Validity of the Multidimensional Ethics Scale for a Sample of Thai Physicians

Research in ethical decision-making has received considerable attention in the realm of the business community in the last three decades due in part to numerous high profile scandals (e.g., Enron). The medical community has been less engaged in this line of investigation as the primary scholar focus has been in biomedical as opposed to social science/humanities. However, recently researchers and their methods have been attracted to the medical field. The purpose of this paper is to explore whether an ethical decision-making measure prominent in the business literature can be applied to the medical contexts.


Introduction
Ethics in medicine has a profoundly long history from the early followers of Hippocrates to the disciples of Taoism and traditional Chinese medicine. However, it has not been until recently that a concerted effort has been made to study the decision-making behaviour of physicians (Malloy et al., 2008).
Much of the contemporary applied ethics research has focused upon the perceptions and practices of members of the business community (Cohen et al., 1996) with development and use of questionnaires to assess ethical judgement and ideology (Forsyth, 1980). Of the many instruments used to assess ethical decision-making in this context, the Multidimensional Ethics Scale (MES) developed by Reidenbach and Robin (1988) has received considerable attention. The purpose of this study was to determine if the MES was a viable instrument to be used in medical contexts.
The MES was originally developed to assess ethical decision-making in business (Reidenbach & Robin, 1988;1990). To this end, 33 items were designed to tap into five ethical decision domains: Deontology, Utilitarianism, Relativism, Egoism, and Justice. Deontology considers ethical conduct to be duty-based in which the outcome of our actions are of secondary importance. Utilitarianism argues that the outcome is the primary goal of ethical conduct and the process one employs is of less immediate concern. Relativism purports that outcome and process are particular to the situation and that one must be flexible when deciding what is or is not ethical conduct (when in Rome do as the Romans do). Egoism is individually-based and directs each of us to pursue the greatest pleasure for ourselves as a means to seek the 'good'. Finally, justice has its roots in Aristotelian theory whereby equals ought to be treated equally.
Through the use of exploratory factor analysis (EFA), the number of items was eventually brought down to eight, measuring a total of three dimensions that combined to form the final version of the MES. The first dimension is Moral Equity, representing the notion of right and wrong is a second dimension that taps into social concepts learned through experience. The final dimension, Contractualism, represents the notion of obligation and social contract.
Although the scale, as developed by Reidenbach and Robin, consists of eight items divided over three dimensions , it has been used in different variations since its inception. For example, using the same initial Regardless of which version has been used, MES has been exclusively applied within a business context. Thus, to assess its applicability for use with a different population, it is important to test and validate the factor structure of the instrument in a sample drawn from a new population (Bollen, 1989, Byrne, Shavelson, & Muthén, 1989. This study focuses on assessing the validity of the MES-8 in a medical context with physicians.

Participants and Procedures
Physicians in this study were participating in a larger investigation exploring the cross cultural influences on ethical decision making. Packet A contained scenarios 1 and 2, packet B contained scenarios, 3 and 4, packet C contained scenarios 5 and 6 and packet D contained scenarios 1 and 7. Age and sex of the participants as well as the number of participants who completed the MES-8 per scenario can be found in Table 1.
Overall, 23.5% of the respondents indicated they were general practitioners. A cross section of specialties is also represented with physicians self-identifying 24 different areas of specialization. Nearly all physicians (98.4%) completed their medical training in Thailand. In terms of religious affiliation, the vast majority of physicians (95.9%) identified themselves as Buddhists.

Translation
When conducting cross-cultural research it is essential that the language used across cultures is equivalent. To minimize the impact of language differences the questionnaire underwent a translation and back-translation process (Sekaran, 1983). The MES and scenarios were originally compiled in Canada (English) and then translated by language experts in Thailand. Scenarios were then translated back into English by Canadian experts. The back translations were reviewed by the research team to ensure accuracy. While no translation is entirely error free, we believe this method minimizes the potential threats to the study's validity.

Analyses
The use of exploratory factor analysis was deemed appropriate for several reasons. First, as detailed previously, different factor structures have been found across studies (Cohen, et al., 1996;Hanson, 1992;McMahon & Harvey, 2006). Second, to our knowledge this is the first study that has employed MES with a sample of physicians. Third, our sample was drawn from Thailand which is clearly a different cultural context from the United States where MES was developed. Fourth, confirmatory factor analysis was not advisable due to the low number of items comprising two of the subscales of MES. At least three items per subscale are needed for the validation of a multidimensional scale, (Bollen, 1989;Marsh & Hau, 1999), while MES contains two subscales consisting of merely two items. Thus, seven EFAs, (one for each scenario) were conducted to assess whether the original factor structure (Reidenbach & Robin, 1990) held for a sample of physicians. Many of the earlier factors analytic studies of the MES employed principal components analysis (PCA) and varimax rotation PCA strictly speaking is a data reduction technique. Since our aim was to explore the underlying factor structure of the already distilled 8-item MES, maximum likelihood extraction is well suited for this purpose (Costello & Osbourne, 2005 27 criterion validity). To this end, a single item measuring whether the presented scenario was deemed ethical or unethical on a 7-point Likert scale was included as the outcome variable. Seven multiple regressions, using the enter method, were conducted; one for each scenario using the data obtained from the physicians.

Results
Bartlett's test of sphericity was significant for all seven EFA's (p<.001) and the KMO measure of sampling adequacy was satisfactory (Norusis, 1988; see Table 2). The EFAs show some varying results, although four out of seven scenarios (i.e., scenarios 1, 2, 4, and 6), all items are clustered into one overall factor. However, there is some support for the existence of a two-factor structure as well. When two factors were identified (scenario seven), the first two factors were combined (Moral Equity and Relativism), while the two items intent to measure Contractualism form to one factor. However, in this case the combined factor is by far the most important factor (i.e., it explains the most variance). Also, scenarios 3 and 5 showed multiple cross loadings and displayed no clearly interpretable factor solution.
The results of the EFAs suggested the existence of a one-factor solution, or possibly a combination of the first two factors (Moral Equity the Relativism) of the original MES-8. The latter was supported by high inter-factor correlations between the two factors. All correlations between these factors are 0.7 or above (See Table 3). This supports the notion that the two factors are measuring, to a large extent, the same concept. Correlations between Moral Equity and Constructualism or between Relativism and Constructualism did not exceed 0.7. However, all correlations were statistically significant at p<0.05, suggesting that all three factors are strongly related to each other.
Results of the multiple regressions showed that the MES-8 is a significant predictor of the univariate measure of ethics conducted with a sample consisting of physicians. The scale explained between 43% and 62% of the variance in the ethical judgement measure.
The results also showed consistently that Moral Equity was the most important predictor of the univariate ethics measure. In all cases, Moral Equity had the largest standardized Beta, indicating its relative importance.
Combining the results of the EFA and the multiple regressions, it can be assumed that a large amount of variance is shared between at least the first two subscales (Moral Equity and Relativism), which can be combined into one subscale. Also, Contractualism can be considered a part of one overarching factor, combining all three original subscales. The results of the multiple regressions showed that if Contractualism was regarded as a separate factor, its impact on the univariate ethics measure was small and in most cases not significant.

Discussion
The purpose of this study was to assess the validity of the MES-8 when completed by a sample of physicians. Our results did not support the original three-factor structure of the MES-8 as developed by Reidenbach and Robin (1990), but rather a one-factor solution, and to a lesser extent a two-factor solution in which Moral Equity and Relativism were combined. Four out of the seven scenarios were used and provided support for the existence of one dominating factor. This is consistent with the findings of other researchers who found weak evidence for the multidimensional structure of the MES (McMahon & Harvey, 2007;Nguyen & Biderman, 2008;Tansey, Brown, et al., 1994). Reidenbach and Robin (1990) suggest that single factor findings may reflect a construct such as ethical judgment is being tapped that is characterized by overlapping ethical philosophies. Indeed, the results of our regression analyses found that the MES 8 was a strong predictor of the ethical judgement of Thai physicians. A two-factor solution similar to that found in the present study was also observed by Reidenbach et al. (1991).
Among other suggestions to explain the existence of a two-factor solution, the authors mentioned the natural relationship expected between what people perceive to be culturally acceptable and what is fair or just. If the meaning of fairness is closely related to what is deemed culturally acceptable, the two involved subscales should be closely related. In Thailand, Buddhism is the dominant religion, there may exist a more ambiguous sense of justice that would incorporate what is culturally acceptable. The ultimate aim of Buddhism is non-violence, there may be a broader acceptance of varying cultural paths leading to the end. For example, the Itivuttaka states that "all the means that can be used as bases for right action are not worth the sixteenth part of the emancipation of the heart through love. This takes all the others up into itself, outshining them in glory" (Hooks, 2007, p. 41). In the current exploratory study, the factor structure of the MES 8 was not invariant across the seven scenarios employed. This is consistent with recent studies by McMahon and Harvey (2007) and Nguyen and Biderman (2008) who found that scenario based factors played a significant role in understanding their MES data. It is important to note that in the present study, we used a grounded approach to scenario selection, to ensure that the ethical dilemmas articulated in each scenario were salient to physicians.
Our scenarios were derived from focus group sessions in which physicians were asked to describe their most commonly encountered ethical dilemmas (see Malloy et al., 2008). As This study, did not intentionally vary the presented situations based on any particular ethical principle or moral dimension. However, given the results found in this initial study, follow up investigations could vary and assess scenario based factors in a systematic way.

Conclusion
In conclusion, for a sample of physicians the MES-8 used in this study was dominated by one general factor. Even though the existence of Contractualism as a separate dimension has been acknowledged in this study as well as previous studies, its usefulness is limited due to the fact that it is comprised of only two items (Bollen, 1989). Furthermore, compared to the combined subscale of Moral Equity and Relativism, Contractualism has little power on predicting ethical judgment. The results of our study mirror those who have used the MES-8 within a business context. The MES-8 is seen as a valid instrument in assessing the ethical decision making of physicians.