Religiousness and Ethical Behavior

Contents

1 Introduction to religion and ethics

2 Hypotheses
2.1 For religious people only? Religious participation promotes moral behavior. . . except for non-believers
2.2 The impact of religiousness languishes when the victim is not thy neighbor, but thy society

3 Data description and summary statistics
3.1 Independent variables
3.2 Dependent variables
3.3 Control variables

4 Results
4.1 H1: Interaction between different religious dimensions
4.2 H2: Context-dependent judgments of unethical behavior

5 Conclusion

6 Discussion

7 References

A Previous studies

B Summary Statistics

C Regression results

1 Introduction to religion and ethics

Though the discussion about the determinants of ethical decision making has been a constant area of interest within economics and business science for about 30 years (KishGephart, Harrison, and Trevino, 2010), it has revived especially due to recent accounting scandals such as Enron and WorldCom as well as the recent financial crisis (Walker, Smither, and DeBode, 2012).

Although moral behavior may lead to losses in efficiency - such as paying higher wages -, ethical misconduct can also be costly for organizations in form of lost reputation and business, as well as for entire economies, for corruption dampens economic growth due to lower investments and misguided incentives (see, for example, Mauro, 1995). Generally, moral laws that enhance social cooperation can promote market efficiency and solve collective action problems (Uslaner, 1999; Shleifer, 2004). Thus, numerous studies investigated the factors that favor ethical decision-making. However, as moral behavior is a complex concept and its assessment difficult, the issue has not yet been resolved and research still struggles to reliably predict an individual’s ethical decision (Kish-Gephart, Harrison, and Trevino, 2010).

To elucidate the determining factors influencing moral decision making, Bommer, Gratto, Gravander, and Tuttle (1987) introduced a behavioral model that, among others, differentiated between social, work and legal environment, as well as individual factors, for instance demographic characteristics and status. In a similar approach, Trevino (1986) developed a ”person-situation interactionist” model that combined situational and individual moderators. Here, individual factors include ego-strength, field dependence and locus of control whereas the immediate job context, organizational culture and characteristics of the work are summarized as situational factors.

Most empirical studies define situational factors as variables that shape the decision maker’s environment and include her peer group, institutional incentives, organization size and hierarchy levels, whereas individual variables refer to characteristics such as sex, age, ethnic origin and education or employment level (Ford and Richardson, 1994). Being an individual factor of ethical behavior, many times and across disciplines, religiosity has been proposed to considerably motivate an individual’s ethical attitudes and actions (Clark and Dawson, 1996; Weaver and Agle, 2002).

For Christians, the Ten Commandments inter alia determine what is to be considered ethical, and the rules that have to be obeyed. For Buddhists, the five precepts provide inter alia a basis of moral principles. These are just two examples of various religions that exhibit specific values and moral rules their followers should live by (see, for example, Uslaner, 1999; Ali, Camp, and Gibbs, 2000; Parboteeah, Hoegl, and Cullen, 2008).

As Weaver and Agle (2002) state, religions generally proclaim a specific attitude in terms of what is right and what is wrong, and are thereby expected to loom large on ethical decisions. Moreover, these moral judgments claim absoluteness and membership implies compliance with them ”on the basis of divine origin” (p. 359 Clark and Dawson, 1996) and faith, rather than on reason (Vitell and Paolillo, 2003). On the other hand, the claim of absoluteness does not apply to non-adherents (De George, 1986).

Yet surprisingly, empirical studies do not provide conclusive evidence supporting the proposed impact of religiosity on ethical decision making. The issue appears to be more complex than first conjectured (Weaver and Agle, 2002).

Several studies have investigated the influence of religiousness on numerous forms of behavior, such as drug abuse, cheating in exams and participation in voluntary work¹. Tan and Vogel (2009) showed in an experimental game that an agent’s religiousness has a positive impact on trust towards others, on trust placed in her, as well as on her trustworthiness.

With respect to business ethics, Hegarty and Sims (1978) found in a laboratory experiment no significant positive influence of religious value orientation on grad students’ decisions to refuse kickback payments. However, adding a threat of punishment led to a lower extent of misconduct. Since many religions embody a concept of afterlife punishment, this result suggests a stronger impact of situation-dependent punishment than of prospective punishment. Similarly, Kidwell, Stevens, and Bethke (1987) did not find religious denomination or church attendance to influence the understanding of moral behavior in a sample of 100 managers and Ibrahim, Howard, and Angelidis (2008) could not show religious managers to care more about corporate ethics, while Cullen, Parboteeah, and Hoegl (2004) and Longenecker, McKinney, and Moore (2004) demonstrated religious executives to be less prone to justify unethical decisions. McNichols and Zimmerer (1985) conjectured that religious undergrad students emphasize disapproval towards unethical, yet somewhat venial, behavior more strongly than their non-religious fellow students². Those that attended church more regularly disapproved of dubious business practices on average more and business students that indicated strong religiousness were also more concerned with corporate social responsibility (CSR) than less religious students (Conroy and Emerson, 2004; Angelidis and Ibrahim, 2004; Ibrahim, Howard, and Angelidis, 2008). For a sample of MBA students, however, the findings were again inconclusive. Neither religious upbringing nor religious practice were significant or crucial factors in predicting the notion of CSR; only religious beliefs regarding charity and responsibility towards one’s neighbor were found to be positively, albeit moderately, correlated with CSR attitudes (Agle and Van Buren, 1999). Brammer, Williams, and Zinkin (2007) investigated a large cross-country sample and concluded that religious denominations could not be linked to a broader notion of CSR. Analogous to CSR, the relationship between religiosity and consumer ethics was not significant (Vitell and Paolillo, 2003).

When distinguishing between intrinsic and extrinsic religiousness, where intrinsically religious people lead their lives in accordance with their religious beliefs and extrinsically religious people implement a religion to suit oneself (Donahue, 1985), Intrinsics were found to evaluate ethically questionable scenarios as less unethical than Extrinsics, but also to be less endorsing ethically questionable scenarios(Clark and Dawson, 1996; Walker, Smither, and DeBode, 2012).

Eventually, it becomes evident that the empirical results presented above are often contradictory and the picture that emerges is inconclusive. The above mentioned studies differ in regard to sample, definition of religion and methodology, why a more refined overview is provided in Appendix A. Whereas most of the cited studies indicate a positive relationship between religion and ethics, others suggest that religiousness is not a significant determinant of ethical behavior or attitudes (e.g. Hegarty and Sims, 1978; Kidwell, Stevens, and Bethke, 1987; Brammer, Williams, and Zinkin, 2007). Weaver and Agle (2002) contributed largely to the comprehension of these contradictory findings.

Firstly, they argue that unidimensional measures such as religious adherence or participation cannot be an adequate proxy for an individual’s religiosity. A more suitable approach is to use several dimensions to derive a comprehensive understanding of how religiosity shapes ethical decision-making (see, for example, Cornwall, Albrecht, Cunningham, and Pitcher, 1986). The varying definitions of religiosity further impede the comparison of empirical findings in order to draw comprehensive conclusions. Secondly, the choice of religious dimensions is sometimes made without awareness for a conceptual framework or instead guided by the empirical results (Parboteeah, Hoegl, and Cullen, 2008). For instance, Conroy and Emerson (2004) estimated in which way frequency of prayer and church attendance, subjective degree of religiosity and religious adherence affect the willingness to justify unethical behavior, and considered church attendance to be the most suitable measure because of higher significance levels. Furthermore, many conclusions about the relationship between ethics and religion are difficult to interpret due to non-representative samples - often, samples of students’ or managers’ ethical judgments were studied without accounting for the fact that ethical attitudes may vary with age and maturity (Walker, Smither, and DeBode, 2012). Lastly, measuring ethical behavior remains difficult. It often suffers from social desirability bias, especially when both religiosity and ethical behavior are evaluated concurrently (Weaver and Agle, 2002). Hegarty and Sims (1978) and Tan and Vogel (2009) for example avoided this bias by conducting an experimental game to assess ethical decisions of their study participants. Also the notion ”ethical behavior” is somewhat ambiguous. To name but a few examples of different forms of evaluating moral decision-making: Tan and Vogel (2009) focus on actual ethical actions, whereas Parboteeah, Hoegl, and Cullen (2008) ask for prospective behavior and again other studies ask for past immoral actions.

A more complex framework of moral decision making has been introduced by Rest and Barnett (1986). It outlines the process of moral decision making in four distinct stages. The first stage implies the realization that a situation calls for an ethical judgment and thus entails moral sensitivity. The second stage is characterized by the judgment of what a person ought to do in the given situation. Further stages include the motivation of a specific ethical course of action that demands priority over other individual values and finally the intention to follow one’s personal reasoning even under pressure. Naturally, an individual’s religion may have an impact on each of these four stages (Weaver and Agle, 2002; Longenecker, McKinney, and Moore, 2004). This study, as many others, relies on Rest and Barnett’s (1986) framework of moral decision-making and focuses on the third stage.

This paper attempts to eschew these caveats by differentiating different dimensions of religiosity following Cornwall, Albrecht, Cunningham, and Pitcher (1986), Barro and McCleary (2003) and Parboteeah, Hoegl, and Cullen (2008), as well as referring to a large Dutch population sample³. Both the willingness to advocate unethical behavior in specific situations and the results obtained in an anonymous two-player trust-game as well as indicators for trust serve to represent ethical behavior. Motivated by Parboteeah, Hoegl, and Cullen’s (2008) findings that suggest that religious practice, commitment and belief in church authorities positively influence ethical judgments separately⁴, it will also be assessed how much religious participation, i.e. ”belonging” (Barro and McCleary, 2006), influences the relationship between religiosity and ethical judgments. The initial assumption is that religious people may be per se less inclined to exculpate unethical decisions, less inclined to opt for betrayal and to distrust; even more so when attending church regularly as this promotes the building of a social network amongst like-minded people, reinforces religion-induced moral teachings as well as enhances an effective sanctioning system (see, for example, Tittle and Welch, 1983; Graham and Haidt, 2010). Here, religiosity refers to religious beliefs as well as to religious commitment. Moreover, this study explores whether religiousness implies a stricter compliance of moral laws towards one’s neighbor but not towards society as a whole which has to bear the cost for instance of bribery and tax evasion. Hence, the expectation is that religious and non-religious people share the same propensity to ”victimless” unethical behavior, but differ regarding stealing and adultery.

The empirical findings suggest that religious dimensions interact with religious practice and that the interaction exerts additional influence on the assessment of ethically dubious scenarios. However, the results do not provide evidence in favor of the hypothesis that religiously oriented people differentiate between misbehavior that harms a single person and misbehavior that harms society as a whole.

The structure of the paper is as follows. The subsequent section presents the main conjectures and their theoretical framework. The empirical analysis then describes the considered data, dependent and independent variables as well as obtained results. Eventually the paper concludes with a discussion of the main findings and an outlook on future research prospects.

2 Hypotheses

2.1 For religious people only? Religious participation promotes moral behavior. . . except for non-believers

As noted inter alia by Weaver and Agle (2002), the relationship between religion and ethical behavior is not trivial, as religiousness is a multidimensional individual characteristic. Following Cornwall, Albrecht, Cunningham, and Pitcher (1986) who differentiated between the cognitive, behavioral and affective dimension of religiousness, this study aims to shed light on the interaction within these dimensions with respect to ethical behavior.

The cognitive dimension covers religious beliefs, such as the belief in heaven and hell. Religious behavior such as praying and participation in church services and donations account for the behavioral dimension of religiousness. Lastly the affective dimensions covers spiritual - here referred to as religious - commitment and refers for example to the importance of God in one’s life. Parboteeah, Hoegl, and Cullen (2008) considered these dimensions separately, finding that beliefs in church authorities, religious practice and an individual’s commitment towards her religion, but not religious beliefs correlated with a lower level of willingness to advocate ethically dubious decisions.

However, one could argue that these dimensions interact, if and when for instance sermons affect religious people more efficiently than non-religious. The conjecture that religious practice in form of church attendance amplifies the effects the two other dimensions have on ethical decision-making seems plausible. Tittle and Welch (1983) were able to link religious participation to a lower level of deviant or unethical behavior particularly in an environment with ”general normative ambiguity” and a high concentration of nonaffiliated individuals. The explanation they provide builds on the notion that religious communities often exhibit features that foster compliant behavior, including sanctioning non-compliant behavior with some form of social expulsion. Secular societies tend to lack those compliance-inducing mechanisms and thus the relative impact of religiousness

should be more pronounced. Since the data to investigate this relationship is a large Dutch population sample, this relationship should be applicable: In the Netherlands, only about 50% of the population profess to be affiliated to a religious denomination. Furthermore these 50% generally also practice their religion, facilitating the differentiation between religious and non-religious people (Renneboog and Spaenjers, 2012).

Graham and Haidt (2010) argue that taking solely religious beliefs into account as catalyst for certain forms of behavior misses the actual point of social coordination⁵. Rather, participation in church services provides a sense of commitment to both the religious community and the church. Applied to this reasoning, beliefs in God, life after death, heaven and hell as well as spiritual commitment do trigger the attendance in the first place, but ritual practices help commit to the proclaimed set of values through being a part of an emergent entity within church. Therefore one would expect religious people that attend church more regularly to stick to religion- and ritual-induced rules, as for example the Ten Commandments, more strictly than those who participate in religious services less frequently.

A similar approach has been proposed by Weaver and Agle (2002). For them, focusing on observable measures of religiousness such as church attendance increases the risk that individual differences in motivations or cognition may be overlooked. They argue that while religion may influence some people, it may not have an impact on other people - the difference between these groups is not the frequency of church attendance, but the salience of the individual’s religious identity, that is, how strong religious role expectations have been internalized through repeated social interaction. Thus, the more salient the religious identity of a person is, the more probable it is that the role expectations linked to the religious identity translate into actual behavior and decision-making. This reasoning can be employed to explain a possible interaction between different religious measures. Thus follows the first hypothesis:

H1: Religious commitment and beliefs each have a negative effect on an individual’s willingness to justify ethically objectionable behavior as well as on betrayal decisions in the trust-game and measures of distrust. This relationship is comparatively stronger for religious people who attend church services more frequently.

2.2 The impact of religiousness languishes when the victim is not thy neighbor, but thy society

Brammer, Williams, and Zinkin’s (2007) empirical evidence suggests that religious people discern individual and corporate ethical behavior. Religiousness did not lead to a broader conception of CSR, and so one could ask whether this also holds for other, non-businessrelated forms of conduct where the victim is a large social group.

In the present questionnaire, subjects were asked to indicate whether dubious behavior could be justified on a scale from one ”never be justified” to ten ”always be justified”. The situations that had to be judged ranged from freeloading, claiming state benefit that one is not eligible for, over accepting a bribe and joyriding to adultery. When comparing these transgressions one can conclude that they refer to two different categories; one being misbehavior that directly affects ”thy neighbor”, and the other being misbehavior that affects society as a whole. With respect to the latter, Buchanan (1965) bluntly asked to whom one’s neighbor corresponds in a large group (p. 12 ). The implication is that people tend to refrain from following moral laws in a large group since a single person being directly affected by one’s behavior cannot be discerned.

He primarily argues that the choice whether an individual follows moral laws or a private maxim heavily depends on the surrounding environment, namely the behavior of others. More specifically, if one expects the peer group to decide in accordance with universal moral laws, it is best to comply - vice versa in a situation where nobody adopts moral laws, it is best to refrain from it as well in order to avoid exploitation. Provided that probabilities can be assigned to these two different states of nature, the decision between following the categorical imperative or following one’s own maxim now can be illustrated applying expected utility theory. However, this reasoning is only appropriate assuming that an individual’s choice does not alter the choice of behavior of others. In large groups this is not the case, the aggregate outcome of these decisions can be considered a given state of nature, and thus he suggests that people prefer to maximize personal utility. In small groups however the decision of each one depends to a large extent on the decision of other members. Under the assumption that c.p. an environment where all members follow ethical laws is preferred, the outcome that indeed all members follow these laws seems very likely. Empirical experiments generally support Buchanan’s (1965) conjecture (e.g. Kerr, 1989). In the given data this also seems to be applicable since unethical decisions that hurt society are on average more justified than those that hurt a specific person⁶

Within religious communities the effect of small numbered groups on ethical conduct may be even higher due to more severe mechanisms enforcing compliant behavior. Regular attendance of church services and participation in one’s religious neighborhood can contribute to an internalization of commitments to specific social norms, sustained by an effective sanctioning system (Tittle and Welch, 1983). However, this system of religioninduced social conformity may languish when the victim in the given situation is a large group, that is society as a whole. Even more so as moral laws proclaimed by religions often refer to victim-specific behavior, such as committing adultery and cheating. The Ten Commandments for example do state that one must not steal, but fare evasion may not be necessarily considered a theft. Therefore the second hypothesis to be tested states that:

H2: Religiosity has a positive influence on the difference between the willingness to advocate unethical conduct towards a specific person and towards society. This relationship is comparatively stronger for religious people who attend church services more frequently.

3 Data description and summary statistics

Following Trautmann, van de Kuilen, and Zeckhauser (2013) this study is based on data from the Longitudinal Internet Studies for Social Studies (LISS) panel, a large Dutch population sample. The panel of 2011 has been conducted by CentERdata with a total of about 9,000 respondents. In addition to information on their religious background and the evaluation of ethical behavior, the dataset includes inter alia information on age, the level of education and political orientation. However, due to the design of the questionnaire the relevant subsample considered in the study includes on average only between 39 and 109 subjects, so the sample may not be demographically representative.

The empirical analysis builds on both Probit and OLS regression models. The OLS regression serves to confirm obtained signs in the Probit regression. The interaction terms in particular have been studied using OLS since their interpretation in the Probit specification is not always clear (Hoetker, 2007). Moreover robust standard errors were used for all regressions to control for heteroscedasticity.

3.1 Independent variables

The set of independent variables comprises several dimensions esteemed crucial in evaluating one’s religiousness, namely adherence, beliefs, commitment and participation. To test for a general effect of religious adherence on ethical judgments the dummy variable ”adherent” has been included into the empirical analysis with 0 corresponding to non-adherent.

To assess religious beliefs, commitment and practice, several items on the LISS questionnaire were considered. First, the cognitive component included beliefs that are expected to apply to most religions (Barro and McCleary, 2003). Subjects indicated whether they believe in life after death, in heaven and hell, in a devil and in purgatory, in Adam and Eve as well as in the purpose of prayer where answers can be either ”no” - 0 or ”yes” - 1. These answers were then added to form a degree indicator whose maximum value is 7. Cronbach’s alpha for this measure is 0.89. In order to be able to differentiate only between people with strong and weak aggregate beliefs, a dummy variable based on this degree indicator was created⁷.

Second, it is assumed that an individual’s notion of God proxies her religious commitment. For Cornwall, Albrecht, Cunningham, and Pitcher (1986), religious commitment as a part of the affective dimension covers inter alia the role of God and faith in one’s life, one’s affection for God and willingness to obey His word. Unfortunately these data are not available. Nevertheless, the answer ”I believe without any doubt that God exists” to the question as to the notion of God hints a strong faith and thus a willingness to live accordingly. Religious commitment is scaled from 1 to 6, where the values 5 and 6 indicate a strong believer and values between 1 and 4 a person with a weak belief in God or none at all⁸.

The behavioral component consists of the frequency of participation in church services. Answers are scaled from ”never” - 0 to ”every day” - 6. Again, based on these answers a dummy variable takes the value 1 for regular churchgoer and 0 otherwise⁹.

Finally, the interaction term claimed in 2.1 is for the cognitive dimension derived by multiplying Strong aggregate beliefs × Church attendance and for the affective by multiplying Strong commitment × Church attendance, respectively.

3.2 Dependent variables

The dependent variables of this study are on the one hand the willingness to advocate unethical decisions, for instance committing adultery or accepting a bribe, and on the other hand the results of a trust game and measures of trust. In Rest and Barnett’s (1986) framework, the first variable corresponds to the third stage where, after the definition of morally acceptable action, moral values that accompanied this definition are given

[...]

¹ In several studies stronger religiosity has been related to significantly lower levels of drug consumption (e.g. Khavari and Harmon, 1982; Amey, Albrecht, and Miller, 1996; Engs and Mullen, 1999), but not to lower levels of cheating in exams or higher levels of participation in voluntary work (Smith, Wheeler, and Diener, 1975). Most empirical results differ as to how individual religiousness influences delinquency due to conceptual and methodological differences, but indicate a moderately negative relationship (see, for a detailed review, Baier and Wright, 2001). On the other hand, several studies have suggested religiosity to reliably predict an individual’s propensity to participate in voluntary work (see inter alia Parboteeah, Cullen, and Lim, 2004).

² For instance, subjects were asked to report their attitude towards behavior such as using company resources for private purposes.

³ The study is based on the Longitudinal Internet Studies for Social Studies (LISS) panel 2011, a large Dutch population sample conducted by CentERdata with a total of about 9,000 respondents.

⁴ Nota bene: Parboteeah, Hoegl, and Cullen (2008) differentiated between beliefs in religion versus in church authorities. Belief in religion corresponded to believing in afterlife, hell and heaven, whereas belief in church authorities covered for instance believing in church to provide answers to moral problems. For this study, only belief in religion has been considered, that Parboteeah, Hoegl, and Cullen (2008) did not find to have a significant influence on ethical judgments.

⁵ The analogy they draw are university sport events, where the event triggers the gathering, but the ritual behavior such as cheering for the home team emerges from the sense of community the university provides.

⁶ The mean of average acceptance of unethical behavior is 2.06 when the victim is the public compared to 1.90 when the victim is a single person. A paired t test results in t=3.0706, implying a significant difference.

⁷ The variable indicating strong aggregate beliefs takes the value 1 if the degree indicator is more than 66% of the maximum degree of religious beliefs and 0 otherwise.

⁸ Answers between 1 and 4 include ”I do not believe in God” and ”At some moments I do believe in God, at other moments I don’t”, whereas 5 and 6 refer to ”I believe in God, although I have my doubts” and ”I believe without any doubt that God exists”.

⁹ Regular churchgoer were identified by answers ranging from ”once a week” to ”every day”.