constructing a 33-item multidimensional ethics scale incorporating these dimensions, and applied them to marketing scenarios. In R&R (1990) the authors.
A Validation and Extension of a Multidimensional Ethics Scale
ABSTRACT. Reidenbach and Robin (1988, t990) proposed and refined a multidimensional etl~cs scale. This study, replicates and extends their work by examining the generalizability of the scale beyond marketing to accounting, and to subjects from across the United States and other countries. P,esults indicate that, in general, the scale holds for this different sample and context. However, an additional utilitarian construct emerged in the current study as important for accounting academics in their ethical decisionmaking. We also found that when we refined Reidenbach and Robin's measure of intention to make a particular choice, a sodal desirability bias or "halo effect" was identiffed. Methodological implications for business ethics research are also presented.
& the field of business ethics matures, research efforts expand from exploratory and descriptive
Jeffrey R. Cohen is Associate Professor of Accounting at Boston College. He is a CM.A. and a KPMG Peat Marwick Faculty Fellow. His articles have appeared in the Journal of Accounting Research, Decision Sciences and The Organizational Behavior Teaching Review.His work on Ethics has appeared in Journal of Business Ethics, Issues in Accounting Education, Management Accounting, and The CPAJournal. Laurie IV.. Pant is Associate Professor of Accounting at Suffolk University. She holds an M.B.A. and a D.B.A. and an M.Ed. She serves on the editorial board of Issues in Accounting Education. Her articles on Ethics have appeared in Journal of
Business Ethics, Issues in Accounting Education, Management Accounting and The Organizational Behavior Teaching Review. David Sharp is Assistant Professor of Accounting at University of Western Ontario. He holds a Ph.D. and an M.Sc. He serves on the editorial board of the Journal of International Accounting, Auditing and Taxation. His articles have appeared in The Midland Corporate Finance Journal and Sloan
Management Review.
Journal of Business Ethics 12:13--26, 1993. © 1993 KluwerAcademic Publishers. Printed in the Netherlands.
Jeffre£ Cohen Laurie Pant
David Sharp
studies to the identification and measurement of variables. An essential first step is the careful definition of theoretical constructs, and validation of observable variables to measure them. The "ethicalunethical" construct is of central importance in this context. Much quantitative ethics research is based on presenting subjects with vignettes, and asking them to state, on a bipolar scale, the extent to which they are ethical. (See the review of business ethics research by Randall and Gibson, 1990.) The problem with asking a unidimensional quesfon for a construct as complex as "ethical" is that the reliability of the measure is highly suspect (Nurmally, 1967). Reidenbach and Robin (hereafter R&R) (1988), in a review of the moral philosophy literature, identiffed five major dimensions to the "ethical-unethical" construct. These were justice, relativism, egoism, utilitarian, and deontological. R&R then made an important contribution to business ethics research by constructing a 33-item multidimensional ethics scale incorporating these dimensions, and applied them to marketing scenarios. In R&R (1990) the authors refined and validated their scale into a more parsimmfious 8-item instrument capturing three orthogonal factors (a broad-based moral equity dimension, a relativist dimension, and a contractualist dimension). This paper reports a validafon of that study using a more culturally heterogeneous population, and its extension to ethical dilemmas in public accounfng. Specifically, the major purpose of this study is to examine whether similar factors on a multidimensional ethics scale emerge when tested on a sample of accounting academics using both R&R's marketing scenarios, and additional accountng scenarios. This will enable us to determine whether the factors identified by R&R are sample specific or content specific. An accounting sample and accounting sce-
14
Jeffrey R. Cohen, Laurie W. Pant and David Sharp
narios were chosen because the accounting profession has a long-standing emphasis on ethics. In particular, we tested for one dimension, utilitarianism, which did not emerge in the R&R studies, because it is very amenable to the cost-benefit principle which permeates accountants' decisionmaking. The remainder of the paper is organized as follows. The next section presents the background and motivation. Then follows the methodology including the pretest, instrument validation and sample description, and the results. The last section contains the discussion including methodological recommendations for future business ethics research.
Background and motivation Business ethics researchers are already using the R&R multidimensional ethics scale. For example, Tsalikis and Nwachukwu (1988) used the longer 33item instrument and two of the retail scenarios to determine how black and white U.S. university students differ in their evaluation of business ethics. In a subsequent study (1991), the same authors used a subset of 20 items to examine how U.S. and Nigerian business students reacted to six vignettes dealing with bribery and extortion. Although R&R have called for testing and validating the scale in order to ensure its generalizability, Tsalikis and Nwachukwu neither questioned nor tested its reliability and validity. R&R (1990) state "It seems plausible that the dimensions could maintain their validity across a wide variety of business ethics applications, but this requires continual testing. If the measures are independent of context, then the diversity of potential applications expands greatly ..."
(p. 6s0).
Members of some business disciplines are especially subject to ethical considerations in the performance of their duties. Marketing, for example, is concerned with the responsible behavior associated with marketing and sales programs (see Hunt and Vitell, 1986). Accountants provide information for a decision-making public. As professionals, accountants are entrusted with prescribing and abiding by the rules for presenting financial information. To discharge this obligation professionally, an accountant's behavior must be consistent with the highest of
ethical ideals. The accounting profession has taken this responsibility seriously (Cohen and Pant, 1991), as evidenced by the Report of the National Commission on Fraudulent Financial Reporting, the promulgation of nine "expectation gap" auditing standards, and the issuance of domestic and international codes of conduct (see the AICPA's Code of Professional Conduct and IFAC's Guideline on Ethicsfor ProfessionalAccountants). Recognizing the importance of high ethical behavior for the accounting profession, one objective of this study is to validate the R&R instrument by testing its applicability to accounting academics using accounting vignettes. One very interesting result from R&R's (1990) reduced multidimensional ethics scale was that utilitarianism did not emerge as a factor. R&R viewed the absence of utilitarianism as noteworthy, stating, "Moreover, in debriefing analyses it was obvious that respondents had a difficult time in understanding and applying the concepts inherent in utilitarian thinking" (13.647). In contrast, Harington (1991) and Fritzche and Becker (1984) suggest that managers predominantly follow a utilitarian approach to ethical decision making. This should be especially prevalent in the field of accounting where cost/ benefit analysis is a cornerstone of accounting-based decision making (Horngren and Foster, 1991). In fact, Cohen and Pant (1991) argue that because a utilitarian framework is so amenable to a cost/ benefit way of thinking, it should be used by accounting educators interested in integrating ethics into the accounting curriculum. Accordingly, another major objective of this study is to extend the R&R instrument by testing whether a utilitarian construct emerges as a significant factor for accounting subjects. A third objective of this study is to test the generalizability of the scale to subjects from across the United States, and non-Americans currently residing in the United States. R&R appear to have validated their scale using subjects primarily from the southern United States. The original R&R (1988) study used University of Mississippi students while R&R (1990) used professionals from the same area~ Specifically, the state, "105 small business operators in a different but contiguous state evaluated the scenarios using the 14 scale items. Using the same analysis and item reduction criteria, the 14 items were reduced to 8 items, which formed the three
Multidimensional Ethics Scale
factor structure shown in Table II and used in the current study" (p. 542). This could be important in that the ethical factors that emerged in their study may have been a function of the sample. R&R (1988) implicitly recognize this issue when discussing the content of a factor, stating "Alternatively, the construct identified by factor three seems to suggest a more formalized set of rules and duties, perhaps based on family and rdigious training since most of the respondents are indigenous to the so-called 'Bible Belt'" (p. 877), A final objective of this study is to refine the univariate measure used by R&R to assess intention or willingness to perform an unethical act. R&R (1988 and 1990) asked subjects to state on a sevenpoint scale the probability that they would undertake the action described in each vignette. Although this is a widely used approach, it may result in subjects overstating their ethical intentions, because of the well-known "halo" or "social desirability" effect prevalent in survey research (see Rossi et at., 1983). To test for this effect, our study adds a question asking subjects to assess the probability that their peers or colleagues would undertake the same action. This wilt allow us to examine and control for a halo effect.
Methodology The study is in two parts. In the first (the pretest), we reduce the original 33 items of R&R's 1988 study to a 15-item instrument following R&R's guidelines. In the second, we test the reliability of this instrument on a sample of accounting faculty.
Pretest
The first stage in refining the R&R multidimensional ethics scale involved a comprehensive pretest conducted with the original R&R (1988) 33-item instrument, augmented by three questions providing univariate measures of ethical evaluation and behavioral intention. The 92 subjects who took part in the pretest included 57 full-time first year MBA (21 international and 36 U.S.) students and 35 upperlevel undergraduate accounting majors at a highly selective private university.1 The objective of this
15
pretest was to reduce the 33 items to a more manageable number using a sample that included a substantial number of non-U.S, citizens, and subjects who had received formal accounting training. These subjects were presented with R&R's three retail scenarios and the original scale items of R&R (1988). See Table I for a description of the retail scenarios used in the pretest. Subjects were told "This study is designed to validate a questionnaire that measures the importance of beliefs in business decision-making . . . There are no right or wrong answers. All individual responses are guaranteed anonymity and will be used strictly in the aggregate to validate the scale." The rationale for these i n s t r u c t i o n s was to attenuate the %ociat desirability" effect that might occur if the subjects were told that this was an ethics study. A valid multidimensional instrument captures those dimensions of ethical decision-making identiffed in R&R (1988) which are actually used by subjects in their responses. Only those scale items which correlate highly" with one dimension, and do not highly correlate with any other, are included. For this reason, exploratory factor analysis, using a 5-factor solution, was used as a first step. Following R&R, we used orthogonal rotation. Although there is no reason to suggest that the underlying five constructs are uncorrelated (indeed, given that they are all "ethical" constructs, we would expect them to be correlated), the strong attraction of orthogonality is statistical convenience. We purged items using the criteria specified by R&R (1990, p. 642) including the consistency of the loading across scenarios, size of loadings, and low inter-item correlation with other dimension items. First, a scale item was eliminated if its highest loading on aW factor in at least two out of the three scenarios was less than 0.6. Then, the remaining 20 items were subject to another factor analysis in which an item was eliminated if it significantly" loaded on more than one factor in at least two scenarios. This enhances the convergent and discriminant validity- of the instrument. From this process, a 15-item (plus 3 univariate measures) multidimensional ethics scale emerged, which was used in the main questionnaire in the second, validation, part of our study. Three of the items that emerged from the pretest were utilitarian. Thus, even using the original retail scenarios, a utilitarian factor emerged in the pretest as an im-
16
Jeffrey R. Cohen, Laurie W. Pant and David Sharp TABLE I Pretest scenarios
Automobile A person bought a new car from a franchised automobile dealershipin the local area.Eight months after the car was purchased, he began having problems with the transmission. He took the car back to the dealer, and some minor adjustments were made. During the next few months he continually had a similar problem with the translmssion slipping. Each time the dealer made only minor adjustments on the car. Again, during the thirteenth month after the car had been bought the man returned to the dealer because the transmission still was not functioning properly.At this time, the transmissionwas completely overhauled. Action: Sincethe warranty was for only one year (12 months from the date of purchase), the dealer charged the full price for parts and labor.
Salesman A young man, recently hired as a salesmanfor a local retail store, has been working very hard to favorablyimpress his boss with his selling ability. At times, this young man, anxious for an order, has been a little over-eager.To get the order, he exaggerates the value of the item or withholds relevant information concerning the product he is trying to sell. No fraud or deceit is intended by his actions, he is simply"over-eager. Action: His boss, the owner of the retail store, is aware of the salesman'sactions but has done nothing to stop such practice.
NeighborhoodStore A retail grocery chain operates several stores throughout the local area including one in the city's ghetto area. Independent studies have shown that prices do tend to be higher and there is less of a selection of products in this particular store than in the other locations. Action:
On the day welfare checks are receivedin the area of the city the retailer increasesprices on all of his merchandise.
portant factor for a sample that included international students and accounting majors. The responses to the scenarios were also evaluated within the context of the multitrait-multimethod (scenario) approach of Campbell and Fiske (1959). This was conducted to determine if all three retail scenarios should be included in the main questionnaire. The results of the multitrait-multiscenario analysis indicates that only two out of the three scenarios showed high convergence (automobile and neighborhood store) leading us to drop the third (salesperson). R&R (1988) in contrast, found that the automobile scenario failed to meet convergence criteria.
Instrument validation The multidimensional ethics scale that emerged from the pretest is presented in Table II. The objective of this part of the study was the validation and refinement of the scale on a sample of
accounting academics with relatively broad geographical and cultural diversity, using the two retail scenarios that demonstrated convergent validity in the pretest, and four additional accounting scenarios. The accounting scenarios were taken from Claypool et aL (1990) who in turn had adapted them from Armstrong (1985) and Loeb (1971). These accounting scenarios have been tested and used for different purposes across a wide variety of subjects. For example, ClaypooI et aL had utilized the scenarios to examine factors that CPAs and theologians would consider salient when reacting to ethical dilemmas. By including both the retail and accounting scenarios we could test the reliability and validity of the multidimensional ethics scale both within and across different business contexts. Table III presents the scenarios used in the study.
Subjects The questionnaires were mailed to a sample of 336
Muttidimensionat Ethics &ale
17
TABLE II Multidimensional ethics scale
J ust
Unjust
Fair
Unfair
E
Cuiturally Acceptable
Culturally Unacceptable
individually Acceptable
Individually Unacceptable
Traditionally Acceptable
Traditionally Unacceptable
Acceptable to my Family
Unacceptable to my Family
Self Promoting
Nor Self Promoting
Self Sacrificing
Not Self Sacrificing
Personally Satisfying
Not Personally Satisfying
Produces the Greatest Utility,
Produces the Least Utility
Maximizes Benefits While Minimizes Harm
Minimizes Benefits While Maximizes Harm
Leads to the Greatest Good for the Greatest Number
Leads to the Least Good for the Greatest Number
II
Violates an Unwritten Contract
Does Not Violate an Unwritten Contract
Obligated to Act this Way
Not Obligated to Act this Way
Violates an Unspoken Promise
Does Not Violate an Unspoken Promise
The probability that I would undertake the same action is
High
The probability that my peers or colleagues would undertake the same action is
High
Low
Low ,,,
The action described above is
Ethical
Unethical
Jeffrey R. Cohen, Laurie W. Pant and David Sharp
18
TABLE III Scenarios used in main questionnaire ACCOUNTING
sibling CPA Z's sister, Susan, is the treasurer and a 26% stockholder of ABC Corporation. The president of ABC Corporation asked Z if he would perform the annual audit of ABC Corporation. Action:
CPA Z accepts the audit engagement.
Bankruptcy CPA Z serves as the auditor for Widget & Co. Widget's market share has declined drastically, and Z knows that Widget will soon be bankrupt. Another of Z's audit clients is Solid Company. While auditing Solid's accountants receivable, Z finds that Widget & Co. owes Solid $200,000. Action:
CPA Z warns his client, Solid Company, about Widget's impending bankruptcy.
CommunityService CPA Z, in addition to practicing public accounting, is heavily involved in community activities. He is especially well known for his passionate promotion of higher voter participation in elections. High-Voter Company, a newly formed compaW, has developed a revolutionary promotional process that the company's officers claim will greatly increase voter turnout. The president of High-Voter asked Z if he would perform the initial audit of High-Voter. Action:
CPA Z accepts the audit engagement.
Mc~'ger CPA Z is considering a merger with CPA F. To facilitate the negotiations F requests access to Z's files of dient work papers, income tax returns, and correspondence, Z's clients are unaware of the proposed merger. Action:
CPA Z grants F access to the files.
RETML
Automobile A person bought a new car from a franchised automobile dealership in the local area. Eight months after the car was purchased, he began having problems with the transmission. He took the car back to the dealer, and some minor adjustments were made. During the next few months he continually had a similar problem with tile transmission slipping. Each time the dealer made only minor adjustments on the car. Again, during the thirteenth month after the car had been bought the man returned to the dealer because the transmission still was not functioning properly. At this time, the transmission was completely overhauled. Action:
Since the warranty was for only one year (12 months from the date of purchase), the dealer charged the full price for parts and labor.
NeighborhoodStore A retail grocery chain operates several stores throughout the local area including one in the city's ghetto area. Independent studies have shown that prices do tend to be higher and there is less of a selection of products in this particular store than in the other locations. Action:
On the day welfare checks are received in the area of the city the retailer increases prices on all of his merchandise.
Multidimensional Ethics Scale accounting academics listed in Hassetback's (1990) directory of accounting faculty. To ensure more culturally and geographically diverse respondents than those used by R&R, our sample included faculty with international interests, namely, members of the International section of the American Accounting Association, as well as academics from across the United States. We reasoned that they were most likely to contain respondents familiar with a~.d sympathetic to non-American culture and values. Twenty-eight of the questionnaires were returned as undeliverable leaving a total of 308 possible respondents. Ninety responses were generated from the first mailing and an additional 23 responses were received from the second mailing. The total of 113 responses represented an effective response rate of 37%. The questionnaire took on average approximately 20 minutes to complete. To test for possible nonresponse bias, we employed the wave approach (Armstrong and Overton, 1977). This involves a comparison of respondents from the early and late waves. Out of a possible 108 questions, (18 questions for each of the six scenarios) onty 9 resulted in significant (p < 0.05) differences between the first and second wave. Accordingly, for all subsequent data analysis, the results are presented for the aggregate sample?
Results Factoranalysis Following R&R, the responses to each of the six scenarios were tested by principal components factor analysis using die varimax rotation. Tabachnik and Fidetl (t989) state "The goal of Principal Components Analysis is to extract maximum variance of factor ioadings by making higher loadings higher, tow ones lower for each factor" (p. 610). Thus, this approach is more likely to lead to cleaner and more interpretable factors, which is especially useful if the factors are to be used later fi3r predictive purposes. The results of a natural factor soluuon are presented in Table IV. We retained only those factors with an eigenvalue greater than 1.0. In the discussion that follows, only items with a loading greater than 0.60 are considered to be included in a factor. Examining Panel A (the accounting scenarios) of
19
Table IV, a natural three-factor solution emerged for two of the scenarios, while one had a natural twofactor and the other a four-factor solution. Similar to R&R, a justice/relativist dimension emerged as a factor in all four scenarios, and a deontological ("contractualism") factor was present in three out of four scenarios. However, what is striking in these results is that elements of a separate utilitarian factor emerge in three of the scenarios while the fourth scenario includes the utilitarian elements with the justice/relativist elements. Clearly, our subjects incorporate utilitarian analysis in their ethical decision-making process. Turning to Panel B, we can examine the factor an@is of retail scenarios. This will enable us to explore whether the factors are scenario specific (accounting) or are a function of the subjects in our sample (accounting academics). The automobile scenario generated a natural five factor solution. Again, as in R&R, a relativist/justice factor and a deontological factor emerged. However, among the other three factors there emerged a factor that included all three utilitarian items. The results for the other retail scenario (neighborhood store) were much cleaner in that factors tended to align with the theoretical constructs. A strong relativist/justice dimension was a major factor while a utilitarian dimension also emerged. Taken together, the results from both the accounting and retail scenarios led strong support to the generalizability of R&R's notion of the importance of a broad-based relativist/ justice dimension to explain ethical decision making. However, the results also strongly suggest that elements of a utilitarian framework are evident in the ethical decision maldng of accounting academics. Because different factors and a different number of factors emerged across the scenarios, we could not replicate R&R's muttitrait-multicontext analysis that tests for the convergent and discriminant validity of the scales. This raises questions about the usefulness of the assumption of factor orthogonality if the underlying constructs are correlated. However, calculation of Cronbach's alpha demonstrates that the reliability of each of the factors is very high, ranging from 0.83 to 0.96 with an average of 0.89. This far exceeds the standard of reliability (0.6) that Nunnally (1967) advocates for exploratory work. In addition, we calculated a single Cronbach's alpha separately for each scenario using the complete multidimen-
20
JeffreyR. Cohen, Laurie W, Pant and David Sharp TABLE IV Factor structures for the six scenarios
PANEL A Accounting Scenarios Justice/Relativist B C M
Variables
S*
Just/Unjust Fair/Unfair Culturally acceptable/Unacceptable Individually acceptable/Unacceptable Traditionally acceptable/Unacceptable Acceptable to my family/Unacceptable
0.77 0.75 0.75 0.80 0.75 0.68
0.76 0.75 0.67 0.74 0.75 0.67
0.80 0.85 0.87 0.89 0.86 0.78
Utilitarian S t3 C
S
Deontological C M
Egoist B
Hybrid S
0.79 0.8t 0.82 0.77 0.77
0.76
Self promoting/Not Self satisfying/Not
0.65
0.84 0.86
Personally satisfying/Not Producing the greatest utility/Least Maximizing beneflts/Maximizing harm Leads to greatest good/Least good
0.64
0.72 0.89 0.79 0.72
0.64 0.79 0.78 0.71 0.61 0.79 0.69 0.79 0.83
Violates an unwritten contract/Does not
0.84
0.75
Obligated to act this way/Not obligated Violates an unspoken promise/Does not
0.75
0.61
0.79
0.63 0.84
0.78
* S = Sibling scenario B = Bankruptcy scenario C - Community service scenario M - Merger scenario Note: A cutoffofa loading of 0.6 was used to include an item into a factor PANEL B Retail Scenarios
Variables Just/Unjust Fair/Unfair Culturally acceptable/Unacceptable IndividnaUy acceptable/Unacceptable Traditionally acceptable/Unacceptable Acceptable to my family/Unacceptable Self promoting/Not Self satisfying/Not Personally satisfying/Not Producing the greatest utility/Least Maximizing benefits/Maximizing harm Leads to greatest good/Least good
Justice/Relativist M N 0.84 0.84
0.80
Utilitarian A N
Deontological A N
Egoist A
Hybrid N
0.76 0.83 0.72 0.68 0.75 0.68
Relativist A
0.88 0.88 0.76
0.64
0.70 0.80 0.83 0.79
0.67 0.72 0.75
0.66
Muhidimensionat Ethics Scale
21
Tatde tV (Cominued) Violates an unwritten contract/Does not Obligated to act this way/Not obligated Violates an unspoken promise/Does not
0.90
0.87
0.83
0.86
0.73 0.85
* A = Auto scenario N ~ Neighborhood store scenario
sionat scale. Again, the reliability was very high with the alpha coefficients ranging from 0.95 to 0.97. This demonstrates an extremely high reliability and consistency of the measurement scale.
Measure of intention Because of the welt-known "social desirability" or "halo" effect (see Rossi et at., 1983) we included two questions to measure intention. In addition to repeating R&R's question: "The probability that I would undertake the action is," we also asked subjects "The probability that my peers or colleagues would undertake the same action is." Table V presents the means (standard deviations) and the results of the paired t-tests for the comparisons of the measures of intention in each of the six scenarios. Subjects responded that it was less likely that they would perform an action than would their colleagues. In five of the six scenarios, the results of a paired t-test were highly significant (p < 0.001) and for the sixth scenario, the difference was in the expected direction, but marginally significant (p < 0.10). These results confirm that in business ethics research there is a need to control for this "halo" effect when measuring behavioral intention.
The use offactorsdSr predictabitity One benefit of a multidimensional ethics scale is that it should better capture the complexity of ethical decision-making. This includes both the evaluative measure (ethical-unethical) and the intention or willingness to perform an action. Therefore, following the procedures of R&R (1990) the factor scores that emerged in each scenario were used to explain
and predict the univariate measure of ethical evaluation and the two measures of intention. The three separate multiple regression results (one for each univariate measure) for each scenario are presented in Table VI. In addition, the regression coefficients are presented in order to determine the importance of each factor in explaining evaluation and intention/willingness. This will enable us to assess the importance of such factors as utilitarianism in accounting-based ethical decision making. Examining Table VI we see that the factors have significant explanatory power in all six scenarios for both measures of intention (I would do the action, and My colleagues would undertake the action) and for evaluation (How ethical was the action). Examining the standardized regression coefficients, it appears that for both measures of intention and the measure of evaluation, a relativist/justice factor is most important. In addition, in the majority of cases, the utilitarian factor is the second most important factor. Following R&R, our final analysis compares the predictive ability of a multidimensional ethics scale with a univariate measure of ethics to explain both measures of intention. The results are presented in Table VII. Looking at the first measure of intention (the probability I would undertake the same action), in the two retail scenarios the multidimensional scale explained a larger portion of the variance than the univariate measure. This is consistent with R&R (1990) who found that for both of these retail scenarios, the multidimensional scale explained a greater amount of the variance. However, in none of the accounting scenarios was the multidimensional scale more explanatory. Similarly, for the second measure of intention, the probability that my colleagues would undertake the action, the multidimensional ethics scale was equal to or superior in the two retain scenarios, but weaker than the univariate
22
Jeffrey R. Cohen, Laurie W. Pant and David Sharp TABLE V Comparisonof measures of intention
Probability I'd undertake action
Mean (Std. Dev) Probability my colleagues would undertake action
Means (Std. Dev.) Difference
value
Accounting Sibling Bankruptcy Community Service Merger
6.41 5.6t 3.36 6.77
(1.32) (1.72) (2.08) (0.81)
5.94 5.14 3.14 6.05
(1.49) (1.72) (1.80) (1.22)
0.46 0.47 0.18 0.75
(0.11) (0.13) (0.10) (0.11)
4.31 3.62 1.77 6.86
0.0001 0.0005 0.0789 0.0001
Retail Automobile NeigborhoodStore
6.49 5.32
(1.18) (1.98)
5.77 4.87
(1.38) (1.94)
0.71 0.43
(0.11) (0.08)
6.44 5.31
0.0001 0.0001
Scenarios
Mean (Std. Dev.)
Paired t-test
Significance
Note: Response Scale Extremes were (1) Highly Probable and (7) Highly Improbable.
measure of colleague intention in three of the four accounting scenarios. Thus, the ability to predict intention in our study is contingent on the context.
Conclusion This study extends R&R's seminal work on multidimensional ethics scale development by incorporating a different type of subject from another business discipline and from a broader geographic and cultural distribution, and by using additional ethical contexts from accounting. The results support the importance of a relativist/justice factor, but also support the use of a utilitarian factor, in accountants' ethical decision making. This finding holds across both retail and accounting scenarios, suggesting that these factors are not context-speeific. The results also suggest that the assumption of orthogonality of constructs does not appear to have been justified either in R&R's (1988, 1990) work or in this study. It appears that elements of philosophical orientations are intertwined. This should be recognized when interpreting results derived from the scale. However, the use of multiple items will enhance the reliability of the scale instrument. The replication in this study of the importance of the relativist/justice elements and the finding that an accounting-based sample was also highly driven by
an utilitarian construct emphasizes the importance of testing this scale in numerous settings. For example, in other business disciplines the use of a cost/ benefit principle may also be so embedded in the subject area that it lends itself to the use of a utilitarian construct. Or in other countries, ethical decision-making may be so influenced by culture that the use of an ethics scale constructed from a U.S. sample would be irrelevant to the potential subject. For example, the scale might not be appropriate in an Asian culture, where the emphasis on familial and communitarian values is strong. Thus, another important step would be to more explicitly test the validity of the scale in an international setting. Another important methodological implication for business ethics research is the confirmation of the existence of a "halo" effect. In this study, subjects consistently reported that they would be more ethical than their peers. This suggests that business ethics research which asks respondents only whether they would do an action probably overestimates the ethical behavior of its sample, and so may limit the generalizability of a study's findings. Surprisingly, the multidimensional ethics scale in the accounting scenarios had less predictive abilityfor all but one of the intention questions. In contrast, we confirmed R&R's (1990) finding in the retail scenarios that the multidimensional scale had greater
Neighborhood Store
Auto
Retail
Community Service Merger
Bankruptcy
sibling
Accounting
0.49 (0.0001) 0.49 (0.0001)
0.61 (0.000 i) 0.64 (0.001) 0.76 (0.0001) 0.85 (0.0001)
R~
0.46 (0.0013) 0.62 (0.0001)
0.73 (0.0011 0.78 (0.0001) 0.75 (0.000I)
0.23 (0.0001) 0.11 (NS)
0.31 (0.001) 0.34 (0.0001)
0.26
(0.0001)
0.73
Utilitarian
(0.0001 )
f12
fl,.
Justice/ Relativist
Ethical
--0.45 (0.000 l) --0.14 (0.04)
--0.21 (0.0001) --0.55 (0.0001)
--0.13 (0.03)
Deontological
fl~
0.02 (NS)
--0.15 (0.009)
Egoist
fl,
TABLE VI Multiple regression on univariate measures of evaluation and intention
0.29 (0.0001)
0.12 (0.04)
Hybrid
f16
0.21 (0.002)
Relativist
b,)
r~
0.60 (0.0001) 0.22 (0.0001)
0.57 (0.0001) 0.58 (0.0001) 0.66 (0.0001) 0.76 (0.0001)
0.21 (0.00tl) 0.33 (0.0001) 0.32 (0.0001)
fl~
0.60 0.32 (0.0001)(0.0001) 0.42 0.09 (0.0001)(NS)
0.71 (0.000t) 0.66 (0.0001) 0.74 (0.0001) 0.73 (0.0001)
fl~
--0.35 (0.0001) 0.003 (NS)
-0.t3 (0.0001) --0.46 (0.0001)
--0.04 (NS)
fl~
My Intention
0.03 (NS)
-0.22 (0.0005)
fl~
* fl's are standardized regression coefficients (iv-values in parentheses).
Neighborhood Store
Auto
Retail
Community Service Merger
Bankruptcy
Accounting Sibling
R 2
Table VI (Continued)
0.26 (0.003)
0.17 (0.007)
fl~
0.17 (0.005)
fl~
0.24 (0.0001) 0.10 (0.005)
0.60 (0.0001) 0.43 (0.0001) 0.59 (0.0001) 0.72 (0.0001)
R2
fl~
0.31 0.28 (0.0003)(0.0012) 0.27 0.09 (0.003) (NS)
0.72 0.22 (0.001) (0.003) 0.61 0.27 (0.0001) (0.0003) 0.64 0.39 (0.0001)(0.0001) 0.73 (0.0001)
fl~
--0.t4 (0.09) -0.08 (NS)
--0.18 (0.0001) --0.44 (0.0001)
-0.19 (0.002)
fl~
--0.04 (NS)
f14
Colleagues' Intention
0.21 (0.02)
0.10 (0.09)
fl~
0.26 (0.002)
flo
~2
¢%
©
4~
Muhidimensional Ethics Scale
25
TABLE VII Comparison of simple and multiple regressionR2on measuresof intention Scenario
Accounting Sibling
My Intention
Univariate Multidimensional Difference
Bankruptcy
Univariate Multidimensional Difference
Community Service
Univariate Multidimensional Difference
Merger
Univariate Multidimensional Difference
Retail Auto
Univariate Multidimensional Difference
Neighborhood Store
Univariate Multidimensional Difference
predictive ability than a univariate measure. This may be explained by the accounting profession's long tradition of a formal code of ethics. Subjects who are inculcated with a tradifon of a code might view behaviors within a lens of"covered by the code vs. not covered by the code." For these subjects, a definition of "ethical" must be explicitly stated, otherwise responses to vignettes may become an exercise in code interpretation. A future study could compare the relevance of a multidimensional scale for subjects from professions with established codes of ethics (e.g., accounting, law) and subjects from a profession in which the code might be a relatively new phenomenon (e.g., a manager in a corporation). Finally, our validation exercise confirms R&R's suspicion that the validity of the multidimensional
Colleagues' Intention
0.76 0.57 -0.19 0.68 0.58 -0.10 0.82 0.66 -0.16 0.92 0.76 -0.16 0.50 0.60 +0.10 0.09 0.22 +0.13
Univariate Multidimensional Difference Univariate Mulndimensional Difference Univariate Multidimensional Difference Univariate Multidimensional Difference Univariate Multidimensional Difference
0.62 0.60 -0.02 0.36 0.43 +0.07 0.68 0.59 -0.09 0.77 0.72 -0.05 0.09 0.24 +0.15
Univariate Multidimensional
0.10 0.10
Difference
0.00
scale needs to be tested across different business ethics applications. Although some of the criteria used by accountants (or accounting-trained subjects) to judge whether an action is ethical or not differed from those used by R&R's subjects, significant similarities between the two disciplines were also found. Thus, R&R's 33-item scale may well provide the basis of multidimensional scales, but a scale must be constructed and validated for each application studied.
Acknowledgement We would like to acknowledge the helpful comments provided by Jean Romeo and Marry Roth, and the research assistance of M. Rocio Luhring.
26
Jeffrey R. Cohen, Laurie W. Pant and David Sharp
Notes T-tests were conducted to determine if responses to individual scale items were different between the international and U.S. M~B.A. students, and between M.BA. and undergraduate students. In general, M23~ students viewed the actions as less ethical than the undergraduate students and would be less likely to undertake a similar action. A comparison of the international and U.S. M33A. students found significant differences only on the neighborhood store scenario which involved a retailer raising prices on the day welfare checks arrive. The international students viewed the action as more ethical and would be more likely to undertake the action than the U.S. MN.A. students. -~ Because Gilligan (1982) suggests that women have different moral development than men, we also examined a breakdown of the results by gender. The sample consisted of 79 males, 30 females, and 3 no responses. A comparison of the results found in general no significant differences between the responses of men and womeru The only exception occurred in the first scenario which involved a familial relationship in which a CPA accepts an audit engagement of a company where his/her sister is the treasurer and is a major stockholder. Female respondents viewed the action as less ethical than male respondents. It may be that males felt a more protective attitude towards the sister that resulted in a willingness to take an action they" would otherwise have found inappropriate.
References Armstrong, J. S. and Overton, T. S.: 1977 (Aug), 'Estimating Nonresponse Bias in Mail Surveys', Journal of Marketing Research,pp. 396-402. Armstrong, M.: 1985, 'Internationalization of the Professional Ethic by Certified Public Accountants: A Multidimensional Approach', Doctoral Dissertation, University of Southern California. Campbell, D. T. and Fiske, D. W.: 1959 (Mar), 'Convergent and Discriminant Validation by the Multitrait-Multimethod Matrix', PsychologicalBulletin56, pp. 81-105. Claypool, G. A., Fetyko, D. F. and Pearson, M. A.: 1990, 'Reacting to Ethical Dilemmas: A Study Pertaining to Certified Public Accountants', Journal of BusinessEthics 9, pp. 699-706. Cohen, J. K and Pant, L. W.: 1991, 'Beyond Bean Counting: Practitioners' Views Concerning Mechanisms for Institutionalizing Ethical Standards',JournatofBusinessEthics 10, pp. 45-56.
Fritzche, D.J. and Becker, H.: t 984, 'Relating Management Behavior to Ethical Philosophy- - An Empirical Investigation', Academy of ManagementJournal 27(1), pp. 166175. Gilligan, C.: 1982, In a Different Voice (Harvard University Press, Cambridge, Mass.). Harington, S.J.: 1991, 'What Corporate America is Teaching About Ethics', Academy of Management Executive 5 (I), pp. 21--30. Hasselback, J. R.: 1990, Accounting FacultyDirector (Prentice Hall, Englewood Cliffs, New Jersey). Horngren, C. T. and Foster, G.: 1991, Cost Accounting: A Managerial Emphasis, 7th ed. (Prentice Hall, Englewood Cliffs, New Jersey). Hunt, S. D. and Vitell, S, 1986 (Spring), 'A General Theory of Marketing Ethics',JournalofMacromarketing, pp. 5-16. Loeb, S.: 1971, 'A Survey of Ethical Behavior in the Accounting Profession', Journal of AccountingResearch 9, pp. 287306. Nunnally,J. C.: 1967, 'Psychometric Methods (McGraw-Hill Book Company, New York). Reidenbach, R. E. and Robin, D. P.: 1988, 'Some Initial Steps toward Improving the Measurement of Ethical Evaluations of Marketing Activities', Journal of BusinessEthics 7, pp. 871-879. Reidenbach, tL E. and Robin, D. P.: 1990, 'Towards the Development of a Multidimensional Scale for Improving Evaluations of Business Ethics',Journal ofBuzinessEthics 9, pp. 639-653. Rossi, P. H., Wright, J. D. and Anderson, A. B.: 1983, Handbook of Survey Research (Academic Press, Orlando, Florida). Randall, D. M. and Gibson, A. M.: 1990, 'Methodology in Business Ethics Research: A Review and Critical Assessment', Journal ofBusinessEthics, pp. 457-472. Tabachnick, B. G. and Fidell, L S.: 1989, UsingMultivariate Statistics,2nd ed. (Harper Collins Publishers, New York). Tsalikis, J. and Nwachukwu, O.: 1989, 'Cross-Cultural Business Ethics: Ethical Beliefs Difference Between Black and white',Journal ofBusinessEthics 7, pp. 745-754. Tsalikis, J. and Nwachukwu: 1991, 'A Comparison of Nigerian to American Views of Bribery and Extortion in International Commerce', Journal ofBusinessEthics 10, pp. 85--98.
Boston College, Wallace E. Carroll &hod of Management, Chestnut Hitt, MA 02167-3808, U.&A.
