureaucratic hierarchy (O’Neil & Pienta, 1994). The ethical rules for advancement are external to the manager. Those who hold moral concerns that conflict with what the
58 group collectively agrees to may be considered troublemakers, creating more pressure to conform to move up (Quinn et al.). The assimilation process required for advancement may be more difficult for low relativists who refuse to bend ethical rules. In contrast, the moral anchor of the relativist is more likely to be attached to the anchor of the organization’s culture (Quinn et al.). For this reason, the low relativist’s strict adherence to a moral code could impact motivation to lead negatively. For example, describing a foray into a management position, one individual said: I do think it’s important to have some principles and that kind of thing. I actually feel good that I’ve been able to hang on to those. When you go through all this, especially if you try management. I’ll tell you the things… remarkable… people don’t realize. (Papavero, 1999, p. 57) It is possible that, in addition to lower commitment to the organization, a universalist may be less likely to imitate successful others who do not share their moral code. This could also impact their motivation to lead. Moore (2008) proposed that moral disengagement fosters organizational corruption by rewarding decisions that advance organizational goals, whether or not these decisions are ethical, and thereby dampening individual moral awareness. A talent in prioritizing organizational goals above all else has been shown to be a top leadership skill, especially valued in times of crisis or uncertainty (Bligh, Kohles, & Meindl, 2004). It seems reasonable to expect that possession of this skill would affect advancement. For example, Scott Sullivan advanced quickly at WorldCom in part because of his willingness to misrepresent financial statements (Jeter, 2003). Andrew Fastow was advanced by the leadership at Enron in part because it was understood that he would do “whatever it took” to make Enron’s numbers (Mclean & Elkind, 2003). It is possible that
59 some organizations reward those most willing to collude in corrupt practices (Eichenwald, 1995). Ethics and advancement are rarely studied (Moore, 2008). The structures and processes that support organizational survival and growth could influence unethical behavior without direct intention on the part of the organization. Dominant groups align their interests with the corporation (Cyert & March, 2002; Thompson, 1967). Organizational norms will reflect the norms of these groups, even when they sanction corrupt behavior (Moore). It is possible that strong performers who are less sensitive to ethical issues advance more quickly into leadership positions (Moore). These same individuals will then create a climate which models, rewards, or further embeds corrupt practices into the social structure (Moore). This could create strong situation pressures that cause perpetuation of corrupt actions throughout the organization (Sims & Brinkman, 2002). Moore proposes that individuals higher in moral disengagement will advance more quickly through the organizational hierarchy than those low in moral disengagement. It is possible that relativism makes it easier to sublimate personal ethics, which could lead unintentionally to moral disengagement. For example, Cynthia Cooper, a whistle-blower at WorldCom stated in retrospect that she was different than her colleagues because she “refused to overlook actions that were contrary to her principles. When evaluating her priorities, she would not succumb to the pressures placed by superior figures” (Kumar, 2007, p. 5). Ms. Cooper appears to have been able to maintain her moral congruence in this case by whistle blowing. Moral congruence is defined as the condition and process of
60 achieving consistency between self-selected moral values and the manifest behaviors of the individual. In a qualitative study, Rodriquez (2005) found that morally congruent managers felt loyal to their internal convictions and had an internal locus of control. These managers viewed being morally congruent as a life-long process of discovery, sense making, alignment, critical self-reflection, and self-correction. They saw incongruence as the door to congruence, such that a loss of inner peace was a wakeup call to change, and they believed that leaders must be morally congruent to generate congruence in others. The cognitive dissonance produced by ethical conflict may be interpreted in a far different way than as a call for personal growth and development. To eliminate the cause of cognitive dissonance, the individual may leave the organization as a way to reduce associated stress. Further, avoiding leadership roles could be considered as a parallel to leaving the organization, insofar as this avoidance could be seen as a coping mechanism. The ethical reasoning process could be influenced by moral philosophy. Ethical reasoning is thought to occur in steps that include: (a) identifying the dilemma, (b) developing an ideal solution, (c) formulating an intention to act, and (d) ethical action (Peterson, 2003). Relativism may influence the formulation of intention to act based on potential outcomes. The relativist may rationalize intentions to act in a way that is ethically acceptable based on the situation: Organizational goals must be achieved at any cost. This rationalization could also result in an increase of their perception of PO fit, removing any influence that low PO fit may have had on motivation to lead. This adaptation process may take place on a regular basis in many daily scenarios of potential ethical conflict in order to decrease the discomfort caused by cognitive dissonance. Further, the instances of adaptation
61 increase with hierarchical level, and those cognizant of this may experience lower motivation to lead. Again, the low motivation to lead that may result from poor PO fit could be likened to the intention to leave, where the hierarchical level is abandoned rather than the organization. In an exploration of the individual decision to refuse an advancement offer, it was found that anticipation of compromising values was a major factor in the decision to turn down a promotion (Papavero, 1999). Relativists may rationalize that ethics at work do not have to match ethics in personal life, as what is ethical changes with the situation. Idealism and the motivation to lead. Aggressiveness, materialism, high achievement motivation, and traditional sex role divisions are antiethical to the “person-centered, humble, nurturing, and interpersonally sensitive orientation of high idealism” (Cui, Mitchell, Schlegelmilch, & Cornwell, 2005, p. 26). This seems to indicate that idealism is negatively related to motivation to lead. Idealists have been found to be more ethically sensitive than those low in idealism (Bass, Barnett, & Brown, 1999; Chan & Leung, 2006). Idealism has been found to be positively related to ethical perceptions, judgments, intent, and behavior (Shaub et al., 1993). Idealists are less likely to engage in organizational and interpersonal deviance (Henle, Giacalone, & Jurkiewicz, 2005), and, in China, they were more likely to report the unethical behavior of peers (Chiu & Erdener, 2003). The high ethical sensitivity of idealists may occur because they place a greater importance on ethics and social responsibility than those low in idealism (Tansey, Brown, Hyman, & Dawson, 1994).
62 The ethical sensitivity of idealists may be attributable in part to their almost singular focus on costs for none when considering consequences (Forsyth, 1992). Idealists have been found to be low in Machiavellianism, where manipulative, persuasive, and deceitful behavior is used to achieve goals (Bass et al., 1999; Leary, Knight, & Barnes, 1986). However, idealists are willing to be disloyal and they are more likely to lie or engage in an immoral act if they perceive that human welfare will benefit (Byers & Powers, 1997; Forsyth). On the other hand, Forsyth and Schlenker (1977) found that idealists see obedience as positive behavior. This gives one explanation as to why Chonko, Wotruba, and Loe (2003) discovered that idealists find ethics codes more useful than pragmatists. This could also explain why idealists have been found to experience higher levels of intrapersonal role conflict. It could be difficult to be simultaneously obedient and cause no harm to others. Idealists are more likely to experience incompatible expectations, especially when organizational values are not clear (Sims & Keon, 2000; Tsai & Shih, 2005). In addition, due to low pragmatism, idealists can become divorced from practice. Organization-professional conflict can occur when professionals are forced to focus on profit rather than professional goals, or when organizational demands diverge from accepted professional behavior, especially when these demands are unethical. In a cross-sectional study of 319 accountants at various organizations, Shafer et al. (2002b) used structural equation modeling to examine the relationships among professionalism, organization-professional conflict, and organizational commitment. Organizationprofessional conflict was found to have a negative relationship with organizational commitment. As the idealist is more committed to their profession, they may be more
63 likely to experience organization-professional conflict. They may therefore be less committed to the organization. The association of idealism with organization-professional conflict, and further, the relationship of this conflict with lower organizational commitment, suggests that highly idealistic individuals may be reticent to take a leadership role. Further, idealists may be harder on themselves in regards to failure (Forsyth, 1992), which may lessen their attraction to leadership situations that they may view as more risky. With their concern for costs and protecting the general welfare, they may avoid leadership roles that require decisions that harm other employees. This may be especially true if, given the idealist’s higher ethical sensitivity, they are more likely to perceive unfair or unethical behavior and sense that they would be required to impose the consequences of this behavior on their subordinates. For example, when discussing the decision to leave a leadership position, an individual stated: You’re affecting people’s lives and their families with these things. That’s what I found offensive. You don’t really have the power to make things better for them but you have the responsibility for being the one that hits them with it, whether it’s a bad salary or whether it’s a layoff. (Papavero, 1999, p. 58) Conceptualizing and Studying Motivation to Lead Where leadership emergence describes the “what” of people who lead (that is, their individual characteristics), motivation to lead answers the question of why they want to lead. Before Chan’s (1999) proposal of the motivation to lead construct, some work was done in this area, most notably that of House and Singh (1987). They proposed three psychodynamic attributes of people who are motivated to lead: (a) high power motive, (b) high activity inhibition, and (c) low affiliation need. Although power and control may be important motivators for some, this cannot be assumed to be true for all
64 leaders. The individual differences of leaders seem a logical avenue to explore to discover more about the diverse and complex motives that drive their attraction to leading. Theoretical model of the motivation to lead. Chan (1999) recognized that a theoretical framework was needed to link individual differences and various leadership behaviors. He noted that prior research discredited the role of individual differences. Later, individual differences came back to the fore, but much of the work used bivariate correlations rather than multivariate models. Chan also agreed with Lord and Hall (1992) by identifying a criteria problem affecting research in this area, in that leader perception, leader emergence, and leader effectiveness were often treated as equivalent. Chan’s (1999) intent was to differentiate leader emergence and performance. He also suggested that the research focus should move in a different direction regarding individual differences. Rather than measuring direct relationships between individual differences and performance, we should consider that “non-cognitive constructs such as personality and values may be linked to leadership performance through the process of leadership development” (Chan, p. 86). It is important to recognize that Chan did not suggest that motivation to lead predicts leader effectiveness. However, he did conjecture that motivation to lead might relate to leader effectiveness indirectly by predicting morale and job satisfaction. Chan (1999) defines motivation to lead in terms of a definition of motivation where internal processes determine direction (the decision to lead), intensity (effort given to leading), and persistence (leading during adversity). The motivation to lead construct
65 represents individual differences that can affect these leadership behaviors. Individual differences included in motivation to lead are considered relatively stable. However, motivation to lead could interact with external factors such as domain and task. For example, motivation to lead may change if one takes part in leadership training. Further, motivation to lead integrates leader development and leader performance by including past leadership experience in the framework. Learned knowledge and skills from leadership experiences are also antecedents of motivation to lead. Leadership experiences cause one to seek out more training and further development occurs. A feedback loop is created where each experience interacts with motivation to lead to create different performance outcomes. The motivation to lead construct. Chan (1999) based the dimensions of motivation to lead on Meyer and Allen’s (1991) model of organizational commitment. Meyer and Allen identified a multidimensional construct of commitment with affective, normative, and calculative types. The sources of each type of commitment are different: affective commitment is sourced in a need for achievement, calculative commitment in job investment, and socialnormative commitment in socialization in the organization. Affective commitment is thought to be related to intrinsic motivation and relational psychological contracts, whereas social-normative commitment is thought to be related to extrinsic motivation, and both calculative and social-normative commitment are thought to be related to transactional psychological contracts (Meyer, Stanley, Herscovitch, & Topolnytsky, 2002).
66 Chan (1999) also identified conceptual similarities between the commitment model and two major social behavior theories: Fishbein and Ajzen’s (1975) theory of reasoned action (TRA) and Triandis’ (1980) theory of interpersonal behavior (TIB). TRA sees intent to act as based on individual attitude regarding outcome valence and perceived social norms. Similarly, TIB explains behavior through four constructs: cognition, affect, social norms, and personal norms. The common components of these models (attitude/affect, cognition, norms) were then mapped to three possible dimensions of motivation to lead: people like to lead (affective-identity motivation to lead), people make a rational decision to lead (calculative or instrumental motivation to lead), and people feel it is their duty to lead (social-normative motivation to lead). Chan reasoned that it made more sense to predict a strong non-calculative dimension in those with a motivation to lead because there is often a cost associated with leading. It is important to clarify the meaning of non-calculative motivation to lead. Although Chan’s (1999) description could lead one to believe that this dimension indicates the level to which someone disregards the cost of leadership, Hiller (2005) pointed out that the antecedents to non-calculative motivation to lead actually heavily emphasize the choice to lead as a selfish one based on rewards and benefits. For example, a person high in non-calculative motivation to lead would give a low rating to this item: “I would only agree to be a group leader if I know I can benefit from that role.” A person low in non-calculative motivation to lead would weigh individual costs and benefits, and lead only if there were a net benefit. They would consider all types of costs, including non-economic ones. However, there may be individual differences in awareness of costs and the weights assigned to them.
67 Someone higher in non-calculative motivation to lead would lead even if there were no net benefit. However, this does not mean that those higher in non-calculative motivation to lead do not consider the costs. To paraphrase Hiller (2005), a person higher in non-calculative motivation to lead is not necessarily someone who simply is not aware of the costs of leadership. Rather, an individual higher in non-calculative motivation to lead would disregard the costs, even if they were aware of them. Non-calculative motivation to lead will be looked at closely in the dissertation study, as Chan, Ong, and Chah (1999) suggested that the choice to lead is a social dilemma, where an individual must choose between their own interests and those of the collective. As such, if costs or benefits are inordinately high, an imbalance could be created in the leadership pool to the detriment of the collective. Previous research was used to identify possible antecedents to motivation to lead: general cognitive ability, personality traits, values, self-efficacy beliefs, and past leadership experience. Focus groups were then used to develop items that measured each of the three motivation to lead dimensions. The instrument resulting from Chan’s (1999) work incorporated measurements of the Big Five personality factors developed by Goldberg (1999), the Individualism-Collectivism values measure developed by Singelis, Triandis, Bhawuk, and Gelfand (1995), and the Leadership Self-Efficacy scale developed by Feasel (1999). Leadership experience was measured using biographical data and selfreports, and cognitive ability was measured using results from previously administered standardized tests. Using three samples (1,594 Singapore military recruits, 274 Singapore students, and 293 U.S. students) and hierarchical regression analysis, Chan found patterns and paths between antecedents and the motivation to lead dimensions. He cited
68 limitations in that the age range was narrow (17 to 21) and situation was not included in the model. However, the model was developed with situation in mind, and the present study seeks to extend Chan’s work by exploring a situational variable, PO fit, in relation to motivation to lead. Chan’s (1999) results showed that personality, values, and leadership experience were all related to motivation to lead, both directly and through leadership self-efficacy. Both leadership self-efficacy and past leadership experience were related to motivation to lead, suggesting that motivation to lead “is a dynamic construct that is partially changeable through social-learning processes and experience” (Chan & Drasgow, 2001, p. 496). Chan did not find that cognitive ability predicted motivation to lead. Relevant studies using Chan’s motivation to lead construct. Chan, Rounds, and Drasgow (2000) studied the relationship between vocational interests and the motivation to lead construct. Using Holland’s (1973) RIASEC model of occupational interests, motivation to lead was found to be orthogonal to occupational types. Chan et al. concluded that motivation to lead is independent of vocational interests. It is therefore not expected that participants in the present study with different occupations would vary in level of motivation to lead. Chan (2001) later conducted a two-year long longitudinal stability study (at oneyear and two-year intervals) of the motivation to lead scale and several antecedents of motivation to lead (personality, individualism/collectivism, leadership self-efficacy). This study used a subset of the Singapore military sample from his original study (Chan, 1999). The results showed that motivation to lead was stable; more stable than individualism/collectivism values, but less so than the personality measures. Motivation
69 to lead did change with increased work experience. Affective-identity motivation to lead and social-normative motivation to lead increased over time. Also, the non-calculative motivation to lead of those who became officers during the study increased. Limited support was also found for a developmental feedback loop between motivation to lead and leadership experience. Given these results, the present study will include control variables for age, organization tenure, work experience, previous leadership experience, and current job level. Chan’s (1999) motivation to lead construct has been used in several other studies. Cintrón (2004) studied the motivation to lead of Hispanic women using Chan’s motivation to lead scale, along with an acculturation scale and an emotional intelligence scale. She found that emotional intelligence and biculturalism predicted motivation to lead, which may support Chan’s finding that emotional stability is related to motivation to lead. However, no significant results were found regarding acculturation and the motivation to lead. Cerff (2006) used regression analysis with a sample of 200 university students in South Africa and found that hope and self-efficacy predicted motivation to lead. However, her study did have some limitations in that the sample was sourced in one region in Cape Town. Erickson (2005) explored the antecedents of motivation to lead across the lifespan and in relation to vocational interests using a sample of 63 leaders at a Pentagon office. Using hierarchal multiple regression, he also looked at a possible situational factor (collective efficacy, or a shared belief of a workgroup in the team’s capabilities) and its relationship to both motivation to lead and the motivation to lead antecedent of selfefficacy. His sample was older (28 to 62), more educated, and more experienced than
70 Chan’s (1999). Erickson’s results supported the internal reliability of the motivation to lead scales. His results also indicated that antecedents to motivation to lead may change over the lifespan and when occupations change. Erickson’s results also supported Chan and Drasgow’s (2001) findings that motivation to lead changes with work experience, but they contradicted Chan et al.’s (2000) findings that motivation to lead is independent of vocational interests. Finally, Erickson found that the situational factor, collective efficacy, had no effect on motivation to lead. However, he cites the limitations of his single context sample as one explanation for this result. Erickson did suggest PO fit as a possible situational factor worthy of future research, which lends further support for the present study. Studies on motivation to lead and situation. Others have studied motivation to lead and situation, but have not used Chan’s (1999) construct. Kabacoff (2002) studied the relationship between emotional drivers and leadership behaviors with a large (N = 1,300) sample of U.S. and Canadian managers. He used the Individual Directions Inventory (IDI) to measure motivational factors and the Leadership Effectiveness Analysis (LEA) to measure leadership behaviors. Citing limitations in current research, including that of Chan and Drasgow (2001), he studied a “wide range of personal motivators and leadership behavior within a broad array of organizational settings“ (p. 1). He suggests that leadership requirements are driven by context, and that emotional drivers must be matched to these requirements. For example, someone who thrives on affiliative experiences would not be a good match for a position requiring dominant, controlling behaviors. This work gives some support for the idea that situation may be related to motivation to lead.
71 Richter (2001) developed a survey measuring the correlation between encouragement, opportunity, and the success of past leadership with the motivation to move up from teaching to school administrator positions. She also used qualitative data to triangulate the quantitative survey results. Richter found that educators were motivated to move into leadership positions when they were encouraged to do so. She found a very strong relationship between familiarity with the leadership role and the motivation to lead. Richter also found strong links between both financial interests and positively affecting children’s lives, and the motivation to lead. She did not find any relationship between mentoring or past leadership success, and motivation to lead. She cited many contextual elements of the educational system structure that could be responsible for these results. The sole path to an increased salary was through an advanced degree and an administrative role. However, she questioned the financial motive, and the structure that compels it, by positing that it may not attract people who are motivated for reasons that will make them effective leaders. In an article from a newsletter of the Chronicle of Higher Education, Jacobson (2002) interviewed academics that rejected promotions. The academics reported that they believed their quality of life would be impacted adversely, and that higher pay was not worth this sacrifice. Many said they were happy with research and teaching, and were not interested in the administrative aspects of higher-level jobs. They preferred to move up in their fields, rather than climb an administrative career ladder. Whetstone (2001) contrasted police officers that sought promotions and those who did not. Officers that did not pursue promotion often cited a discrepancy between the effort required and the pay. They felt officers had more flexibility with assignments and
72 schedules, which indicates that work-life balance was also a concern. Some officers were not attracted to the duties of a sergeant. They did not want to “lose touch with the streets” (p. 155). Organizational complaints were also found. Officers not interested in promotion distrusted their managers and the selection process itself. Howard and Wilson (1982) studied the motivation to lead by contrasting studies from the 1950s and 1970s. They found that in the 1970s sample, the motivation to move up the corporate ladder was significantly reduced and expectations regarding work life were much lower. The 1970s sample was also found to be much less interested in dominating others. Howard and Wilson did find, through qualitative means, that this did not necessarily indicate a desire to follow, but rather a rejection of the leadership role and the organizational hierarchy. They attributed these differences to a problem with fit between organizational and personal values. A qualitative study of six software engineers who rejected promotions to management positions was conducted by Papavero (1999). The results indicated that the engineers were not motivated to lead for a number of reasons. They considered the costs associated with leading to be too high. These included increased emotional and time demands, and pressure to violate their principles (e.g., laying subordinates off, lying to their subordinates, and making unreasonable and unfair demands on their subordinates). This may indicate that the engineers had a low level of non-calculative motivation to lead. It may also indicate that idealism caused them to avoid affecting individuals negatively. The engineers also stated that they valued people first, whereas the organization’s values could be summed up as “success at all cost,” indicating a possible low level of PO fit and a less relative moral philosophy.
73 Several of the parallels between software engineers, educators, academics, and police officers are striking. It is notable that all of these professionals were given a choice to lead that entailed forfeiting their chosen profession. Idealism might have contributed to their reluctance to leave their profession and potentially cause harm to others. Also, because they would lose a large investment in their profession, this cost may have also been a factor in their decision, indicating lower non-calculative motivation to lead. Motivation to Lead Antecedents The literature discussed to this point argues that PO fit may affect motivation to lead. However, a more detailed analysis of how situation may affect the antecedents motivation to lead, and therefore the motivation to lead dimensions themselves, gives a more detailed view of the role of PO fit as a situational factor. Personality trait antecedents and situation. Chan (1999) found four of five personality traits (extraversion, agreeableness, conscientiousness, and emotional stability) to be direct antecedents of motivation to lead. The fifth trait, openness to experience, was related to motivation to lead though past leadership experience and leadership self-efficacy. Shin and Holland (2004) found that PO fit moderated the prediction of job performance from these traits. Although the present study concerns motivation to lead attitudes, Shin and Holland’s results suggest that this analysis has value. Extraversion was positively related to both affective-identify motivation to lead and social-normative motivation to lead through leadership self-efficacy. However, it is possible that extraversion could be affected by situation. Some confirmation of this comes from a study of extraversion, situational factors, and evolutionary principles by
74 Campbell, Simpson, Stewart, and Manning (2003). They used small work groups consisting of men who were evaluated by an attractive man or woman, or not evaluated at all. In this intrasexually competitive situation, Campbell et al. found that “more extraverted men were significantly more likely to emerge as leaders, but only in the female-evaluator condition” (p. 1556). This result indicates that extraversion is a personality trait that is used selectively. A person may choose to behave in an extraverted way and take a leadership role only when there is a perceived benefit, or norms require it. This may explain the absence of a relationship between extraversion and non-calculative motivation to lead in Chan’s (1999) study, where situation was not taken into account. Extraverts may be more likely to calculate leader costs and benefits based on the situation. Although the female evaluator situation may not be related to PO fit, these results show that situation can change the extravert’s decision to lead; that is, situation can change the demonstration of a behavior expected from a relatively stable individual difference that is an antecedent of motivation to lead. Agreeableness was positively related to non-calculative motivation to lead and social-normative motivation to lead in Chan’s (1999) study. However, situation might change an agreeable person’s calculation of costs and benefits and sense of duty. For example, agreeable individuals have been found to be more committed to their organizations (Cohen-Charash & Spector, 2001). However, if an individual perceives a low level of organizational justice, they are less committed to the organization, even if they are agreeable (Cohen-Charash & Spector). Software engineers (Papavero, 1999) and police officers (Whetstone, 2001) perceived unjust environments and exhibited low motivation to lead. An unjust environment does not necessarily indicate low PO fit;
75 however it may be related to a person’s perception of values mismatch. Therefore, PO fit could have an effect similar to low justice, and change the relationship between agreeableness and motivation to lead, giving a less positive relationship than if PO fit were not considered. Conscientiousness was positively related to social-normative motivation to lead and affective-identify motivation to lead through leadership self-efficacy. Similar to agreeableness, when in a specific situation, a conscientious person might decide that it is best not to take a leadership position if they are convinced it is impossible to do a good job. For example, one of the reasons that software engineers (Papavero, 1999) gave for declining promotions was that they believed they would have to make unreasonable demands on their people due to a lack of resources. Therefore, PO fit could change the relationship between conscientiousness and affective-identity motivation to lead. Idealism may also change this relationship, given the idealist’s strong desire to cause no harm to others. Emotional stability was positively related to non-calculative motivation to lead. It is entirely possible that situational aspects, such as high conflict or limited resources, could affect emotional stability negatively. Siegall and McDonald (2004) found that low PO fit had a strong association with burnout (emotional exhaustion; depersonalization of co-workers, customers, and administrators; and feelings of diminished personal accomplishment). In a high-stress context, an emotionally stable person might exceed some level of tolerance that causes them to become more calculative in their decision to lead, or more cognizant of costs (actual or perceived). We know that stress can affect leader performance. For example, Sissem (2004) found that positive leadership in the
76 context of adult education was reduced when leaders were placed under stress. PO fit has been shown to be negatively related to stress levels (Choi, 1998). Therefore, PO fit could contribute to stress, and alter the relationship between emotional stability and noncalculative motivation to lead, giving a less positive relationship than if PO fit were not considered. Openness to experience was positively related to affective-identify motivation to lead and social-normative motivation to lead through previous leadership experience and leadership self-efficacy. One possible situational variable that might change this is an exclusionary promotion process. If someone is open to experience, but they are low in PO fit, they may not be given the opportunity to gain leadership experience and, consequently, leadership self-efficacy. Therefore, the relationship between openness to experience and affective-identify motivation to lead and social-normative motivation to lead may diminish when low PO fit is present. Values antecedents and situation. Collectivist values, both horizontal (collective harmony and equality) and vertical (accept social hierarchies and subordinate goals to majority or authority), were positively related to non-calculative motivation to lead and social-normative motivation to lead (Chan, 1999). However, some organizational cultures may foster collectivist values by inspiring trust and feelings of membership, while others do not. PO fit has been positively related to level of trust (Tikanmaki, 2001). Therefore, PO fit may affect the relationship between collectivist values and non-calculative motivation to lead and social-normative motivation to lead.
77 Horizontal individualism (individuality and uniqueness) was negatively related to non-calculative motivation to lead and social-normative motivation to lead (Chan, 1999). Organizational culture might affect this value. A very bureaucratic organizational culture that demands consistency and unquestioning loyalty might not fit a person with these values. However, an organization that values individual contributions and is less concerned with creating a homogenous workforce might attract this type of individual to a leadership role. Therefore, PO fit on horizontal individualism may change the relationship between horizontal individualism itself and each of non-calculative and social-normative motivation to lead. Vertical individualism (achievement oriented and competitive) was positively related to all three motivation to lead dimensions (Chan, 1999). However, if achievement is not appreciated and rewarded (e.g., in an organization high in nepotism) this might not be the case. Additionally, vertical individualists working in organizational cultures with a high level of affiliation may not be attracted to leadership positions because they would have to care for their subordinates’ social needs. In other words, PO fit on vertical individualism may be a factor, as the value of vertical individualism may not be supported by the organizational culture. Finally, a related point concerning the individualism and collectivism constructs themselves is given by Ryckman and Houston (2003). They note that it may be more accurate to “conceptualize individualism and collectivism as two separate dimensions in which cultures and individuals can be classified as high or low on both dimensions” (p.135). Workers, and organizational cultures, could be collectivist and individualist to different degrees at the same time. These combinations (i.e., additional dimensions of
78 individualism/collectivism) may need to be considered in future research on motivation to lead. Leadership antecedents and situation. Leadership self-efficacy and past leadership experience were positively related to affective-identify motivation to lead and social-normative motivation to lead. The promotion process may be key to this result. If individuals are generally promoted because they share the values of existing leadership, or because they are less idealistic, others may not be given the opportunity to gain leadership experience and leadership self-efficacy. Similar to openness to experience, PO fit could affect leadership selfefficacy, and therefore affective-identify motivation to lead and social-normative motivation to lead. Summary of motivation to lead antecedents and situation. The majority of the personality and values antecedents of the motivation to lead dimensions have the potential to be impacted by PO fit and idealism. In general, lower PO fit and higher idealism are each expected to predict lower levels for each motivation to lead dimension. However, the antecedents of the affective-identity dimension are mainly personality differences that are more stable and may be less likely to change. Summary of Literature Review This review reveals that employees are self-selecting away from leadership positions. This may be occurring for a variety of reasons, but a central theme appears to be values incongruence and idealism. Exploring this phenomenon is important because it may be contributing to a situation where the values of leaders may be less diverse than is desirable. For example, in a study of women and leadership, Billing and Alvesson (1989)
79 point out that “organizational selection and socialization processes seem to lead to a mainstreaming of candidates where proposed beneficial women-specific attributes are lost” (p. 16). We may have converged on this state because those with values that match the organization (or those whose values do not match, but are willing to change) may be more likely to be motivated to lead. The identification of those who are avoiding leadership in their current organization, but who possess leadership ability, could bring rich information, generating positive changes in the organization. This information may identify other situations that could be affecting their decisions. Also, those outside the dominant culture may be better able to question and extend the organization’s values to make positive change. It may be beneficial to find ways to encourage their contributions. With a diversity of values in leadership, we would be better able to work together to balance moral strengths and weaknesses in each other and create more ethically resilient organizations. This study takes a first step by identifying additional factors that may be related to motivation to lead.
80 Chapter 3: Methodology Overview The two main objectives of this research were to determine if PO fit predicts an individual’s motivation to lead, and to determine if idealism predicts motivation to lead. A secondary purpose was to determine if PO fit predicts motivation to lead when moderated by moral relativism. The answers to these questions are significant because they may indicate that certain individuals are not assuming leadership roles and contributing fully to their organizations. This may result in less diversity in leadership values than is desirable. Restatement of Hypotheses H1: Lower levels of PO fit will predict lower levels of general motivation to lead, over and above personal, job, and organization characteristics. H2a: PO fit will not be associated with affective-identity motivation to lead. H2b: Lower levels of PO fit will predict lower levels of affective-identity motivation to lead, over and above personal, job, and organization characteristics. H3: Lower levels of PO fit will predict lower levels of non-calculative motivation to lead, over and above personal, job, and organization characteristics. H4: Lower levels of PO fit will predict lower levels of social-normative motivation to lead, over and above personal, job, and organization characteristics. H5: Lower levels of PO fit will predict lower levels of general motivation to lead when relativism is low, over and above personal, job, and organization characteristics. However, lower levels of PO fit will not predict lower levels of general motivation to lead when relativism is high.
81 H6: Lower levels of PO fit will predict lower levels of affective-identity motivation to lead when relativism is low, over and above personal, job, and organization characteristics. However, lower levels of PO fit will not predict lower levels of affective-identity motivation to lead when relativism is high. H7: Lower levels of PO fit will predict lower levels of non-calculative motivation to lead when relativism is low, over and above personal, job, and organization characteristics. However, lower levels of PO fit will not predict lower levels of non-calculative motivation to lead when relativism is high. H8: Lower levels of PO fit will predict lower levels of social-normative motivation to lead when relativism is low, over and above personal, job, and organization characteristics. However, lower levels of PO fit will not predict lower levels of social-normative motivation to lead when relativism is high. H9: Higher levels of idealism will predict lower levels general motivation to lead, over and above personal, job, and organization characteristics. H10a: Idealism will not be associated with affective-identity motivation to lead. H10b: Higher levels of idealism will predict lower levels of affective-identity motivation to lead, over and above personal, job, and organization characteristics. H11: Higher levels of idealism will predict lower levels of non-calculative motivation to lead, over and above personal, job, and organization characteristics. H12: Higher levels of idealism will predict lower levels of social-normative motivation to lead, over and above personal, job, and organization characteristics.
82 Research Design The present study used a quantitative non-experimental predictive design. The design allowed the prediction of motivation to lead from PO fit, idealism, and relativism over and above control variables. Using a predictive design is also consistent with previous research that assessed the prediction of commitment from PO fit (Cable & Judge, 1996; Chatman, 1991; O’Reilly et al., 1991). An earlier study also predicted motivation to lead from personality, values, leadership, self-efficacy, and past experience (Chan & Drasgow, 2001). One survey was used to collect all data. In addition to the predictor and criterion variables, the data collected included several descriptive variables concerning personal, job-related, and organization-related characteristics, which were used as control variables: (a) age, (b) gender, (c) ethnicity, (d) educational background, (e) number of years of work experience, (f) number of years of leadership experience, (g) job level, (h) number of years in position, (i) employment status (full-time or part-time), (j) number of employees in organization, and (k) number of years at organization. These variables were used as controls to predict motivation to lead over and above personal, job, and organization characteristics. Operational Definition of Variables The variables in this study were all measured at the individual level of analysis. Three predictor variables (PO fit, relativism, and idealism) and three criterion variables (affective-identity motivation to lead, non-calculative motivation to lead, and socialnormative motivation to lead) were measured.
83 PO fit is defined as the values similarity between individuals and their organizations, measured by asking for the individual’s subjective assessment. Relativism is a moral philosophy where ethics change given the situation. Idealism is a moral philosophy that strongly prefers to find solutions that do not harm others. Both relativism and idealism were measured using Forsyth’s (1980) Ethics Position Questionnaire. Affective-identity motivation to lead is seeing one’s self as a leader, noncalculative motivation to lead is not including the cost of leadership in the decision to lead, and social-normative motivation to lead is leading to benefit the group. All motivation to lead variables were measured directly using Chan’s (1999) motivation to lead instrument. General motivation to lead was not measured directly as it was derived from the three first-order motivation to lead dimensions. Instrumentation PO fit was measured with three questions created by Cable and DeRue (2002). The three PO fit items are shown in Appendix A. The response scale ranged from 1 (strongly disagree) to 7 (strongly agree). Scores for the three items were summed (giving a range of 3 to 21) to obtain the degree to which participants perceived that their values matched that of the organization and the organization’s employees. Cable and DeRue found predictive validity by comparing results from these questions to those from a measure of congruence of reported individual and organizational values. Cable and DeRue found a coefficient alpha reliability of .92 in their multi-organization sample, as compared to alpha reliability of .91 in the present study.
84 Moral philosophy was measured using Forsyth’s (1980) Ethics Position Questionnaire (EPQ). The EPQ measures two orthogonal dimensions of moral philosophy, idealism and relativism, using 10 items per dimension. The 10 relativism items and 10 idealism items are shown in Appendix A. Each item was rated using a scale of 1 (strongly disagree) to 5 (strongly agree). Scores for each dimension were summed individually (giving ranges of 10 to 50) to determine level of idealism and level of relativism. Forsyth found that the measures were not affected by social desirability, and he found discriminant validity in that the measures were not related to the Defining Issues Test (DIT). Forsyth also found predictive validity as the EPQ results mapped to predicted moral judgment processes. Forsyth found coefficient alpha reliabilities of .80 and .73 for idealism and relativism respectively, as compared to .89 and .84 in the present study. Motivation to lead was measured using Chan’s (1999) motivation to lead instrument. This is a three-dimensional measure, with 9 items for each dimension. The motivation to lead items are shown in Appendix A. The introductions to the items were modified to focus the participant on their current organization when considering the questions. Each item was rated on a scale of 1 (strongly disagree) to 5 (strongly agree). Scores were summed for individual dimensions (giving ranges of 9 to 45) to obtain levels for affective-identity, non-calculative, and social-normative motivation to lead. Because these dimensions were correlated, Chan and Drasgow (2001) found support for a secondorder general motivation to lead measure underlying the three first-order dimensions. Scores for each dimension were summed, and the result was divided by the number of dimensions to obtain the level of general motivation to lead. This gave general motivation to lead the same range as the first-order dimensions (9 to 45).
85 Chan (1999) found strong incremental validity for the motivation to lead construct over its antecedents. He also found strong internal validity, as the construct was not equivalent to any of the other constructs (personality, values, leadership, self-efficacy, and past experience) used to measure its antecedents. Although his sample was diverse in occupation, culture, and gender, it was not diverse on age. However, Erickson (2005) performed a validation study of Chan’s construct using older participants, and found internal reliability. For the general motivation to lead scale, low reliability (α = .54) was reported in one study of Latinas (Cintrón, 2004), and for all motivation to lead scales, low reliability was reported in another study of South Africans (Cerff, 2006). However, low reliability did not occur in the present study. The coefficient alpha reliabilities for affective-identity, non-calculative, and social-normative motivation to lead found by Chan and Drasgow (2001) in their study using three samples were .84-.91, .80-.94, and .65-.75 respectively, as compared to those of the present study, which were .85, .85, and .81. Sampling A priori power calculations. In order to achieve adequate power for this study, a priori sample size calculations for small, medium, and large effect sizes were performed for the first type of inferential test, which was hierarchical multiple regression. A priori sample size calculations for small, medium, and large effect sizes were also performed for the second type of inferential test, which was moderated multiple regression. Every effort was made to obtain a sample size that met the requirements of the largest required sample size (assuming a medium effect) resulting from these calculations.
86 A free sample size calculator for hierarchical regression (Soper, 2007a) was used to find minimum sample sizes for small, medium, and large effects. The results are shown in Table 1. The result for a medium effect showed that a sample size of 75 would be required for the hierarchal multiple regression tests to achieve adequate power. Tabachnick and Fidell (2001) give another way to calculate sample size for multiple regression that tests the ratio of the number of participants to predictors (N ≥ 104 + number of predictors). With 21 control variables and two predictors (PO fit and idealism) the required sample size was 127.
Table 1. Hierarchical Multiple Regression: A priori Power Calculation Final Block Effect Size
Calculation Parameters
Sample Size
Small (.02)
Alpha level: .05 Predictors in previous blocks: 21 Predictors in final block: 1 Desired statistical power level: .80
406
Medium (.15)
Alpha level: .05 Predictors in previous blocks: 21 Predictors in final block: 1 Desired statistical power level: .80
75
Large (.35)
Alpha level: .05 Predictors in previous blocks: 21 Predictors in final block: 1 Desired statistical power level: .80
46
Four analyses were performed to determine the required sample size for the moderated multiple regression tests. First, the same calculator used for the hierarchical multiple regressions was used to calculate sample size for the moderated multiple regression tests (Soper, 2007a). The results are shown in Table 2. The required sample size for a medium effect was 77.
87
Table 2. Moderated Multiple Regression: A priori Power Calculation Not Considering Coefficient Differences PO fit X Relativism Effect Size
Calculation Parameters
Sample Size
Small (.02)
Alpha level: .05 Predictors in previous blocks: 23 Predictors in final block: 1 Desired statistical power level: .80
408
Medium (0.15)
Alpha level: .05 Predictors in previous blocks: 23 Predictors in final block: 1 Desired statistical power level: .80
77
Large (0.35)
Alpha level: .05 Predictors in previous blocks: 23 Predictors in final block: 1 Desired statistical power level: .80
48
The second analysis for moderated multiple regression used a table offered by Aguinis (2004, p. 114) that shows required sample sizes for varying power levels and moderator group correlations, assuming equal sample sizes for the groups. The table considers the differences between the regression coefficients for the two moderator groups to determine the power that a sample size will give. The power becomes lower as the differences between the regression coefficients become smaller. The results gathered by looking up entries for one group with a zero coefficient (expected effect for high relativism) and each of small (.1), medium (.3) and large coefficients (.5) for the second group (low relativism) gave a required sample size of 400 for a medium difference and 120 for a large difference. The third analysis used Aguinis, Boik, and Pierce’s (2001) MMRPOWER tool to calculate power considering the differences in the sizes of the two moderator groups (low
88 and high relativism), in addition to differences in the regression coefficients. The results are shown in Table 3. The result for a medium coefficient difference showed that a sample size of 1,500 was required for the moderated multiple regression tests to achieve adequate power for a medium interaction effect size for low relativism.
Table 3. Moderated Multiple Regression: A priori Power Calculation Considering Coefficient Differences PO fit X Low Relativism Coefficient
Calculation Parameters
Sample Size
Small (.10)
Alpha level: .05 PO fit X high relativism coefficient: .01 Statistical power level: .85
90,000
Medium (.30)
Alpha level: .05 PO fit X high relativism coefficient: .01 Statistical power level: .83
1,500
Large (.50)
Alpha level: .05 PO fit X high relativism coefficient: .01 Statistical power level: .88
300
For the fourth analysis, Tabachnick and Fidell’s (2001) calculation (N ≥ 104 + number of predictors) was used with PO fit and idealism, 21 control variables, and relativism as an additional predictor, giving a required sample size of 128. The results of all analyses for moderated multiple regression produced four sample sizes ranging from 77 for a hierarchical multiple regression analysis not accounting for differences in moderator group sizes (i.e., when the moderated regression was treated as a more simple hierarchical multiple regression, with the interaction term included as an additional
89 predictor) to 128 using a participants to predictors ratio calculation, and from 400 using a table source assuming equal moderator group sizes to 1,500 for differing group sizes. Because one sample was used for both the hierarchical multiple regression and the moderated multiple regression tests, an attempt was made to satisfy the largest of all requirements, which is that of moderated multiple regression. As it was not known if it would be feasible to obtain a sample of 1,500 in this study, a sample size of 400 was considered adequate, and it was assumed that moderator group sizes would be balanced. The balanced group assumption seemed reasonable, as the groups would be created by gathering scores one standard deviation above mean relativism and one standard deviation below mean relativism, rather than by splitting using a dichotomous moderator value, such as gender. In fact, it was found in the present study that the groups were sufficiently balanced, as 45% of the participants were highly relative. Selection of participants. A non-random sampling design and a convenience sampling method were used in the present study. Participants with a variety of attributes were required, as levels of motivation to lead, PO fit, and moral philosophy are each present for any individual who works in an organization. The goal of recruitment was to include participants with diverse educational backgrounds, job levels, and experience who work in organizations of varying sizes. Participants were adults, age 18 and over, who were employed in organizations of various sizes. Self-employed participants were not sought. The survey included questions on employment status that were used to qualify participants.
90 Each person from a list of 279 personal and professional contacts received a recruitment e-mail containing a request for participation (see Appendix B). The list was compiled by gathering 98 contacts from a personal e-mail address book, which consisted mainly of friends (most of whom were previous work colleagues) and several family members. Internal e-mail lists of current coworkers at various hierarchical levels from engineering, marketing, and support groups totaling 181 individuals were then added to the list. The same recruitment text was sent to 6,141 Northcentral University students and faculty using the internal messaging system. Note that some text in the e-mail recruitment content can be attributed to MacPhee (2006). Procedures Participants completed an online survey (shown in Appendix C, along with the informed consent and debriefing pages) hosted on a third-party survey site (www.surveymonkey.com). Written permission was obtained to use the motivation to lead and EPQ instruments included in the survey. The questions regarding PO fit are published in numerous forms in various studies. Therefore, because they are not part of a named instrument, no permission was sought to reuse them. Secure transmission of survey results was ensured by using https encryption, and it was not possible to open the survey using http rather than https. User identifiers were not collected and IP addresses were not included in collected data. Survey reliability was enhanced by requiring a password to enter the survey. In addition, internal reliability was tested for all scales. The informed consent page stated explicitly that participation was being requested for a research project. An indication of consent was recorded in the form of an answer to a question on the consent page. This record of consent was logged with a timestamp. The
91 informed consent page contained contact information that participants used to ask questions about the study. The debriefing page also contained contact information that participants were free to use to ask questions about the study. The results database was accessible to the researcher only through the use of a username and password. The survey site publishes a privacy policy stating that they will not access or disclose research data. Data was backed up once an hour, and then backed up at a central site once a day. The servers containing the data were located in a locked cage at a staffed data center with environmental controls. Deleted data was restorable for up to 14 days. Data Analysis The SPSS (Version 15) statistical software package was used for all statistical calculations. Descriptive statistics were conducted to obtain a demographic profile of the sample. Means, standard deviations, and reliability estimates were then calculated for the study variables. Cronbach coefficient alphas were calculated using the raw scores of each predictor and criterion variable. The resulting reliability estimates gave the internal consistency for each scale used to measure the study variables. Before regressions were performed, Pearson correlation was used to verify relationships for PO fit and each of general, affective-identity, non-calculative, and social-normative motivation to lead. Correlations were also verified for idealism and each of general, affective-identity, non-calculative, and social-normative motivation to lead. Before regression testing, PO fit, relativism, and idealism were mean centered to reduce multicollinearity. Hierarchical multiple regression was used to identify control
92 variables and then to test the prediction of each motivation to lead type from each of PO fit and idealism. Hierarchical moderated regression (Aquinis, 2004) was used to test the moderating role of relativism for the relationship between PO fit and general motivation to lead, PO fit and affective-identity motivation to lead, PO fit and non-calculative motivation to lead, and PO fit and social-normative motivation to lead. Methodological Assumptions, Limitations, and Delimitations As all measures appeared in one instrument, it was anticipated that common method bias could be introduced. An individual rating both predictor and criterion variables may attempt to maintain consistency in similar questions (Podsakoff et al., 2003). However, the scales used for the predictor and criterion variables had explicitly different items, and the scales for the predictor and criterion variables differed in size. The survey guaranteed the participants' anonymity (reducing evaluation apprehension and social desirability threats) and the criterion constructs were presented before the predictor constructs, all of which are good remedies for common method bias (Podsakoff et al.). Online research can incur some limitations, including coverage error (sample does not reflect target population), sampling error due to self-selection, measurement error due to misinterpretation or fraudulent responses, and nonresponse error (Bartlett, 2005). However, a meta-analysis found the quality of data collected online to be equivalent to data collected via traditional methods, suggesting that online survey research does not carry as much risk as first thought (Gosling, Vazire, Srivastava, & John, 2004).
93 Ethical Assurances Institutional Review Board (IRB) approval was obtained before data was collected. Participants were not deceived or misled, and their responses remained anonymous. Participants were volunteers who were recruited as individuals, and they did not receive compensation or payment. No institutional sponsorship was pursued. Social science studies can present risk in the form of social, legal, economic, or psychological outcomes (Kraut et al., 2004). Ethics principles and federal regulation prescribe that the chance of harm be no greater than that presented in ordinary life (Kraut et al.). Online participation in research may be less risky in that there is less social pressure to complete the survey. However, online experiments can cause unpleasant feelings or distress. This occurs without the benefit of a researcher present to mitigate harm. To address this risk, the survey included researcher contact information and participants were encouraged to contact the researcher. For online research, debriefing documents can be presented on the research web site. However, because the researcher is not present, it is difficult to determine if the debriefing material has met the needs of the participants. Therefore, the debriefing page contained contact information and participants were encouraged to contact the researcher. Walther (2002) notes that academic research enjoys a privileged position in regards to telephone surveys (they are not blocked by “do not call” lists), suggesting more latitude for academics conducting online research. However, online research may pose special risks that must be managed. Research data is vulnerable to theft during transit and when stored on public servers (Smith, 2003). Data collected during this study did not contain identifying information (with the exception of the IP address, which was
94 available in transit and stored by the survey hosting service, but was not included in the study results) to minimize risk of a breach of confidentiality. Because participants could have taken the survey while at work, a high level of security was needed to guarantee that their responses were not viewable by their employer. SSL encryption of responses was used to accomplish this. Federal regulation requires written consent by human participants. However, IRBs can waive this requirement and allow participants to give consent by clicking a button. Kraut et al. (2004) suggest breaking informed consent into multiple pages or testing for understanding when the subject is at a greater than normal risk. Mueller, Jacobsen, and Schwarzer (2000) suggest that informed consent for online experiments should be as short as possible, as Internet users are not pressured to participate, and they are, to a very high degree, volunteering to participate. To address this issue, the informed consent form was placed on a separate page for this survey and kept as short as possible. The level of data security should match the risk. In this study, sensitive data was handled by a third party to relieve the researcher of this responsibility. Summary This chapter reiterated the hypotheses, and gave descriptions of the research method and design used for the present study. The sampling method, participant characteristics, and study procedures were described. The methods used to collect, process, and analyze the data were then discussed. Finally, assumptions, limitations, and ethical assurances were given.
95 Chapter 4: Findings Overview This chapter describes the data collected for this study and the results of analyses of the data. Demographics, descriptive statistics, reliability test results, and correlations for all scales and control variables are presented. A confirmatory factor analysis for the general motivation to lead scale is also included to verify the presence of this secondorder measure. Results of hierarchical multiple regression analyses used to determine the effect of the PO fit predictor on each motivation to lead criterion are presented. Results of tests of the interaction of PO fit and relativism for each motivation to lead criterion are then given. Finally, test results showing the effect of the idealism predictor on each motivation to lead criterion are provided. Data Preparation All survey responses (including incomplete surveys) were downloaded from the host website. A separate data collector was used for friends and family versus co-workers and Northcentral University students and faculty. The resulting data was analyzed using SPSS (Version 15.0) statistical software. The possibility for out-of-range errors was minimized through the online data collection method, and an inspection of minimum and maximum values revealed no out-of-range values. A total of 1,141 responses were received, 46 of which were friends and family. Of these, 5 did not give consent, 31 were essentially empty responses (no demographic or scale responses), and 21 gave demographic responses only. These cases were excluded from the subsequent analyses. SPSS missing value analysis (MVA) was performed on the remaining 60 incomplete
96 responses to determine if the omitted predictor variables were missing at random (MAR) for criterion variables. If missing cases can be classified as MAR, then listwise deletion is advised (Allison, 2001). Separate variance t-tests showed PO fit and idealism to be MAR for all types of motivation to lead. Relativism was MAR for non-calculative and social normative motivation to lead, and borderline MAR (p = .045) for affective-identity motivation to lead. Due to these findings, and because Allison also considers listwise deletion to be robust to regression, these cases were removed from the analysis, giving a total of 1,024 complete responses. A custom Java program and SPSS were used to calculate reverse codes, compute scale totals, and create dummy-coded control variables. Sample Description Participation was requested from 6,141 students and faculty at a distance-learning university, 98 friends and family, and 181 work colleagues, giving a total of 6,440 potential research participants. The total number of complete responses collected was 1,024, yielding a response rate of 16%. Of these, 40 were friends and family, and 984 were work colleagues, university students, or faculty. The participants were mainly older workers, with a median age of 41 to 50 years old. However, 37% were 40 years of age and younger, and 14% were 32 and under. The sample was fairly balanced in terms of gender at 47% male and 53% female. The majority of participants were White (79%), but the sample was also made up of 10% Black, 4% Hispanic, and 3% Asian participants. The educational level was very high, with 77% of participants possessing a graduate degree. Not surprisingly given the age of the participants, the majority (70%) had 16 or more years of work experience. Leadership experience, job tenure, and organization tenure were more balanced, all with medians of
97 6 to 10 years. Organization size was also balanced, with a median of 101 to 500 employees. The job level was high, as a majority of the participants (59%) characterized themselves as professionals. Of the rest, 34% were first line, middle, or executive managers, and only 7% were technical or clerical. Most participants worked full time (92%). Personal characteristics are shown in Table 4, and job and organization characteristics are shown in Table 5. Selected characteristics were included in regression analyses in order to control for their effect. Predictor and criterion means and standard deviations for all characteristics are shown in Table 9 through Table 19 in Appendix D. Common Method Variance As all measures were taken using one survey, Harman’s one-factor test (Podsakoff et al., 2003) was used to estimate the extent of common method variance. If a single factor emerges or if one factor accounts for the majority of the covariance in the predictor and criterion variables, then common method bias may be present. The secondorder general motivation to lead score, relativism, idealism, and PO fit were entered into an unrotated factor analysis. The analysis gave two factors, with the largest accounting for less than a majority of covariance at 32%. This suggests that common method bias is not a concern in this sample. Nonresponse Bias Armstrong and Overton (1977) showed that late responders are similar to nonrespondents. They suggested that comparing early and late responders could give information as to whether nonrespondents would provide different replies than respondents. Using their extrapolation technique, the means of early and late responders
98 Table 4. Personal Characteristics Characteristic Age
Gender
Educational level
Work experience
Leadership experience
Ethnicity
n
%
18 to 23 years
9
.9
24 to 32 years
136
13.3
33 to 40 years
239
23.3
41 to 50 years
359
35.1
51 years and over
281
27.4
Male
485
47.4
Female
539
52.6
Some high school
0
0
Completed high school
2
.2
Some college
37
3.6
Completed college
42
4.1
Some graduate school
154
15.0
Graduate degree
789
77.1
Less than 1 year
2
.2
1 to 5 years
55
5.4
6 to 10 years
95
9.3
11 to 15 years
157
15.3
16 years or more
715
69.8
Less than 1 year
88
8.6
1 to 5 years
247
24.1
6 to 10 years
218
21.3
11 to 15 years
164
16.0
16 years or more
307
30.0
White non-Hispanic
811
79.2
Asian or Pacific Islander
26
2.5
Hispanic
37
3.6
103
10.1
47
4.6
Black non-Hispanic Other Note. N = 1024.
99 Table 5. Job and Organization Characteristics Characteristic Job tenure
Job level
n
%
50
4.9
1 to 5 years
343
33.5
6 to 10 years
274
26.8
11 to 15 years
144
14.1
16 years or more
213
20.7
Clerical
22
2.1
Technical
52
5.1
599
58.5
53
5.2
Middle manager
160
15.6
Executive
138
13.5
Part-time
81
7.9
Full-time
943
92.1
1 to 50 employees
205
20.1
51 to 100 employees
122
11.9
101 to 500 employees
253
24.7
501 to 1000 employees
110
10.7
1001 employees or more
334
32.6
Less than 1 year
112
10.9
1 to 5 years
342
33.4
6 to 10 years
236
23.0
11 to 15 years
105
10.3
16 years or more
229
22.4
Less than 1 year
Professional First line manager
Employment status
Organization size
Organization tenure
Note. N = 1024.
were compared using one-way ANOVA on gender, ethnicity, educational level, work experience, leadership experience, job level, general motivation to lead, idealism, relativism, and PO fit. No significant differences (p > .05) were found between the two groups.
100 Descriptive Statistics Scale means, standard deviations, correlations, and reliabilities for study variables are shown in Table 6. Means for the predictor variables PO fit, idealism, and relativism were consistent with previous studies. The participants reported high levels of PO fit, i.e., they perceived that their values were aligned with their organization and coworkers. They were more relative than idealistic, indicating that they take situation into account and are pragmatic when making ethical decisions. The motivation to lead criteria means were slightly higher than those found for Chan and Drasgow’s (2001) U.S. sample of students (N = 290), which may reflect the relatively high job level and amount of leadership experience for the sample used in the present study. The means found by Chan and Drasgow for general, affective-identity, non-calculative, and social-normative motivation to lead respectively were M = 32.08, SD = 5.99; M = 31.24, SD = 7.39; M = 34.22, SD = 5.59; and M = 30.79, SD = 4.99.
Table 6. Coefficient Alphas, Correlations, Means, and Standard Deviations for Study Variables Study Variable 1. PO Fit 2. Idealism 3. Relativism 4. Affective-identity MTL 5. Non-calculative MTL 6. Social-normative MTL 7. General MTL
1
2
3
4
5
6
.91
.03
-.07*
.13**
.15**
.15**
.89
.01
-.12**
-.14**
.84
-.07* .85
7
M
SD
.19**
14.79
4.81
-.07*
-.15**
26.16
7.04
.06*
.10**
.08
34.99
7.40
.26**
.37**
.76**
32.89
6.05
.85
.31**
.71**
35.01
5.72
.81
.73**
32.05
4.99
.88
33.32
4.11
Note. N = 1024. MTL = motivation to lead; Coefficient alphas are presented in boldface along the diagonal. *p < .05 (two-tailed). **p < .01 (two-tailed).
101 Tests of Statistical Assumptions All predictor and criterion correlations required for linear regression analysis were present. As expected, lower levels of PO fit were statistically significantly related to lower levels of each motivation to lead type. Also, as expected, higher levels of idealism were statistically significantly related to lower levels of each motivation to lead type. Idealism and relativism were found to be orthogonal, as expected given previous findings (Davis et al., 2001; Forsyth, 1980). Although Forsyth et al. (2008) suggest that predictions from either of these dimensions should consider the other, this study is concerned with contributions of distinct aspects of each to motivation to lead, and therefore, their discriminant validity was relevant. However, given Forsyth’s suggestion, and although not formally hypothesized, each dimension was tested as a moderator of the other for each hypothesized relationship to motivation to lead (tests of a three-way interaction for PO fit, relativism, and idealism; and a two-way interaction for idealism and relativism). These tests did not change the statistical significance level of any results for tests not including the additional moderator. The hierarchical multiple regression and moderated multiple regression tests used to evaluate study hypotheses, and the t-tests used to compare residuals, require that certain statistical assumptions be met. Univariate linear relationships between each predictor and criterion are assumed (Cohen & Cohen, 1983). These relationships were tested by examining scatter plots and correlations. All relationships appeared to be linear. PO fit had a significant positive relationship with each motivation to lead criterion, and idealism had a significant negative relationship with each motivation to lead criterion. Therefore, the linearity assumption appeared to have been met.
102 The relativism moderator had a significant positive relationship with the noncalculative and social-normative motivation to lead criteria, and a significant negative relationship with the PO fit predictor and the affective-identity motivation to lead criterion. Although it had been thought that a moderator and predictor should not be related (Baron & Kenny, 1986), it was later found that this concern was unfounded (Aguinis, 2004). Kenny (2004) also later stated that correlation between predictor and moderator has no special interpretation. Univariate normality for all predictor and criterion variables was examined using histograms and normal probability plots (see Figure 4 through Figure 17 of Appendix D). All skewness and kurtosis values were between -1 and 1. Because skewness for PO fit approached -1 at -.95, a square root transformation was performed. However, although there was some improvement in skewness, using the transformed variable did not change the overall significance or direction of the effects for regression. Therefore, the transformed variable was not used. Examination of plots for residual versus predicted values, and histograms and normal probability plots of the regression standardized residuals indicated that the homoscedasticity (homogeneity of residual variance) assumption was not violated and that the residuals were normally distributed. Examination of the Durbin-Watson statistic for each regression performed showed that the independence of residuals assumption was not violated. All statistics were within the acceptable range of 1.50 to 2.50. The homogeneity of error variance assumption, which requires that the distribution of residuals remain constant across moderator groups, must be met for
103 moderated multiple regression using categorical moderators. Although the regressions in the present study use continuous moderators, subgroups were created using a median split to test this assumption with Aguinis’ (2008) ALTMMR tool. No violations of this assumption were detected. Multicollinearity between predictors and controls was checked by examining the variance tolerance statistics for each regression coefficient. All predictors and controls, with the exception of the professional job level (with a variance tolerance of .07), had a variance tolerance larger then .10, indicating most predictors and controls were not highly correlated (de Vaus, 2002). Hypothesis Testing Procedure All hypotheses were tested using hierarchical multiple regression or moderated multiple regression. Predictors were mean centered before being entered. Backward elimination was used to exclude redundant control variables for each regression equation. The set of dummy-coded variables for each control variable was entered in a separate step. When the R2 change value was significant (p < .05) for a step and the set had at least one statistically significant regression coefficient (also at p < .05) in the final step, the set of dummy-coded control variables was used in the next version of the regression (Hardy, 1993). This procedure was repeated until only dummy-coded control variable sets that contributed significantly to the model were retained. Note that for each regression, the backward elimination procedure was performed exactly twice. Also note that the some high school educational level group was empty and was not included in this analysis. After control variables were selected for each regression, new baseline regressions were performed. Control variables were entered first to observe the predictor
104 or interaction effect after accounting for differences attributable to the control variables. Each set of selected dummy-coded control variables was entered in a separate step. For each hierarchical multiple regression the predictor variable was then entered in the final step. For each moderated multiple regression the predictor variables were entered in the next step, and the interaction term was entered in the final step. The baseline regressions included all outliers. Outliers were then removed by deleting cases with standardized residuals with absolute values greater than three. Revised regressions were then performed. None of the revised regressions differed in direction or significance level from the baseline, and the R2 values did not vary from the baseline by more than 2%. Therefore, the results of the baseline models were used to evaluate all hypotheses. Effect size was calculated for each regression, and confidence intervals are given in all regression tables (Soper, 2007b; Soper, 2008). Reported effect size magnitude is characterized here as small at .01, medium at .09, and large at .25 (Cohen, 1988, pp. 75107). A posteriori power analysis was not performed, as power analysis is considered appropriate for study design rather than data analysis (Hoenig & Heisey, 2001; Lenth, 2001; Zumbo & Hubley, 1998). This study had a finite number of a priori inferences, which reduces concerns for Type I errors (Hochberg & Tamhane, 1987). However, given that multiple comparisons may have inflated Type I error, two methods were used as controls during hypothesis testing. First, the experimentwise Type I error rate was controlled for multiple comparisons by using a Bonferroni adjusted statistical significance level. Families were defined by grouping hypotheses which were similar in content and use (Hochberg &
105 Tarhane). Each question (the relationship of PO fit and motivation to lead; the relationship of PO fit, relativism, and motivation to lead; and the relationship of idealism and motivation to lead) was considered an experimental family using this guideline, giving a Bonferroni adjusted alpha level of .0167 (.05/3). Second, Benjamini and Hochberg’s (1995) false discovery rate (FDR) method, which is considered liberal but suitable for a priori inference testing (Anderson, Burnham, & Thompson, 2000), was used. This method controls the proportion of errors among rejected null hypotheses. This preserves power by controlling the most relevant errors. The significance level cutpoint found using this method was p
