Journal of Academic Ethics

, Volume 9, Issue 3, pp 235–256

Does Education Influence Ethical Decisions? An International Study

Authors

    • Gabelli School of BusinessRoger Williams University
  • Caryn L. Lecca
    • Gabelli School of BusinessRoger Williams University
  • Jennifer C. Murphy
    • Gabelli School of BusinessRoger Williams University
  • Elizabeth M. Sturgis
    • Gabelli School of BusinessRoger Williams University
Article

DOI: 10.1007/s10805-011-9134-3

Cite this article as:
Bernardi, R.A., Lecca, C.L., Murphy, J.C. et al. J Acad Ethics (2011) 9: 235. doi:10.1007/s10805-011-9134-3

Abstract

This study examined whether having attended a public, private or religious affiliated grade and/or high school influenced a college student’s ethical decision making process. We also examined whether having taken an ethics course in college influences a student’s ethical decision making process. Our sample included 508 accounting students (237 men and 271 women) from Albania, Ecuador, Ireland and the United States. Our analyses indicated no differences in ethical decision making that associated with either grade-or-high-school education. While our data showed no difference in the reported attitudes between students from Ecuador and the United States after controlling for social desirability response bias, we found significant differences between the attitudes students from the United States and students from both Albania and Ireland. While gender was also significant for six of our seven scenarios, social desirability response bias was significant in all of our scenarios.

Keywords

Primary and secondary educationEthics course

Introduction

Many researchers believe that religiosity associates with a person’s moral identity. Vitell et al. (2009, p. 602) define religiosity as “the degree to which an individual is a religious person apart from his/her particular religious beliefs and the way that those beliefs are manifested”. Vitell et al. conclude that religiosity along with self-control has a direct influence on decision-making. Consequently, ethical decision making could be influenced by the environment of an individual’s education (i.e., public, private or religious affiliated). However, studies on the associations among religion, religiosity and ethics produced mixed results. For example, Kennedy and Lawton (1998) found that students at Roman Catholic universities did not differ from public university students in ethical decision-making.

Different countries have different religious and cultural beliefs that could influence ethical ideals, which could also associate with the level of corruption (Transparency International 2008) in a country. Ethics researchers normally use self-reported behavioral answers to circumstances as a research tool; these responses could reflect social desirability response bias. This bias suggests that a person is more likely to answer ethical questions in a way that is consistent with the beliefs of their society and not necessarily how they would actually act.

While Kennedy and Lawton examined the environment of university education, we continue this stream of research by examining whether the environment of primary or secondary education has an effect on ethical decision making in business practices by adding multiple countries to our sample and controlling for social desirability response bias. Our survey was based on Kennedy and Lawton’s (1998) questionnaire that they developed using information from prior studies. Unlike these prior surveys, our questionnaire was not gender specific (i.e., we did not indicate the gender of the person(s) in the scenario).

Literature Review

Definition

Religion typically refers to a specific faith such as Roman Catholic, Methodist, Baptist, Lutheran, Muslim, Judaism, etc. On the other hand, religiosity has been defined in several ways. For example, Barnett et al. (1996) used the strength of one’s religious beliefs; Cornwall et al. (1986) list terms such as cognition, emotional attachment/feelings about religion, and behavior (i.e., church attendance) in their summary of definitions of religiosity. Devonish et al. (2011, p. 169) defined religiosity as “a strong belief in moral principles and doctrines; this belief, in turn, can translate into ethical behaviour in everyday life”. Religiosity has been operationalized in numerous ways; for example, two resent studies used religious affiliation, church attendance, the frequency of prayer/meditation (Conroy and Emerson 2004, p. 387) and questions about being religious (Peterson et al. 2010, p. 576). Peterson et al. noted that religiosity has not been studied to the degree that gender and nationality in ethics research.

Conroy and Emerson (2004, p. 391) note that the influence of “religiosity on ethical attitudes is fairly robust in the literature”. These authors found that having taken religion classes and an ethics course(s) was only significant in two of their 25 scenarios. We suggest that a more protracted exposure to religiosity such as attending religious grade-and/or-high schools (i.e., how we operationalized religiosity) could provide significant results. Consequently, we use a student’s educational environment (i.e., public, private or religious) for grade-and-high school in this research.

Religion and Decision Making

While the purpose of religious instruction is to inculcate a particular set of beliefs about God, the primary reason for teaching religion and ethics is that they are important fields that have considerable academic interest. The religions of the world also provide a framework for what is right and wrong. For example, Ali et al. (2000) discuss the Ten Commandments in the context of Christianity, Judaism, and Islam. Ali et al. (p. 353) suggest that six of the commandments deal with “social and ethical issues”. These authors maintain the Ten Commandments provide a framework for organizational issues that reflect ethical behavior. However, this implies that individuals within organizations must be capable of recognizing ethical issues.

Rest (1986, pp. 5–17) proposed a four-component/step model for ethical decision making and behavior in which an individual must: (1) recognize a moral issue; (2) make a moral judgment; (3) put moral concerns ahead of other concerns by establishing intent, and, (4) act on the moral intent. While Rest noted that individuals may not be capable of completing the entire four-step process, an individual’s ability to recognize a moral issue (i.e., the basis) is essential for Rest’s four-step process to function. With respect to the content of an ethics course in higher education, Callahan (1980, pp. 64–67) proposed five goals: (1) understanding moral views and consequences; (2) recognizing ethical issues—Rest’s first step; (3) eliciting a sense of moral obligation – Rest’s third step; (4) developing analytical skills—Rest’s second step; and, (5) tolerating and reducing disagreement and ambiguity. Consequently, Ali et al.’s (2000); Callahan’s (1980), and Rest’s (1986) research implies a link between religion, the goals of an ethics course, and ethical decision making.

Religiosity and Ethics

Calkins (2000) suggested including religion in business ethics because the stories told about religion shaped the decision-making process the many individuals in the business world. Calkins (p. 348) suggested that the closer one is to religion, the better that person will be at making ethical decisions and suggests that “business ethics ought to reclaim unabashedly its religious traditions”. Parboteeah et al. (2007, p. 389) report that prior studies used an individual’s knowledge of, commitment towards, and/or actions supporting their religion as variables, which are components of Cornwall et al.’s (1986) framework. Parboteeah et al. (2004) maintain that behavior is the strongest reflection of an individual’s religiosity. For example, Gallup Polls (2007) found that countries where religiosity was higher had lower suicide rates. The Gallup survey questioned participants from 67 countries about the importance of religion in a person’s daily life, whether they had been to church in the week prior to the poll, and if they had confidence in religious organizations.

Prior research indicates mixed results concerning the association between religiosity and ethics (Weaver and Agle 2002). Wimalasiri et al. (1996) found that individuals with a higher religious commitment scored higher on Rest’s (1979) Defining Issues Test (DIT). Clark and Dawson (1996) and Wagner and Sanders (2001) noted that religious individuals indicated that unethical acts/scenarios were less ethical. Agle and Van Buren (1999) found that attitudes about corporate social responsibility associated with religious beliefs. Razzaque and Hwee (2002) found that religiosity positively associated with four of their six scenarios.

However, Hood et al. (1996) and Vitell and Paolillo (2003) found no correlation between ethical attitudes and religiosity. Although Conroy and Emerson (2004) hypothesized that participating in religious activities may improve a student’s ethical values, their study indicated that religion and ethics courses do not play a role on a student’s ethical perceptions. Kennedy and Lawton (1998) examined the ethical perceptions of business and non-business students at the college/university level; the students in their sample were enrolled in a public, evangelical, and Roman Catholic institutions. They found no difference between business and non-business student’s self-reported ethical behavior. They also noted that the ethical behavior of students at the Roman Catholic institution were not different from the students at the public institutions. A study in consumer ethics found that religiosity is negatively correlated with consumer ethics (Vitell and Paolillo 2003).

Given the contradictions in prior studies, it is important to continue the study of religiosity and ethics using different approaches to operationalizing religiosity to determine whether an association exists between the two. Additionally, several studies relating to religiosity suggested that future studies consider broader demographics (Conroy and Emerson 2004), different populations and settings (Vitell and Paolillo 2003), and account for national influences (Parboteeah et al. 2007). We hypothesize that religiosity as demonstrated by the educational environment in grade-and-high school (i.e., religious, private, or public) will associate with ethical decision-making; our first hypothesis can be stated as (null form):
  1. H1:

    The educational environment (i.e., whether it was public, private, or religious) in grade and/or high school will not play a role in ethical decision-making.

     
Rest’s Defining Issues Test (1979) (DIT) has been used as a measure of ethical sensitivity in accounting and auditing research. For example, Bernardi (1994) found that managers who scored higher on the DIT detected fraud at a higher rate when provided client integrity data than the managers who were not provided with client integrity data or managers who scored lower on the DIT. Arnold and Ponemon (1991) found that auditors who scored higher on the DIT were also likely to disclose sensitive findings even when management threatened retaliation. Ponemon (1992) found that auditors who scored higher on the DIT were less likely to underreport billable hours in attempt to conceal their inability to meet time deadlines. Ponemon and Gabhart (1993, p. 48) examined ethical sensitivity measured by DIT scores and auditor independence; their findings indicate that:
  1. 1.

    Auditors with relatively low DIT P scores were more likely to violate independence rules than those with higher DITs.

     
  2. 2.

    The existence of penalty, or the likelihood of losing a job . . . interaction effects suggest that penalty is most salient for low DIT auditors.

     
  3. 3.

    Auditors with lower DITs assigned greater importance to economic factors, such as client profitability and litigation than auditors with higher DIT results.

     
However, Bernardi and Bean (2007) found that male students scored significantly lower on the DIT for both Rest’s standardization data (1993) (n = 535) and accounting data (n = 1,635). While Rest’s standardization sample included students in numerous fields of education from grade school through doctoral studies, Bernardi and Bean’s meta-analysis sample included college accounting majors from 13 previously published studies. Given the importance of ethical sensitivity in accounting, Bernardi and Bean’s (2007) finding that accounting-majors’ scores on the DIT, which were consistently lower than the scores of the general population throughout and after college, is disturbing. However, Armstrong (1993) indicated that taking an ethics course can increase a participant’s ethical sensitivity. Armstrong found that students who took an accounting-ethics course scored significantly higher on Rest’s DIT. If Armstrong’s (1993) findings on increasing scores on the DIT and the research from auditing about ethical sensitivity and higher DIT scores can be generalized outside of accounting, then we suggest the following as our second hypothesis (null form):
  1. H2:

    Students who have taken an ethics course will not be more sensitive to ethical issues than students who have not taken an ethics course.

     

Socially Desirable Responding and Ethics

While considerable behavioral research has been done in ethics, there is the concern that a significant portion of this research uses self-reported data (Randall and Gibson 1990). While most research does not mention the potential problems associated with self-reported data, the potential for social desirability response bias does exist—subjects not accurately reporting what they would do in a situation (Geiger and O’Connell 2000; Nyaw and Ng 1994). Weaver and Agle (2002, p. 388) note that problems in many studies occur because the authors use measures that are susceptible to socially desirable responding and suggest using measures that do not elicit this type of response.

Social desirability response bias occurs when individuals respond in a manner that underestimates and/or overestimates the probability they would execute an unwanted and/or popular action (Fernandes and Randall 1992). In an attempt to conform with public norms, individuals may present themselves in a positive light, in spite of their genuine outlooks or definite actions. For example, Gendall et al. (1992, p. 1) hypothesize that “donations to charities are typically over-reported, while less desirable behaviors such as smoking and drinking are likely to be under-reported.” As a personality characteristic, a socially desirable response is often “culturally sanctioned and approved” (Crowne and Marlowe 1960, p. 350). High scorers on a socially desirable response scale reflect a tendency to respond in a socially desirable manner. Social desirability response bias was significantly associated with attitudes about bribery and unethical actions (Bernardi and Vassill 2004; Bernardi et al. 2009). Our third hypothesis tests the association between social desirability response bias and ethical perceptions:
  1. H3:

    Students’ self-reported responses will not be associated with social desirability response bias.

     

Gender and Ethics

Socialization theory proposes that women are trained from a young age to reason differently from men about moral issues (Gilligan 1982). Consequently, women’s choices are more likely to be influenced by the pressures of communal norms and morals (Schoderbek and Deshpande 1996). Nguyen et al. (2008) noted that women and men used different styles of moral orientations to solve moral dilemmas. While, women were more care orientated (i.e., stage three in Rest’s model of moral development), men were more likely to be justice orientated in solving moral dilemmas (i.e., stages five and six—the scored stages for Rest’s P score).

There are studies which found that women tend to behave more ethically than men. For example, in a class where ethics was incorporated into the curriculum, women had a significantly higher increase in moral awareness (i.e., ethical sensitivity) than men (Ritter 2006). Ritter also observed that gender associated with personal integrity and external ethics—“males were less sensitive to ethical issues” (p. 159). Smith and Oakley (1997) and Beltramini et al. (1984) found that ethical issues were a greater concern to female students (i.e., Rest’s first step—recognizing the ethical dilemma). Malinowski and Berger (1996) found that female students responded more ethically to scenarios (i.e., Rest’s second and third steps—list alternatives and chose an action). MacLellan and Dobson (1997) and McDevitt et al. (2007) found that women were more ethically sensitive and that men were more than twice as likely to engage in less moral decisions. Roxas and Stoneback (2004, p. 161) noted that: “when the sample including all countries was tested, males were significantly less ethical than females.

However, there are also studies that indicate that women act as ethically as men. Diekhoff et al. (1999, p. 348) indicated that the importance of an individual’s gender “on cheating is weak, at best, and the difference in the gender distributions of our samples is probably relatively unimportant”. In a study including students from Canada, Ireland, Spain and the United States, Bernardi and Vassill (2004) found no differences in ethical perceptions that associated with gender after controlling for social desirability response bias. Izraeli and Jaffe (2000) found that, although there were significant differences in the means of intention for justice versus care orientations in all five scenarios, gender was only significant in two of the scenarios In both these scenarios, men had a higher intention to disapprove the proposed action (i.e., act ethically); consequently, we state our fourth hypothesis as:
  1. H4:

    Gender will not play a role in ethical decision-making.

     

Culture and Ethics

Transparency International (2008) defined corruption as the misuse of entrusted power for private gain. According to Transparency International, the least corrupt countries in their sample were rated as a 9.4 (New Zealand, Denmark, Finland, and Singapore) and the most corrupt countries as a 1.4 (Somalia and Myanmar). However, this rating system is the opposite of what one would expect (i.e., the least corrupt countries have the highest scores on a corruption index). To make our discussion more interpretable to readers, we subtracted Transparency International’s scores from ten; this inversion now depicts the more corrupt countries with the higher scores. Our transformation (Fig. 1) results in scores of: 8.0 for Ecuador, 6.6 for Albania (i.e., the most corrupt countries in our sample), 2.7 for the United States and 2.3 for Ireland(i.e., the least corrupt countries in our sample). The data in Fig. 1 show a distinct separation between the Corruption Perceptions Indices for the United States and Ireland and Albania and Ecuador. These differences suggest a potential for individuals’ reactions to ethical issues to associate with a country’s level of corruption. For example, Bernardi and Vassill (2004) and Bernardi et al. (2009) found that the propensity to bribe a police officer to avoid a speeding ticket (i.e., an form of corruption) associated with attitudes concerning much larger corporate ethical issues for a sample from Canada, Colombia, Ecuador, Ireland, South Africa, Spain, and the United States.
https://static-content.springer.com/image/art%3A10.1007%2Fs10805-011-9134-3/MediaObjects/10805_2011_9134_Fig1_HTML.gif
Fig. 1

Hofstede’s (2010) and transparency international’s (2009) scores

Hofstede (1991, 112) defines culture as a “set of likely reactions of citizens with a common mental programming. . . . reactions need not be found within the same persons, but only statistically more often in the same society”. Hofstede (1980) provides four dimensions related to national culture: individualism, masculinity, power distance and uncertainty avoidance. A fifth construct (long term—short term orientation) was later added by Hofstede (1991); however, of the four countries in our sample, only the United States currently has a score for this dimension. When developing his cultural constructs, Hofstede (1980) sampled over 100,000 employees of a large multi-national corporation. Although there are concerns about the use of Hofstede’s data because of its age and being from only one corporation, Merritt (2000) and Smith (2002) provide evidence that Hofstede’s work is still relevant. Using data from 9,000 commercial airline pilots (Merritt) and 1,000 staff members of international accounting firms (Smith), these authors found that their individualism and uncertainty avoidance constructs were correlated with Hofstede’s (1980) original constructs of individualism and uncertainty avoidance.

Hofstede (1980, 214) describes individualism as the society’s an overall concern about taking care of one’s self and family. Masculinity is a reflection of a society’s preference “achievement, heroism, assertiveness, and material success” (Hofstede 1984, 83–84). The power distance construct reflects a country’s attitude towards obedience, conformity an accepting a hierarchical order with respect to an individual’s place in the society (Hofstede 1984, 84). Finally, uncertainty avoidance reflects a society’s attitudes towards following the rules, employment stability, and stress/nervousness at work (Hofstede 1984, 118–119). For the four countries included in our study, Fig. 1 shows three of Hofstede’s four initial cultural constructs—the range of scores for masculinity is very narrow and therefore not included. The other three constructs show differences between the two groups of countries (i.e., Ireland and the United States versus Albania and Ecuador); the differences range from 28 (uncertainty avoidance) to 45 (individualism), which should be sufficient enough to distinguish between the two groups.

Given the differences in Corruption Perceptions Indices and cultural constructs in Fig. 1 between the two groups of countries and the results of Bernardi and Vassill and Bernardi et al., one could suggest that individuals from the two groups of countries would perceive ethically questionable actions differently. Our fifth hypothesis is:
  1. H5:

    Ethical perceptions of students from countries with higher CPI (i.e., Ecuador and Albania) will not differ from students from countries with lower CPIs(i.e., Ireland and the United States).

     

Methodology

Sample

The participants in this study include 508 undergraduate accounting students from several universities located in Albania (116 students), Ecuador (105 students), Ireland (74 students) and the United States (213 students) as shown in Table 1. The college level of the participants was coded as two for sophomores, three for juniors and four for seniors (i.e., no freshmen). The data indicate that the students were approximately juniors (i.e., total sample average of 2.9).
Table 1

Sample demographics

 

Albania

Ecuador

Ireland

United States

Total

 

Men

Women

Men

Women

Men

Women

Men

Women

Sample

Sample composition

 Gender

36

80

37

68

46

28

118

95

508

 Age

20.6

20.4

20.6

20.4

20.6

20.5

20.1

20.0

20.4

 College level

3.0

2.8

2.5

2.6

3.0

2.9

2.7

2.7

2.9

Education:

 GS Public

27

76

1

5

0

0

96

70

275

 GS Private

9

4

17

31

0

0

3

4

68

 GS Religious

0

0

19

32

46

28

19

21

165

 HS Public

31

71

0

3

0

0

92

66

263

 HS Private

5

8

19

36

0

0

3

5

76

 HS Religious

0

1

18

29

46

28

23

24

169

Ethics College Course

21

41

1

11

20

18

36

28

176

Percent Ethics Course

58.3

51.3

2.7

16.2

43.5

64.3

30.5

29.5

34.6

SDRB

3.8

3.8

7.0

8.4

4.9

6.9

4.9

5.9

4.8

College level

Sophomore (2), Junior (3), and Senior (4)

SDRB

Average Social Desirability Response Bias Score

For the participants from Albania and the United States, public education in grade (81.8%) and high school (79.0%) represented the majority of the sample. However, in predominately Roman Catholic Ecuador and Ireland religious education in grade (69.8%) and high school (67.6%) represented the majority of the sample. In both Albania and Ireland, about 50% of the students had taken a college ethics course; whereas, only about 30.0% of the college students from the United States and 11.4% of the college students from Ecuador had taken a college ethics course.1

Survey

Our survey requested personal responses regarding ethical decision making among male-and-female college students with and without a religious grade and/or high school education. The information requested in the first section of the survey (Appendix A) asked participants for their gender, home country, what type of grade school they attended, and what type of high school they attended. For grade-and-high-school education, we gave four options for our subjects to pick from: public, private with no religious affiliation, private with religious affiliation, and other. For gender, male students were coded as one and female students were coded as zero. For a student’s response for grade-and-high-school education, each level was coded as one for the option selected (i.e., public, private, religious, or other) and the other three options were coded as a zero. We asked students for their home country because we wanted to ensure that the sample from each country did not include students who were studying abroad (i.e., maintain each country’s distinct culture).

Rettig and Rawson (1963, p. 244) proposed varying the intensity of the following six factors that should influence individual’s propensity to engage in unethical behavior: (1) the expectancy of the gain; (2) the reinforcement value of the gain; (3) the expectancy of the censure; (4) the reinforcement value of the censure; (5) the severity of the offense; and, (6) the individual’s reference group. Kennedy and Lawton (1998) used Rettig and Rawson’s survey to study the association between religiousness and business ethics. In this research, we proposed seven scenarios to our participants—four related to business issues and three related to student issues (Appendix B). We adapted three questions used by Fritzsche and Becker (1984) dealing with business issues of: bribing a government official (p. 169); releasing pollutants (p. 171); and a personal integrity issue (p.173). Our fourth business question was from Jones and Gautschi (1988) that dealt with selling a product that could cause cancer. Our three student questions were adapted from Rettig and Rawson’s original scenarios (pp. 244–245). While Kennedy and Lawton’s questionnaire framed questions where the subject was a high-ranking man, our questions did not specify gender. Consequently, perceptual differences should not have influenced the decision making process. Finally, we used a scale from one (seven) indicating that the participant thought the individual in the scenario definitely would not (definitely would) take the action proposed in the scenario.

The other part of the survey (Appendix C) was Paulhus’ (1986) Impression Management Subscale (IMS) of 20 questions that measure an individual’s propensity to respond to questions in a socially desirable manner. Socially desirability response bias (i.e., responding in a manner one judges to be acceptable in that society) occurs even when questions are asked in the third person. This type of responding to survey questions introduces a systematic error into the study’s findings. The 20 questions are coded alternately; for the odd (even) numbered questions, the socially desirable answers were a one or a two (six and seven) for ‘Not True’ (‘Very True’). For example, the first two questions on Paulhus’ IMS are:
  1. (1)

    Sometimes I tell lies if I have to.

     
  2. (2)

    I never cover up my mistakes.

     

The higher the number of socially desirable answers, the more that student would answer ethical questions in a socially desirable way.

A person might suggest that everyone sometimes tell lies and attempt to cover their mistakes. These individuals have made the point about socially desirable responding; their responses reflect honest answers to questions one and two. Consequently, they would answer six or seven (one or two) to question one (two); these responses would not be counted in their score on the IMS. However, if another person responded either one or two (six or seven) to question one (two), these would be scored as socially desirable responses (i.e., society frowns on telling lies and covering up one’s mistakes). Paulhus reports that the range for Cronbach’s alpha is typically from .75 to .86; Cronbach’s alpha for the current sample was .83.2

The data in Fig. 2 show the distribution of the current sample (Panel A) and Bernardi’s (2006) research on social desirability response bias (Panel B). While the current sample includes four countries (n = 508), Bernardi’s sample was three times as large both in its diversity of countries (12) and sample size (n = 1,525). The graphs in Fig. 2 suggest similar distributions of scores on Paulhus’ Impression Management Subscale. The data indicate that both the current data and Bernardi’s data are positively skewed (0.62 and 0.54 respectively) and have a similar level of kurtosis (2.85 and 3.07 respectively).
https://static-content.springer.com/image/art%3A10.1007%2Fs10805-011-9134-3/MediaObjects/10805_2011_9134_Fig2_HTML.gif
Fig. 2

Social desirability response bias scores. Panel a Current Sample [n = 508], Panel b Bernardi (2006) [n = 1,525]. Current Sample: Albania, Ecuador, Ireland and the United States. Bernardi (2006): Australia, Canada, China, Colombia, Ecuador, Hong Kong, Ireland, Japan, Nepal, South Africa, Spain, and the United States Axis: Horizontal—social desirability response bias scores Vertical—number of students in the sample

For the students from Albania and Ecuador, a person translated the survey into each country’s language; then, the survey was back-translated into English by a second person. For Question Seven, which specified a $10,000 bribe, we translated the dollar amount into the currency of Albania and Ireland just prior to sending the surveys to these countries.3 The surveys were given to the contact person who was a professor teaching at a university in the country. The contact person distributed the surveys to business students and returned the completed surveys to the authors.

Each student’s country was coded in a series of four dichotomous indicator variables (i.e., Albania, Ecuador, Ireland and the United States); these variables were coded as a one or zero depending on the student’s home country. For example, had the student been from Albania, the column for Albania in our spreadsheet would have been coded as one and the columns for the other three countries would have been coded as zero. For our analysis, we used the United States was used as our control variable because the studies from which our survey was developed had samples from the United States.

Analysis

Overview

The data indicate that there was no correlation between social desirability response bias and education (Table 1). However, there was an association between gender and socially desirable response bias; female students’ scores were significantly higher (p = 0.001) than male students’ scores. The sample from Albania comprises 31.7 % of our sample and there was no difference between the genders for Albania—both male-and-female students had an average score of 3.8. For the other three countries, there was a significant difference between the average scores for male-and-female students from: Ireland (4.9 versus 6.9, p = 0.006), Ecuador (7.0 versus 8.4, p = 0.033) and the United States (4.9 versus 5.9, p = 0.027).

The data in Table 2 provide an overview of the students’ responses to the survey questionnaire. Panel A shows the students’ average responses by country; however, rather than listing the countries alphabetically, we list them by their average responses for the ‘sum of responses’. This list ranks the countries from highest response (i.e., most likely to approve of the proposed action) to lowest response (i.e., least likely to approve of the proposed action). Except for questions two and four, the order from highest to lowest was Albania, Ireland, the United States and Ecuador. While there is one exception (highlighted) to Albania (Ireland) being the highest (second), there were no exceptions to Ecuador being the lowest. Consequently, one would anticipate that the responses for the students from Albania and Ireland (Ecuador) would be significantly higher (lower) than the responses for the students from the United States in our regression models.
Table 2

Survey responses by country, gender and having taken an ethics course

https://static-content.springer.com/image/art%3A10.1007%2Fs10805-011-9134-3/MediaObjects/10805_2011_9134_Tab2_HTML.gif

Panel B shows the students’ responses by gender. Except for question one (slight negative difference—highlighted) and question four (no difference), female students’ reported intentions were lower (i.e., less likely to approve of the proposed action) than male students’ responses. Consequently, one would anticipate that a student’s gender would be a significant variable in our regression models. Panel C shows the students’ responses by whether or not the student had taken an ethics course in college. It is apparent that the responses of students who had taken an ethics course in college were slightly higher for all questions except Question Two. This indicates that they were less sensitive to the ethical problem in the proposed action (i.e., the opposite of what one would have anticipated).

Student Issues Questions

The data in Table 3 show the results of the stepwise regression models for the three research questions dealing with student issues.4 Question one involved a student who is working in the admissions department at a university and has to consider taking money in order to pay off a personal debt. The model indicates that hypothesis three dealing with SDRB (p < 0.000) and hypothesis five dealing with country (Albania: p < 0.000 and Ireland: p = 0.004) can be rejected. As anticipated, students who responded in a more socially desirable manner (SDRB) were less likely to indicate that they would “borrow” university funds (i.e., act more ethically). The average responses for the students from the United States and Ecuador did not differ after controlling for SDRB, which was not anticipated. The average responses for the students from Albania (anticipated) and Ireland (not anticipated) indicate that these students were more likely to “borrow” university funds (i.e., act less ethically) than the students from the United States (our control group). We failed to reject our hypotheses dealing with education (H1), ethics courses (H2), and gender (H4).
Table 3

Stepwise regression models for student issues

Term

Coefficient

T Stat

Prob T

Partial R2

Adjusted R2

Question 1—Admissions student “borrowing” university funds

 Intercept

2.99

26.26

<0.000

.336

 Albania

0.85

13.24

<0.000

.295

 

 SDRB

−0.08

−4.77

<0.000

.031

 

 Ireland

0.21

2.92

0.004

.010

 

Question 5—Accounting club treasurer needs money for surgery

 Intercept

3.37

21.18

<0.000

.074

 SDRB

−0.08

−3.38

0.001

.045

 

 Gender

0.23

3.14

0.002

.012

 

 Albania

0.28

3.02

0.003

.011

 

 Ireland

0.21

2.01

0.045

.006

 

Question 6—Fraternity treasurer’s friend has serious gambling debts

 Intercept

3.11

24.32

<0.000

.086

 SDRB

−0.08

−3.65

<0.000

.054

 

 Albania

0.32

3.96

<0.000

.020

 

 Gender

0.19

2.79

0.006

.012

 

Albania

Student was from Albania

Ireland

Student was from Ireland

Gender

Coded 1 for male students and 0 for female students

SDRB

Students’ scores on Paulhus’ measure of socially desirable responding.

Question five involved the treasurer of the university’s accounting club in a position where he/she must decide to take money from the club’s funds in order to pay for his/her operation. We chose this question because there is a more serious reason behind taking the proposed course of action than in the first question. The model indicates that hypothesis three dealing with SDRB (p = 0.001), hypothesis four dealing with gender (p = 0.002), and hypothesis five dealing with country (Albania: p = 0.003 and Ireland: p = 0.045) can be rejected. Students who responded in a more socially desirable manner (SDRB) and female students were less likely to indicate that they would to take the club’s funds for the required surgery (i.e., act more ethically). The average responses for the students from the United States and Ecuador did not differ after controlling for SDRB (not anticipated). The average responses for the students from Albania (anticipated) and Ireland (not anticipated) indicate that they were more likely to take the club’s funds for the required surgery (i.e., act less ethically) than the students from the United States (our control group). We failed to reject our hypotheses dealing with education (H1) and ethics courses (H2).

Question six involved the treasurer of a fraternity who is in a situation where he/she must decide to take money from the fraternity in order to help a friend pay off his/her gambling debt and avoid physical injury. This question differs from the first and second questions because it involves putting oneself in risk for another person. The model indicates that hypothesis three dealing with SDRB (p < 0.000), hypothesis four dealing with gender (p = 0.006), and hypothesis five dealing with country (Albania: p < 0.000) can be rejected. Students who responded in a more socially desirable manner (SDRB) and female students were less likely to indicate that they would take funds to pay the gambling debt (i.e., act more ethically). The average responses for the students from the United States and Ecuador did not differ after controlling for SDRB (not anticipated). The average responses for the students from Albania (anticipated) indicate that they were more likely to take funds to pay the gambling debt (i.e., act less ethically) than the students from the United States (our control group). We failed to reject our hypotheses dealing with education (H1) and ethics courses (H2).

Business Issues Questions

Question two involved a conflict of interest for a businessperson who must decide whether to bribe a businessperson to sell a product. The model indicates that hypothesis one dealing with ethics courses (p = 0.001), hypothesis three dealing with SDRB (p < 0.000), hypothesis four dealing with gender (p < 0.000), and hypothesis five dealing with country (Albania: p = 0.015 and Ireland: p = 0.003) can be rejected. While we can reject hypothesis one, it was for the wrong reason; students who had taken a college ethics course were more likely to bribe a businessman to sell a product (i.e., act less ethically). Students who responded in a more socially desirable manner (SDRB) and female students were less likely to bribe a businessman to sell a product (i.e., act more ethically). The average responses for the students from the United States and Ecuador did not differ after controlling for SDRB (not anticipated). The average responses for the students from Albania (anticipated) and Ireland (not anticipated) indicate that they were more likely to bribe a businessman to sell a product (i.e., act less ethically) than the students from the United States. We failed to reject our hypothesis dealing with education (H1).

Question three involved a situation in which a company’s president must decide whether to release pollutants into a river to manufacture a profitable product. This question differs from Question Two because it would involve harming the environment. The model indicates that hypothesis three dealing with SDRB (p = 0.008), hypothesis four dealing with gender(p = 0.001), and hypothesis five dealing with country (Albania: p < 0.000) can be rejected. Students who responded in a more socially desirable manner (SDRB) and female students were less likely to indicate that they would release pollutants into a river to manufacture a profitable product (i.e., act more ethically). The average responses for the students from the United States and Ecuador did not differ after controlling for SDRB (not anticipated). The average responses for the students from Albania (anticipated) indicate that they were more likely to release pollutants into a river to manufacture a profitable product (i.e., act less ethically) than the students from the United States. We failed to reject our hypotheses dealing with education (H1) and ethics courses (H2).

Question four involved a situation in which manufacturing a high profit product that could cause cancer. Question four takes the harm done up one level from question three; instead of just causing injury (i.e., Kennedy and Lawton’s question); we changed the consequence to a severe disease. The model indicates hypothesis two dealing with ethics courses (p = 0.033), hypothesis three dealing with SDRB (p = 0.019), hypothesis four dealing with gender (p = 0.001), and hypothesis five dealing with country (Albania: p < 0.000) can be rejected. Students who had taken and ethics course were more likely to indicate that they would manufacture a high profit product that could cause cancer (i.e., act less ethically), which was not anticipated. (i.e., act more ethically). Students who responded in a more socially desirable manner (SDRB) and female students were less likely to indicate that they would manufacture a high profit product that could cause cancer The average responses for the students from the United States and Ecuador did not differ after controlling for SDRB (not anticipated). The average responses for the students from Albania (anticipated) indicate that they were more likely to manufacture a high profit product that could cause cancer (i.e., act less ethically) than the students from the United States. We failed to reject our hypothesis dealing with education (H1).

Question seven involved a situation in which a junior auditor realizes that a client owes $10,000 in back taxes. However, his/her audit manager has ordered him/her not to report the back taxes. The model indicates that hypothesis three dealing with SDRB (p < 0.000) and hypothesis four dealing with gender (p < 0.000) can be rejected. Students who responded in a more socially desirable manner (SDRB) and female students were less likely to indicate that they would follow the audit manager’s orders to keep this quiet (i.e., act more ethically). We failed to reject our hypotheses dealing with education (H1), ethics courses (H2), and country (H5).

Sum of Responses to Questions

Rest’s (1979) Defining Issues Test (DIT) asks individuals to read six (three) scenarios in the long (short) version and then rate the importance of 12 considerations at the various stages of moral development. In each group of considerations, there are three or four at Stages Five and Six (i.e., the principled stages) of moral development. After rating the 12 considerations, individuals are then asked to pick and rank the top four considerations. The four considerations are used to determine the partial P score for that scenario taking into account the considerations at the principled stages and their ranking by the individual. The P score on Rest’s (1979) DIT is the total of the partial scores for each scenario. As a result, the P score represents the sum of each individual’s principled considerations over the six (three) scenarios of the long (short) version of the DIT. The range of an individual’s P score on the DIT is from Zero (e.g., no considerations at the principled levels) to 90 (e.g., all responses at the principled level).

Using a concept from Rest’s DIT, we summed the students’ responses for all seven of our questions for an index that would evaluate students’ tendency to act ethically over a range of scenarios. As shown in Table 2, the average for the sum of responses was 16.5 for female students and 19.0 for male students (i.e., lower scores indicate intentions to act more ethically). The model in Table 5 indicates that hypothesis two dealing with ethics courses (p = 0.034), hypothesis three dealing with SDRB (p < 0.000), hypothesis four dealing with gender (p < 0.000) and hypothesis five dealing with country (Albania: p < 0.000 and Ireland: p = 0.001) can be rejected. While we can reject hypothesis one, it was for the wrong reason; students who had taken a college ethics course were more likely to ascribe to the proposed actions (i.e., act less ethically) than students who had not taken an ethics course. Students who responded in a more socially desirable manner (SDRB) and female students were less likely to ascribe to the proposed actions (i.e., act more ethically). The sum of the average responses for the students from the United States and Ecuador did not differ after controlling for SDRB, which was not anticipated. The sum of the average responses for the students from Albania (anticipated) and Ireland (not anticipated) indicate that they were more likely to ascribe to the proposed actions (i.e., act less ethically) than the students from the United States (our control group). We failed to reject our hypothesis dealing with education (H1).

Additional Analysis

As noted earlier in our analysis (Table 2), the average data for the participants from each country suggest that Ecuador should have been a significant indicator variable in five of the seven models (i.e., questions 1, 2, 5, 6, and 7—Tables 3 and 4) and model for the sum of the questions (Table 5). However, the indicator variable Ecuador was not significant in any of these models. One explanation for this was that the average level of SDRB for the participants from Ecuador was significantly higher (p < 0.001) than for participants from the United States. To test the validity of this explanation, we reanalyzed the data without including SDRB in the modeling process (Table 6). The columns labeled ‘with SDRB’ include the partial R2 data from our prior models (Tables 3, 4, and 5) that included SDRB. The columns labeled ‘without SDRB’ show the results of our additional analysis that did not include SDRB as an independent variable. Our additional analysis reflects our initial assessment from the Table 2—Ecuador was a significant indicator variable. Had we not controlled for SDRB, we would have asserted that, for five of the seven models for our questions (i.e., questions 1, 2, 5, 6, and 7) and model for the sum of the questions, the average responses for the students from Ecuador were more ethical (i.e. lower) than the responses for the students from the United States. Consequently, had we not controlled for SDRB, we would have concluded that the students from Ecuador were more likely to respond in a more ethical manner than the students from the other three countries for five questions of our seven (i.e., except for questions three and four).
Table 4

Stepwise regression model for business issues

Term

Coefficient

T Stat

Prob T

Partial R2

Adjusted R2

Question 2—Company president bribing local businessman

 Intercept

3.79

23.77

<0.000

.154

 Gender

0.48

6.60

<0.000

.092

 

 SDRB

−0.10

−4.28

<0.000

.045

 

 Ireland

0.30

2.97

0.003

.009

 

 Albania

0.22

2.44

0.015

.008

 

Question 3—Vice president deciding whether to release pollutants

 Intercept

2.79

23.39

<0.000

.289

 Albania

0.79

11.21

<0.000

.254

 

 Gender

0.22

3.47

0.001

.027

 

 SDRB

−0.06

−2.68

0.008

.016

 

Question 4—Whether to market high profit product that may cause cancer

 Intercept

2.41

24.99

<0.000

.233

 Albania

0.62

9.89

<0.000

.199

 

 Gender

0.16

3.23

0.001

.020

 

 SDRB

−0.04

−2.36

0.019

.008

 

 Ethics

0.11

2.31

0.022

.006

 

Question 7—Audit client owes $10,000 in back taxes

 Intercept

3.59

24.99

<0.000

.105

 SDRB

−0.13

−5.64

<0.000

.070

 

 Gender

0.33

4.59

<0.000

.035

 

Albania

Student was from Albania

Ethics

Coded 1 (0) for students who had (not) taken an ethics course in college.

Gender

Coded 1 for male students and 0 for female students

Ireland

Student was from Ireland

SDRB

Students’ scores on Paulhus’ measure of socially desirable responding.

Table 5

Stepwise regression model for the sum of students’ responses

Term

Coefficient

T Stat

Prob T

Partial R2

Adjusted R2

Sum of responses to all seven questions

 Intercept

22.30

38.50

<0.000

.310

 Albania

3.23

9.74

<0.000

.159

 

 SDRB

−0.53

−6.29

<0.000

.084

 

 Gender

1.62

6.12

<0.000

.055

 

 Ireland

1.20

3.23

0.001

.012

 

Albania

Student was from Albania

Gender

Coded 1 for male students and 0 for female students

Ireland

Student was from Ireland

SDRB

Students’ scores on Paulhus’ measure of socially desirable responding.

Table 6

Stepwise regression models for student issues

https://static-content.springer.com/image/art%3A10.1007%2Fs10805-011-9134-3/MediaObjects/10805_2011_9134_Tab6_HTML.gif

SDRB entered the model ahead of gender for the models for questions five and six (Table 3) and seven (Table 4); consequently, gender becomes a more significant variable in the models that did not consider SDRB (Table 6). This is consistent with Bernardi’s (2006) research in which he found that female students consistently responded in a more socially desirable manner. The Table 1 data indicate that, except for the students from Albania, female students responded in a more socially desirable manner than their fellow male students for each country. Consequently, our research findings with respect to country differences and gender would also have been misstated had we not controlled for SDRB in this research.

Conclusion

Our research examined whether ethical perceptions were influenced by the environment for education in grade-and-high school, having taken an ethics course in college, social desirability response bias, and gender. We used the participants’ educational experience in grade-and-high school as a surrogate for religiosity in this research. The environment for grade-and-high school education (i.e., our surrogate for religiosity) was not significant in any of our seven scenarios or the sum of the responses to the seven scenarios. A more bothersome finding was that having taken a college ethics course was a factor in only one of our four business scenarios—this one with the opposite sign to what one would have anticipated. Consequently, our data indicate that students who had taken a college ethics course were not more likely to respond to our scenarios in a more ethical manner.

The results of our research are consistent with those of Kennedy and Lawton (1998) with respect to differences in educational environment not playing a role on a student’s ethical perceptions. While Kennedy and Lawton operationalized religiosity as attending a public, evangelical, or Roman Catholic college, the current study operationalized religiosity as attending a public, private or religious grade and/or high school. One possible cause of the lack of a significant contrast could have been the result of our sample. While the sample from the United States and Ecuador had participants from all three educational environments, the sample from Albania (Ireland) did not include students who had experienced a religiously-influenced (public) grade and/or high school education.

The results of our research are also consistent with those of Conroy and Emerson (2004) with respect to ethics courses not playing a role on a student’s ethical perceptions. One possible cause of the finding that having taken an ethics course did not affect ethical decision making was that the percent of students who had taken an ethics courses was higher for Albania and Ecuador. While over 50.0% of the students from Albania and Ireland had taken a college ethics course, only 11.4 (30.0) of the students from Ecuador (the United States) had taken a college ethics course. While the percent of students from Ireland and Albania who took a college ethics course is higher (p < 0.000) than for the students from the United States, the percent of students from Ecuador who took a college ethics course is lower (p = 0.002) than for the students from the United States. Consequently, the students from the country who had the lowest percent of having taken an ethics course also consistently reported the lowest average probability of acting unethically in our seven scenarios. We indicated that having taken a college ethics course relates to the curriculum at the participating universities (See endnote 1). Consequently, students who are required to take ethics courses as part of their curriculum may not ‘take away’ a similar level of ethical sensitivity from these ethics courses as students who take an ethics course as electives (i.e., self select into them).5 While we have no evidence to support this premise, this does provide an interesting area for future research.

Our research also found that social desirability response bias (SDRB) was significant in all seven of our questions. SDRB was also a significant factor in our overall model that summed the students’ responses for the seven individual questions. In all cases, students who responded in a more socially desirable manner (i.e., higher SDRB scores) were less likely to report a willingness to engage in the proposed action (questions one through six) or follow their manager’s instructions not to report the overdue taxes (question seven). While there was no gender difference for socially desirable responding between the male-and-female students from Albania, there was a significant difference between the male-and-female-students’ scores from Ireland, Ecuador and the United States. Consequently, our research finding that gender was significant in the models for six of our questions and in the model for the sum of responses becomes more robust given our control for social desirability response bias (Nyaw and Ng 1994). In the additional analysis section, we demonstrated how our research findings would have been misstated had we not controlled for social desirability response bias (Table 6). These finding suggest that ethics research should directly control for SDRB.

We grounded our fifth research hypothesis on Transparency International’s Corruption Perceptions Index (CPI) and Hofstede’s cultural constructs, which suggested that Ireland and the United States (i.e., one group) and Albania and Ecuador (the other group) should have similar ethical perceptions. While our expectations for the students from Albania and the United States supported our initial expectation, this was not the case for the students from Ecuador and Ireland. Our data (Tables 2, 3 and 4) indicate that rather than perceiving unethical situations as being more acceptable than the students from the United States, the students from Ecuador responded with similar opinions as the students from the United States for all seven of our scenarios after controlling for social desirability response bias. Our data (Tables 2, 3 and 4) also indicate that rather than perceiving unethical situations in a similar way as students from the United States, the students from Ireland perceived unethical situations as being more acceptable than the students from the United States for four of our seven scenarios. One potential reason for this difference is that, while both Hofstede and Transparency International surveyed international business executives, we surveyed undergraduate accounting students. This difference provides an opportunity for future researchers to test our premise concerning corruption and ethical perceptions on a different group of countries.6

Our primary limitation dealt with sample composition. While our data come from four countries, we believe that extending the research to other countries would provide more generalizable results.

Footnotes
1

This difference relates to the curriculum at the participating universities.

 
2

For a more detailed explanation of socially desirability response bias see Paulhus (1986) pages 17 to 51.

 
3

In March of 2000, the US dollar became the official currency of Ecuador.

 
4

Stepwise regression was our initial modeling process because we wanted to determine the order that the variables went into our model and their individual contribution to that model’s explanatory power (i.e., each variable’s partial R2). We used the regression models provided in JMP (SAS Institute, 2009) statistical and discovery software in their analysis.

 
5

To test the premise that the sample from Ecuador ‘drove’ our regression results with respect to having taken a college ethics course, we removed the sample from Ecuador and reanalyzed our data. The results of our modeling processes for the reduced sample were the same—having taken a college ethics course did not result in students intending to behave more ethically in our scenarios.

 
6

For the countries in our study, Gallup’s religiosity scores are Ecuador (67), Ireland (63), and the United States (61). While Albania was not included in Gallup’s Poll, the religiosity scores for all of the 14 former Soviet Union countries are 52 (Uzbekistan) and below. Consequently, Gallup’s measure of religiosity would not have provided significant results.

 

Acknowledgments

The authors thank Kim Zamojcin for her editorial assistance on this paper.

Copyright information

© Springer Science+Business Media B.V. 2011