Role, Values, Person and Context: A Story of ‘Bent’repreneurship

Abstract

We prove a fundamental attribution error connecting rule-breaking behavior to entrepreneurs. We do so in the research context of the US, where we recently sampled from medium-sized venture entrepreneurs and their corporate executive peers (as an applicable reference point). We chose the US not only for its high entrepreneurial activity, but also because of the not uncommon relationship between business leaders and religion. By including various measures of religiosity in the study, we could control for factors that would likely influence rule-breaking, which standard models like the fraud triangle do not explicitly consider. In fact, we add contingency theory ideas to the fraud triangle to determine whether it is the decision conditions that determine rule-breaking rather than the role of the person (i.e., as an entrepreneur). We find that once demographic, religious, firm and industry contingencies are controlled for, any statistically significant influence of being an entrepreneur (relative to being a corporate executive with similar opportunity, motivation, capability and rationalization) disappears when it comes to self-admitted value-bending behaviors at work. Our contribution consists of a novel analysis, results and discussion of the ‘bent’repreneur—adding to conversations on the under-researched nexus of entrepreneurship with religiosity and ethical decision-making.

1. Introduction

Whether it is the headlines from the Internet bust of 2000, or from the 2008 global financial meltdown, or from more recent biotech and crypto surges, entrepreneurs have been associated with rule-bending behavior if not outright fraud (e.g., FTX and Binance, Theranos, Autonomy, Madoff, Napster and so on) (Hägg et al. 2024). Perhaps this association should not be surprising given that, by its very nature, entrepreneurship entails breaking existing traditions through innovations—which often includes challenging existing norms and regulations—in order to successfully enter an industry (Buckley and Casson 2001; Pidduck and Tucker 2022). This phenomenon of entrepreneurial rule-breaking is important to better understand for several reasons, including that: it is not going away; regulators have been slow to reign in the impacts of recent norm-bending technologies on privacy, competitiveness and national security; and the possible harms involved are not isolated to a few customers and their financial losses, but scaled up and outwards to also involve employees, communities, and often more vulnerable groups (e.g., children and teens and their mental health) (Litzky et al. 2006). Also, justifying the exploration of this phenomenon is the challenge of reducing ‘bad’ rule-bending and breaking by entrepreneurs while not stifling their innovative activity—which remains a national economic engine—all prior to fully knowing what may prove to be ‘bad’.

Consider the larger relevant phenomenon here to be rule-breaking by business leaders—by corporate executives and entrepreneurs. Such powerful individuals are being more exposed in the media for their bending and breaching of laws, traditions, norms and morals in the pursuit of what they consider success. In the past, much of the exposure was related to outright fraud (e.g., Theranos, Enron, Arthur Andersen and Worldcom), but more recently it also involves consumer manipulation (e.g., Meta/Instagram’s negative effects on teenagers), anti-competitive behaviors (e.g., Alphabet, Amazon and Apple’s gatekeeping actions), and violations of consumer privacy and employee trust. Such activity has had and continues to have many potential harms, not just relating to regulated criminal conduct but also from partaking in unregulated acts (e.g., involving disinformation about elections, pandemics and more) (e.g., Woiceshyn 2011). Combining the larger and more focused phenomena brings us to pose our research question—Are entrepreneurs more likely to bend the rules—rules that are based on their own values and morals—for their work? (Or, are other more significant factors correlated with such behavior?)

Addressing this question involves a focus on the possible fundamental attribution error involved—the attribution of rule-bending being more likely due to the expected innovative role of an entrepreneur (Arend 2016; Brenkert 2009; Lucas et al. 2022; Robinson 2014; Zhang and Arvey 2009). Additional attributions involve possible beliefs that entrepreneurs deal with less oversight, are in less-regulated industries, and have more on the line than their corporate peers. Addressing this question in the way we do—i.e., with an empirical study involving a peer group and controlling for the impact of religiosity on rule-bending—also fills a gap in the related literature. For example, Daradkeh (2023) asserts that research into entrepreneurial ethics, such as its basis (e.g., religion) and its violation (e.g., in rule-bending), is important but lacking. Drašček et al. (2021) note a distressing decline in empirical research on ethical decision-making. Dvouletý (2023) calls for expanding knowledge on the nexus of religion and entrepreneurship, while Battisti and Deakins (2018) and Pittz and Pittz (2024) call for deeper research into specific areas of entrepreneurial rule-breaking. We personally note a gap regarding work that spans rule-breaking, religion and entrepreneurship together, let alone in the USA. This paper seeks to tackle such concerns with a recent empirical study on entrepreneurs and their peers.

Our objective for this paper is to explore the story behind entrepreneurial rule-bending behavior—to determine whether it is fairly attributable to the role, or to the factors surrounding that role, including personal, organizational and contextual characteristics. We seek to add to the conversations on entrepreneurial behavior, ethical decision-making by business leaders, and individual religiosity’s impact on both. We do so within an appropriate context—the US—where there exists not only important entrepreneurial activity, and strong connections of entrepreneurs with their religion beliefs (e.g., Dougherty et al. 2013), but also many recent cases of ‘bad’ behavior (much of which remains under-regulated). We do so with an appropriate ‘comparative’ approach—where our survey data include a benchmark peer group of business leaders (Longenecker et al. 2006); the responses of medium-sized venture entrepreneurs are compared with the responses of corporate executives of larger firms,1 who also would have the opportunity, means and motive to rule-break (e.g., think Boeing, Enron, and Purdue Pharma). Additionally, we apply an appropriate measure of rule-bending—one that elicits freer self-reporting (i.e., it does not speak to criminality but to the respondent’s own morals and values, it is highly-anonymized, and it allocates blame to the demand of work).2 As such, we contribute new empirical results to the nexus of various streams of literature about the entrepreneur—as a rule-breaker, as an executive, and as a religious person—as well as to research on ethical decision-making, by extending the fraud triangle/means–motive–opportunity model towards a more contingency-based perspective.

In the remainder of the paper, we review the relevant literature, explain the expectations of the study, describe the empirical analysis, report the results and discuss the implications, limitations and future work prior to concluding.

2. Literature Review

There are four main streams of research that are relevant to our work here: the research attributing rule-breaking to entrepreneurship; the research about rule-breaking in business more generally; the research on ethical decision-making and its influences; and the research connecting such decision-making to religiosity, especially for entrepreneurs. We consider each of these streams in turn.

Research attributes rule-breaking with entrepreneurs and entrepreneurship in many ways and for many reasons (Shaver 2009; Wells and Graafland 2012). For example, entrepreneurs have been characterized as experts at exploiting lagging regulations (e.g., regarding digital privacy) as well as moral ambiguity (Fadahunsi and Rosa 2002; Hart 1975; Stewart 1990). Entrepreneurs have been accused of being more challenged by business ethics (Porter and Kramer 2006; Trevino and Brown 2005; Williams 2002). Additionally, entrepreneurs need to be innovators, and that necessarily implies doing things that break with the accepted ways of the past (Brenkert 2009; Longenecker et al. 1988). Further, translating opportunities into reality appears to require immoral practices by some entrepreneurs (Fadahunsi and Rosa 2002).

More generally, rule-breaking in business is common, and done for many reasons, mostly selfish ones, as models like the fraud triangle portray (Cressey 1953; Raval 2018). Given that spotting the opportunity to do so is a major part of such models, and entrepreneurs are defined as having skills in identifying, creating and exploiting opportunities, it is no surprise that they have been associated with such activity (e.g., Arend 2016; Benson and Simpson 2014). Broadly speaking, the tension between one’s material interests and one’s adherence to moral values remains a significant problem for all humans and not just entrepreneurs or corporate executives (Cheung and King 2004), even though the relevant moral issues that arise for entrepreneurs are often critical to their identity (Buchholz and Rosenthal 2005). So, it is no surprise that entrepreneurs, like their corporate executive peers who wield significant decision-making power, bend the rules—and even their own moral values—for several reasons, chief among them being for business success (Fisscher et al. 2005).3 Rule-bending increases chances at success because it often saves costs (e.g., on safety and time) and can take rivals by surprise (Cheung and King 2004). That is not to say that behaving ethically can also be potentially beneficial or, at least not disadvantageous—see Carasco and Singh (2003); Green (1989); Ogbari et al. (2016).

The research on ethical decision-making is broad, encompassing issues from what exactly is ethical in business, to ethical decision-making’s effects and drivers (e.g., Finkelman and Kelly 2012; Greve et al. 2010; Saksena 2001; Wikström 2006). We focus on unethical decision-making and its drivers in this study, comparing entrepreneurs with their peers. The decision to rule-bend is often influenced by many factors rather than being solely attributable to a person’s role (e.g., entrepreneur) or their ethical compass (e.g., their religiosity). For example, it may be influenced by sympathy for those who might be harmed (Buchholz and Rosenthal 2005; Dewey 1971) or how it may be interpreted by employees (e.g., as setting a new level of acceptable behavior—see Treviño et al. 2000). A decision-maker’s morality may also be influenced by cultural background (Brenkert 2009). Further, a decision-maker’s motivation may also be influenced by the uncertainties and pressures characterizing the competitive context (Arend 2023; Hall and Rosson 2006; Staw and Szwajkowskib 1975; Sethi and Sama 1998).

The research connecting religion and spirituality to ethics, and then to business decision-making and entrepreneurship, is more recent, especially for the latter. For example, the literature connecting spirituality to entrepreneurial activity has experienced a recent resurgence (Block et al. 2020; Dyck 2014; Smith et al. 2019; Smith et al. 2021). It has even been suggested that studies of entrepreneurship are incomplete if they do not include the influence of spirituality and religion (Chan-Serafin et al. 2013; Smith et al. 2021). In the context of the US, this is even more so, where Judeo-Christian values are widely taught and are the basis for laws, many services and community events, and even some firms (e.g., Chick-fil-A, Hobby Lobby, and eHarmony) ( Dougherty et al. 2013; Judge and Douglas 2013). Religiosity is often considered a mitigating factor in rule-breaking. For example, one solution to rule-bending is via an emotional commitment to a moral code—which is often possible via religious beliefs and the threat of attributable guilt as punishment (Casson 1991, 1998; Rietveld and Hoogendoorn 2022). Indeed, sensitivity to immorality is based partly on one’s spiritual and religious beliefs (Buchholz and Rosenthal 2005; Chan and Ananthram 2019). Additionally, an ethical commitment—through identification with religion—has further upsides, such as networking and the trust that is built upon the entrepreneur’s stewardship and pride in reputation, which can be leveraged into a climate of confidence within which risky projects can attract investment and take place (Casson 1998; Deller et al. 2018). That being noted, even for deeply religious entrepreneurs, admissions of acting in ways that violated their own moral principles have not been rare (Cheung and King 2004).

One possible reason is that the use of those beliefs as an excuse to do bad (e.g., for ‘the greater go[o]d’) can also occur (e.g., Jackson and Gray 2019; Peifer 2015).

In the next section, we draw upon this literature to pose several tested empirical expectations from our study. We note, however, that while many aspects of entrepreneurial rule-breaking have been investigated in past research, there remains little—empirical or theoretical—on the nexus of rule-bending, entrepreneurship and religiosity. Indeed, significant questions remain about what are the most important factors in the story of ‘bent’repreneurship—in explaining why entrepreneurs bend the values and morals for their work. We believe that the current short-hand of entrepreneurial ‘role’ is hiding a longer-hand truer explanation of rule-bending that involves other practically measurable factors, such as individual, firm and industry attributes.

3. Theory and Expectations

In order to explore the research question and the fundamental attribution error that underlies it, we start with the fraud triangle’s theoretical framing and then consider a modification that includes more contingencies. The fraud triangle (Cressey 1953) explains unethical rule-breaking using three conditions. We argue that entrepreneurs are relatively more likely to meet those conditions. This explains the attribution. Then, we drill down on what factors could affect those conditions and propose that those may not be singularly entrepreneurial. We complete the exercise in theorizing by considering contingencies (drawing on contingency theory—see Morgan 1986; Scott 1981)—that rule-bending at work has more to do with the context at play than the individual.

There are three conditions that can trigger that behavior according to the fraud triangle (see Cressey 1953; Raval 2018)—opportunity, motivation and rationalization.4 This model includes variants like the means–motive–opportunity version that switches capability for rationalization (Pendse 2012). Arguably, entrepreneurs find themselves facing those conditions even more than their executive peers. Opportunity is afforded to entrepreneurs as owner–managers who face imperfect oversight, often even less than executives, often because of greater information asymmetries with potential overseers, and the greater ability to override controls (Bussmann and Werle 2006; Dorminey et al. 2010; Howe and Malgwi 2006; Loebbecke et al. 1989; Lou and Wang 2009; Marden and Edwards 2005). Motivation is very high, both in terms of rewards and pressure on the entrepreneur, who accepts high-powered incentives to succeed, often even more high-powered than executives. Entrepreneurs also often have more at stake—in terms of identity and sole accountability—than their executive peers. Rationalization is often easy to manufacture for entrepreneurs who see others ‘faking it’ or ‘embellishing their numbers’ or ‘cutting corners’ to succeed. Executives may have even an easier or proximate rationalization when the corporate culture is toxic (e.g., Enron). Capability (or ‘means’) is embodied by the entrepreneur who has successfully started a new venture, engineered its business model and retained significant decision-making power. As such, it is expected that rule-bending would be more attributable to entrepreneurs than managers, and even executives, let alone other employees (Arend 2016; Brenkert 2009; Lucas et al. 2022; Robinson 2014; Zhang and Arvey 2009). Of course, it is not just entrepreneurial ventures that are hospitable to these factors; most businesses are intrinsically potentially criminogenic—being full of opportunities for deviance (Greve et al. 2010; Punch 1996).

Using somewhat blunt instruments like the fraud triangle only gets us so far in addressing our research question. The reason such a model exists is, of course, not to address our question, but to mitigate fraud by both internal and external means. For example, internally, oversight can reduce opportunity, appropriate incentives can reduce motivation, and positive culture can reduce rationalization. However, such ‘solutions’ are much less applicable to top executives, CEOs and entrepreneurs, given they all hold more knowledge than their overseers (as well as regarding ways to escape accountability by blaming others), all have more than financial incentives at stake (e.g., reputation), and can all influence culture (e.g., through imprinting). So, the applicability of the model is limited when it comes to our focus on the high-level decision-maker. It is also limited in applicability because it may require some sensitive measures about organizational security protocols, individual negative motivations, and personal and often questionable rationalizations.

What we need to do, then, is open this up to other factors—specifically, those that apply to high-level decision-makers and that are easier to measure—and to other rules being broken besides the criminal ones—in order to obtain more admissions to such behaviors. We want to move the model from being subject to a fundamental attribution error that encapsulates person, motivation and the requirements (e.g., innovativeness) of a role like the ‘entrepreneur’ and deconstruct it to separate out the heterogeneity in people, morals and contexts, all to improve outcomes (i.e., of ethical decision-making). Because this is an exploratory study—one that tests whether the fundamental attribution error can indeed be shown to be an error—we do this in somewhat rough terms. We do it by arguing that, like the best approach to a decision is contingent on the circumstances, ethical decision-making—by any party (including entrepreneurs)—will also be similarly contingent; therefore, we apply contingency theory (e.g., Morgan 1986; Scott 1981). Additionally, we do so by adding several measurable contextual factors to the list of possible explanations for rule-bending behavior. For example, it is likely that individual-level characteristics, like sympathy and the need for social acceptance, are also contributory to rule-bending (Schuchter and Levi 2016).

Based on the reviewed literature and our stated theoretical approach, we now describe a set of empirical expectations as the rough results we expect from the analysis of our data. We do this rather than arguing formal hypotheses, again, because this is an exploratory study where a gap in the literature exists. As a base result, prior to adding contingencies, we do expect to verify the fundamental attribution, as has been argued based on the fraud triangle conditions above. So, our first expectation is that when comparing entrepreneurs to their peer corporate executives, we will observe more rule-bending for the former (i.e., given more opportunity and motivation).

We next add an expectedly strong contingent effect—one that is outside of the traditional triangle model and one that works to counter rule-bending behavior in this sampling context. Religiosity—as various spiritual beliefs and relationships with God—is not only prevalent among US entrepreneurs, but also a strong force advocating for ethical decision-making and reasonably easy to measure. In the fraud model, it would likely constitute an effective means against rationalizing bad behavior. As a result, we expect that adding effects of religiosity will significantly reduce any correlation between entrepreneurs and rule-bending. So, our second expectation is that the effects of religiosity on rule-bending (overall) will be significant (and negative) and that the correlation between rule-bending and being an entrepreneur (versus a peer corporate executive) will be significantly reduced.

Finally, we add a set of further contingencies to all three standard levels of analysis: personal–individual, firm and industry. From the literature (see above), we have seen support for the expectation that personality traits (e.g., sympathy) affect rule-breaking, and contextual characteristics (e.g., the firm’s success; the industry’s turbulence) can affect ethical decision-making (e.g., through motivation and opportunity in the fraud triangle). Introducing these further contingencies is likely to further reduce the entrepreneur–rule-breaking relationship. So, our third expectation is that some basic personality, firm and industry characteristics will have significant effects on rule-bending behaviors (acting in line with how the fraud triangle may categorize them) and that that will remove any significant correlation between the entrepreneur and such behaviors. (This will, in effect, prove that the fundamental attribution was in error.)

4. Materials and Methods

We recruited 168 participants—84 US entrepreneurs and 84 US corporate executives, through the online panel provider CloudResearch, to take part in our IRB-approved research study. These participants were triple-verified (i.e., using reliable recruitment sources, filter questions and consistency checks). We targeted such high-level decision-makers because they play a central role in strategic decision-making (Hunjra et al. 2021). In exchange for an average payment of 30 USD, they completed a 35-minute-long online (Qualtrics) survey. This is a reliable method to access a suitably diverse sample (e.g., Landers and Behrend 2015), and the data quality does not substantially differ from a non-paid random sample (e.g., Behrend et al. 2011) when researchers embed attention checks in the survey as we did (Griep et al. 2023).

As with similar studies, we collected data through a single survey instrument and a single respondent per questionnaire (Mazereeuw-van der Duijn Schouten et al. 2014). Additionally, as with similar studies, we applied a subjective ontological perspective of reality and agent behavior. We assumed an objective observer role in the research, one that was curious about a phenomenon and open to alternative results regardless of the stated expectations. To address concerns about common method bias and common source bias, we applied several procedural remedies (Podsakoff et al. 2003): respondents’ privacy was assured by using a third-party source and the Qualtrics’ option; item ambiguity was mitigated by simplifying questions and concepts; and scale items were separated (i.e., so that respondents were less likely to guess the relationships among the dependent and the independent variables and consciously match their responses to the different variables (Parkhe 1993)). As is standard, we applied Harman’s one-factor test (Podsakoff et al. 2003). The unrotated principal component analysis over all variables revealed five factors with eigenvalues greater than 1.0, which combined accounted for 63% of the total variance, where the largest factor only explained 29.8% of it. Our results are inconsistent with common method effects (given that we did not find statistically significant links among all of the relations posited (Bacharach et al. 2005)).

We also considered social desirability response bias (Trevino and Weaver 2003). To reduce the potential for this bias, our informed consent statement explained that the survey was confidential, anonymized and used for research purposes only. Thus, participants had no reason to present a more favorable picture of themselves than otherwise. Social desirability response bias did not appear to vex the analysis given the high variance in the scores of the various components of spirituality, sympathy and other constructs. Thus, as others have in this type of study, we relied on the empirically supported belief that self-reported behavior is strongly correlated to actual behavior (Gatersleben et al. 2002) to interpret our results.

Like similar studies (e.g., Hunjra et al. 2021), our constructs for spirituality, the use of religious beliefs, industry turbulence and other multidimensional variables were based on several questions each. These constructs were verified with components having load factors over 0.50 (Cua et al. 2001) and Cronbach alphas over 0.7 in value. Like related research, we applied controls on an individual (e.g., age and trait), firm (e.g., size) and industry (e.g., growth and turbulence) level (Brieger et al. 2021; Shin et al. 2022).

We considered two dependent variables to capture how much participants bent their values in their work: D_CPREL, which measures to what extent the participant has compromised their religious or spiritual values because of their work; and D_CPMRL, which measures to what extent the participant has had to compromise their morals because of work. These measures capture a definition of rule-bending on a more personal level—where one feels that s/he has violated their morals and values in pursuing their venture. This measure embodies unethical decision-making beyond the criminal level—where the fraud triangle was originally applied. This is important because substantial harm to stakeholders often occurs from non-criminal but unethical or unregulated behaviors. This measure is advantageous because it avoids the sticky issue of whether the rules involved are considered ‘right’ or ‘wrong’ to the reader, given that the subjects are applying their own standards and not society’s. Further, it is advantageous because the admission of bending is made without personal blame (i.e., it happens ‘because of work’) or intent, making it more likely for it to be (honestly) admitted to.

We used three primary and three secondary explanatory variables: ENT—which indicates whether the respondent is an entrepreneur (i.e., currently an owner, partner, founder, president or corporate executive at a new venture that employs at least 10 FTEs, or started or helped start such a venture within the last 10 years); SP_X—which is a 10-item construct (α = 0.966) measuring the respondent’s spirituality and religiosity (e.g., I have a personally meaningful relationship with God); ENTxSPX—the interaction term (calculated using standardized values for the two previous items); D_IGREL—which captures how much the participant’s spiritual belief/religion allows them to ignore some concerns of others; SP_RUU—which is a 3-item construct (α = 0.883) measuring how much the participant relies on their spiritual beliefs to help them deal with risk and uncertainty; and SP_RIGHT3—which is a 3-item construct (α = 0.862) capturing the role of the participant’s spiritual beliefs in making ‘righteous’ business decisions.

We applied nine control variables to the regression analyses: AGE—which measures the participant’s age; MINORITY—which captures the participant’s minority status; SYMPATH—which is a personality trait measure of sympathy; ADAPTX—which is a 7-item construct (α = 0.729) measuring the participant’s ability to adapt to new contexts; PER_WX—which is a 15-item construct (α = 0.898) measuring the participant’s personal satisfaction with work; FIRMREV—which provides the current size of the firm in annual sales revenues; IND_GR1—which indicates industry growth; INDTUNCX—which measures the industry’s turbulence, instability and uncertainty; and, INDRELX—which measures the degree of religious affiliation in this industry. See

Table 1. Variable Descriptions.

Variable	Description
D_CPREL	(5-pt Likert scale) I have compromised my religious or spiritual values because of my work.
D_CPMRL	(5-pt Likert scale) I have had to compromise my morals because of work.
ENT	Are you currently an owner, partner, founder, president or corporate executive at a new venture that employs at least 10 FTEs; or, have you started or helped start such a venture within the last 10 years? (yes = 1; no = 0) [based on similar criteria in (Sardana and Scott-Kemmis 2010), and (Busenitz and Barney 1997)]
SP_X	10-item construct (C-alpha = 0.966): I have a personally meaningful relationship with God + prayer is important to me + I believe that God is concerned about my problems + My relationship with God contributes to my sense of well-being + I believe that God loves me and cares about me + I often think about religious issues + I believe that God or something divine exists + I often take part in religious services + I often pray + I often experience situations in which I have the feeling that God or something divine intervenes in my life. [based on the Religious Well Being subscale of (Paloutzian and Ellison 1982)]
ENTxSPX	interaction term (calculated using standardized values for the two terms)
D_IGREL	(5-pt Likert scale) My spiritual belief/ religion allows me to ignore some concerns of others (some of the stakeholders affected by my actions) and focus on doing what is right by those beliefs.
SP_RUU	3-item construct (C-alpha = 0.883): My spiritual beliefs allow me to take greater risks + My spiritual beliefs help me deal with business uncertainties + My spiritual beliefs comfort me when facing unknowns.
SP_RIGHT3	3-item construct (C-alpha = 0.862): When I succeed in business, I see that as a sign of the righteousness of my spiritual beliefs + I do the right thing in business to ‘avoid going to Hell’ + I do the right thing in business to ‘get into Heaven’. [based partly on work by (Wong 2008)]
AGE	participant age (22–29 = 1; 30–39 = 2; 40–49 = 3… 80–plus = 7)
MINORTY	participant identifies as a minority (=1) or non-minority (=0)
SYMPATH	(7-pt Likert scale) sympathetic, warm/NOT critical, quarrelsome [based on standard, 2-question Big-5 scale]
ADPTX	7-item construct (C-alpha = 0.729): Many times I have proved that I can cope with difficult situations + I have the ability to adapt positively to challenging circumstances + I am resilient + I quickly master new routines, procedures and new ways of working + I am constantly looking for new ways to improve my performance at work + I like to learn new things + I am not afraid to learn new things. [based partly on innovative behavior work by (Bolton and Lane 2012), and (Karwowski et al. 2012)]
PER_WX	15-item construct (C-alpha = 0.898): I am satisfied with the happiness I feel in my work + My business work enhances my life interpersonally + My business work enhances my life spiritually + My business work enhances my life intellectually + My business work enhances my life physically + My business work enhances my life emotionally + My business work enhances my life overall + I am satisfied with the combination of work and life + I am satisfied with my current individual income + My work helps me live out my life’s purpose + I see my career as a path to purpose in life + My career is an important part of my life’s meaning + I try to live out my life’s purpose when I am at work + I enjoy my work + I am happy to meet the challenges of work. [based partly on research on purposeful work by (Xie et al. 2016), and on work satisfaction by (Chang and Chen 2020)]
FIRMREV	the current size of the firm in annual sales revenues (below $1m USD = 1; $1m up to $10m = 2; $10m up to $100m = 3; over $100m = 4)
IND_GR1	(5-pt Likert scale) there is a lot of growth in my firm’s industry
INDTUNCX	5-item construct (C-alpha = 0.806): My firm’s industry is very turbulent + There is a lot of exit and entry in my firm’s industry + There is great uncertainty in this industry + Predicting demand in this industry is very difficult + The supply chain is unstable in this industry. [based partly on work by St-(Pierre et al. 2023)]
INDRELX	2-item construct (C-alpha = 0.834): There are a lot of religious-affiliated firms in this industry + Most of my firm’s customers are religious. [based partly on work by (Swimberghe et al. 2011)]

for full descriptions of the variables used.5

Table 2. Descriptive Statistics and Correlations.

	Variable	Mean	StdDev	Min	Max	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
1	D_CPREL	−0.619	1.339	−2.000	2.000
2	D_CPMRL	−0.268	1.462	−2.000	2.000	0.536
3	ENT	0.500	0.501	0.000	1.000	0.312	0.257
4	SP_X	0.677	1.122	−2.000	2.000	0.451	0.222	0.212
5	ENTxSPX	0.210	0.974	−2.378	2.378	0.103	0.094	0.000	−0.088
6	D_IGREL	0.262	1.258	−2.000	2.000	0.374	0.318	0.247	0.469	0.024
7	SP_RUU	0.593	1.141	−2.000	2.000	0.341	0.192	0.197	0.811	−0.062	0.410
8	SP_RIGHT3	0.163	1.209	−2.000	2.000	0.458	0.418	0.346	0.683	0.047	0.439	0.751
9	AGE	2.667	1.030	1.000	6.000	−0.211	−0.306	−0.081	−0.076	0.012	−0.131	−0.087	−0.204
10	MINORTY	0.321	0.468	0.000	1.000	0.205	0.057	0.229	0.172	−0.008	0.171	0.160	0.242	−0.149
11	SYMPATH	1.268	1.185	−2.000	3.000	−0.059	−0.161	0.066	0.147	−0.121	−0.069	0.169	0.044	−0.015	−0.054
12	ADPTX	1.434	0.437	0.143	2.000	−0.037	−0.045	0.164	0.256	0.054	0.114	0.321	0.245	−0.013	−0.004	0.263
13	PER_WX	0.761	0.573	−1.250	1.625	0.368	0.342	0.408	0.417	0.011	0.225	0.443	0.493	−0.259	0.079	0.151	0.528
14	FIRMREV	2.732	0.865	1.000	4.000	−0.180	−0.033	−0.297	−0.061	0.074	−0.089	−0.127	−0.139	0.121	−0.126	−0.017	−0.082	−0.263
15	IND_GR1	0.982	0.872	−2.000	2.000	0.288	0.170	0.212	0.348	−0.006	0.223	0.260	0.309	−0.187	0.087	0.257	0.315	0.456	−0.054
16	INDTUNCX	0.068	0.901	−2.000	1.800	0.492	0.359	0.333	0.073	0.191	0.214	0.100	0.163	−0.031	0.135	−0.166	−0.047	0.126	−0.115	−0.011
17	INDRELX	−0.268	1.189	−2.000	2.000	0.595	0.496	0.311	0.446	0.135	0.437	0.379	0.537	−0.266	0.199	0.084	0.211	0.551	−0.175	0.356	0.368

Notes: N = 168. Correlations over 0.141 are significant (p < 0.05); most are significant. A few correlation are over 0.5; survey questions differ regardless (on their face, they measure different things); regardless, multicolinearity not problematic in regressions.

depicts the descriptive statistics and correlations of the variables. We note near-full ranges and healthy variances for most factors (ADAPTX being the exception [with apparently most participants at this level of management believing that they are ‘above average’ in their adaptation skills]). The correlations are high; in fact, most of the correlations are significant between factors. However, this did not translate into multicollinearity in the empirical analyses, given that the VIFs were all below 4.4 in the regressions.

We applied a three-step (hierarchical) OLS regression analysis for each of the dependent variables. In the first step, we included the role (ENT), basic religiosity and their interaction as explanatory variables (see the left column of the set of three for the dependent variable indicated). In the second step, we added three variables that measured different ways that religiosity could be used that are likely to influence rule-bending (see the middle column of the set of three). Finally, in the third step, we added several controls at the personal, firm and industry levels (see the right column of the set of three).

5. Results

Table 3. Regression-Based Empirical Analysis.

Variable	D_CPREL						D_CPMRL
ENT	0.224	***	0.154	*	−0.010		0.218	**	0.088		0.032
SP_X	0.416	***	0.388	***	0.340	***	0.185	*	0.000		−0.040
ENTxSPX	0.140	*	0.106		0.038		0.110		0.047		−0.016
D_IGREL			0.139	†	0.020				0.186	*	0.091
SP_RUU			−0.252	*	−0.145				−0.291	*	−0.161
SP_RIGHT3			0.264	*	0.089				0.522	***	0.329	**
AGE					−0.046						−0.159	*
MINORTY					0.018						−0.115	†
SYMPATH					−0.033						−0.116	†
ADPTX					−0.232	***					−0.185	*
PER_WX					0.122						0.200	*
FIRMREV					−0.064						0.108	†
IND_GR1					0.109	†					0.004
INDTUNCX					0.331	***					0.187	*
INDRELX					0.237	**					0.219	*
adj R2	0.259		0.295		0.526		0.091		0.220		0.389
F-stat	20.448	***	12.674	***	13.338	***	6.584	***	8.870	***	8.078	***

Notes: N = 168. Significances indicated as: † = p < 0.10; * = p <0.05; ** = p < 0.01; *** = p < 0.001. All VIFs < 4.4. Standardized coeficients shown.

depicts the HOLS analysis. Prior to summarizing those results, it is important to note that a simple, first-cut analysis of the raw data—using a two-tailed means t-test (assuming unequal variances)—revealed that entrepreneurs (relative to their corporate executive peers) are more likely to compromise their religious or spiritual values, and their morals, because of work (at the p < 0.001 significance level, with the mean as ‘yes’ for entrepreneurs bending morals and ‘no’ for their peers). This result is consistent with our first expectation—it supports the fundamental attribution. The first regressions are also consistent with that expectation (and attribution)—ENT is highly significant for both D_CPREL and D_CPMRL.

Adding the proxies of ‘religiosity use’ to the regressions provides mixed results and partial support for our second expectation. ENT remains significant for D_CPREL (but less so than without the additional variables). So, it appears that adding depth, or dimensionality, to the rationalization part of the rule-bending framework—through religiosity—significantly reduces the correlation between being an entrepreneur and rule-bending behavior. What is surprising is that half of the religiosity measures expected to correlate with decreasing rule-bending actually work the other way. So, while D_IGREL and SP_RUU act in the correct manner (i.e., indicating more rule-breaking when religion is used to rationalize it, and less rule-breaking when religiosity reduces the uncertainty of the situation—acting to make rule-breaking less ambiguous), SP_X and SP_RIGHT3 do not; they correlate with more rule-breaking. We offer a preliminary explanation for this surprising result—that holding stronger religious beliefs (especially about what is ‘right’) makes one more sensitive to one’s own rule-bending, lowering the bar for admitting such behavior, and thus appearing to increase its occurrence.

The third regressions not only provide the better models of the phenomena, but also eliminate the significance of the role (ENT) altogether in the rule-bending story. This supports our third expectation—i.e., that when motivation factors are added (and more rationalization factors [above those related to religiosity] are added) the influence of the attribution vanishes. To the other part of the expectation, it appears that the added contingencies work in mostly logical directions (e.g., adaptability lowers the need to break rules, sympathy lowers the rationalization to do so and industry uncertainty increases the opportunity and rationalization to do so, as do stakes in the firm). The one contextual factor that does not jibe well is INDRELX, which may have a similar explanation as the other ‘surprises’—i.e., it increases the sensitivity of the decision-maker to what is considered bending a rule, and so more of such behavior is admitted to. This shows that care needs to be taken about specifying ‘what is moral’, given that morality can differ significantly between individuals and social contexts, leading to many ‘grades of acceptability’ for different actions—see Fadahunsi and Rosa 2002).

6. Discussion

Our contribution involves deepening the understanding of ‘bent’repreneurs and what can drive their behaviors, and what this should mean to our teaching and research in our field. Through this empirical study of recent survey data, we have explored the story behind ‘bent’repreneurs and found that there is more to it than a simple relative bias of entrepreneurs towards rule-bending. The story is less about the opportunity that being an innovative, leader–founder entrepreneur affords; instead, it is more about the factors that influence a decision-maker’s motivation and rationality, from contextual pressures to religiosity, respectively. As more relevant controls are added, the fundamental attribution of rule-bending to entrepreneurs is replaced by the finding that such behaviors are contingent, having more to do with the context and the decision-maker’s morality. The story has shifted from the role of entrepreneur to a decision-maker’s values, personal traits, and firm–industry context. Indeed, the results indicate that, once basic demographic, psychographic, firm and industry factors are controlled for, in addition to relevant religiosity measures, the fact that a decision-maker is an entrepreneur (versus a corporate executive) is not statistically significant in explaining self-reported moral compromises at work. Such a result is consistent with other recent research indicating that contingencies matter for entrepreneurial misbehavior (e.g., Theoharakis et al. 2021).

Our results offer several implications—for academia and for practice. We speak to the main ones below. Our significant empirical results contribute to the entrepreneurship literature, especially those streams that connect to spirituality and religion and to ethical decision-making, in terms of providing a deeper and more dimensioned exploration into an important and fundamental attribution error, using a relevant reference point for perspective. We contribute to a relatively under-researched area in entrepreneurship (Chell 1985)—where most previous work has considered either illegal business activity (Fadahunsi and Rosa 2002) or business ethics more generally (Ajagbe and Ismail 2014; De George 1990; Fadeyi et al. 2015; Ogbari et al. 2016). Our work answers the call for more research on moral decision-making and bending (Buchholz and Rosenthal 2005). Our work adds to other studies that seek to update the fraud triangle (Raval 2018; Van Scotter and Roglio 2020); we do so by adding important contingencies like religiosity, personality traits and industry context.

The main practical concern about rule-bending is what such behavior produces—in terms of its potential harms, risks and benefits. Once the rule-bending outcome is better understood, then we can more effectively determine whether, when and how to mitigate or even promote it. Our results identify several factors that are expected to produce harm (i.e., because they are associated with compromising values and morals), some of which can help internal or external overseers to better target mitigating policy (e.g., towards more turbulent industries, and younger, less adaptable decision-makers). By offering such potential targets, we add to the literature that also considers targeting but also suggests mitigating tools. For example, De Castro et al. (2024) target entrepreneurial firms with improved governance, while Antunez et al. (2023) target internationalization issues with leadership. Lucas et al. (2022) consider entrepreneurs and suggest improved regulations to reduce bad behavior, while Nordstrom et al. (2020) suggest moral communities and firm culture. Henle (2006) offers business education as an answer, while Premeaux (2009) points out that visible enforcement (e.g., criminal convictions of rule-breakers) can be an effective approach.

With a deeper understanding of ‘bent’repreneurship, we can focus on how to organize to leverage its benefits while mitigating its harms and risks. This entails offering empirical results like ours that will help others to design businesses and regulations in which people are encouraged to innovate and push boundaries while discouraging their acting in illegal or highly immoral ways; of course, this design continues to present a real practical challenge (Brenkert 2009). The trade-off between entrepreneurial rebelliousness and constrained ethical leadership is not clear: Yes, ethical behavior is good for business, as it avoids legal problems and contributes to preferable employee self-selection, commitment, satisfaction and comfort (Treviño et al. 2000), but the conformity and fear of change it often produces can be fatal to firms in dynamic contexts where (constant) innovation (of more than just technology) is necessary.6

Our analysis of recent survey data of US entrepreneurs has contributed to the related literature on entrepreneurs, their ethical decision-making, and the effects of their religiosity. This paper has also contributed to recent related conversations in this journal, including on spirituality and workplace behavior (Iqbal et al. 2023), opportunity assessment affected by moral valuation (Arend 2021), ethical reasoning at work (Lasthuizen and Badar 2023) and leadership and sustainability (Warner-Søderholm et al. 2024).

6.1. Future Research

More research is needed to learn more about entrepreneurship in relation to morality and ethics (in defined contexts), as right now it is mostly based on myths, idiosyncratic popular stories and simple but erroneous attributions. Such research may help us to develop a model of the ethics of entrepreneurship (Brenkert 2009), which may be based on extensions of the fraud triangle (e.g., involving contingency theory) as we have suggested here. Below, we describe several paths forward for consideration in continuing work on this rule-bending phenomenon.

An interesting way forward is to drill down on how ethical decision-making is processed by the individual within their business context. This may involve sensemaking—rational, emotional, ethical or spiritual (e.g., Diochon and Nizet 2019). On the ethical side, this may entail how commercial and moral costs and benefits and risks are traded-off (e.g., Arend 2021).

This leads to another important area for further research—in accounting for the full effects of rule-bending. At its worst, it can not only harm the decision-maker, their venture and its stakeholders, but also the moral status of an industry as a whole (Brenkert 2009; Machan 1999). The dimensions of the potential harm should be captured in the magnitude of the consequences, their temporal immediacy, their proximity and the concentration of the effects (see Jones 1991), as well as the kinds of effects (e.g., from personal to social, spiritual to financial and short-term to long-term). The expected harms include the added market inefficiencies that arise from unfair advantages gained by rule-bending/breaking (e.g., involving insider trading, bribery, collusion and tax evasion—see Fadahunsi and Rosa 2002). Other harms involve the loss of transactional trust (when the bending is discovered) both inside and outside the firm, which tarnishes the brand and increases contractual costs in the longer term (Cheung and King 2004). It would be useful to determine which such impacts are ever considered by decision-makers, when and why.

Besides accounting for harm, another area for future work would be to focus on the potential upsides of rule-bending/breaking. At its best, such behavior can lead to markets for beneficial goods where none had existed (Fadahunsi and Rosa 2002) and beneficial changes to social morality (e.g., Fleming 2020). This is why society tends to tolerate it in entrepreneurs—it is worth some downsides when entrepreneurs often also bring creative destruction to social morality to help improve norms in ways that increase fairness, freedoms, compassion and justice for more people (Brenkert 2009). Entrepreneurship can create new contexts, but this also often means that new rules must also be created, which is a situation not well studied in our field, let alone one where our field has trained entrepreneurs to do it well (or specified with what processes or with which morally aligned partners it should be done). Such rule-making is an entrepreneurial activity that likely calls for additional improvisation (Casson 1998) in practice, in research and in theory.

Besides these more creative areas for future work, there are the more standard paths for further research to consider—the ones that simply extend the model(s)7 we have studied here. Extensions include sampling from non-US populations and adding further contextual characteristics, which could involve more specifics, such as industry indicators, regulatory maturity, and so on (which would appear as added control variables in regressions). One could also compare how different nations’ entrepreneurs (and their executive peers) feel pressure to bend their own rules for work, and to then compare what those rules—legal, moral and personal—are in those contexts.8 On a more fine-grained level, one could also make that comparison along gender lines across nations to determine any differential impact.9

6.2. Limitations

We realize that such implications are based on empirical studies, all of which have limitations, including our own. Our study suffers from the usual ones for survey-based research in entrepreneurship: sample restrictions involving limits on geography, recruitment, pool size and number of controls; survivor bias; common method bias; recall bias; and non-response bias (note, to the latter, we applied the standard proxy testing for differences between early and late responses and found none of significance). As such, care should be taken when applying our results under different conditions. So, to supplement the future work suggested above, we call for further studies on this phenomenon that use different locations, added controls (e.g., related to personal characteristics, like hubris—see (McManus 2018)—or related to firm characteristics, like strategy—see (Zona et al. 2013)), and so on, to check for robustness and extend the modeling.

6.3. Conclusions

We investigated the fundamental attribution of rule-breaking to entrepreneurs and found it to be an error. This result suggests that when we write and teach about entrepreneurial ethics, it will need to be contingent and grounded, rather than oxymoronic. It may be a convenient short-hand to attribute less ethical decision-making to small- and medium-sized venture operators, but it would be bad policy to target them for additional oversight. To set better policy—that encourages beneficial rule-following—the important contingencies (e.g., regarding individual, spiritual, organizational and industry characteristics) need to be considered more to identify proper targets and better mechanisms. This means that our field’s theorizing needs to be more complete to provide solid logic for such policy (e.g., it needs to extend basic models like the fraud triangle). Additionally, when it comes to complex phenomena like rule-bending, where there are both significant potential benefits and costs, we must more explicitly consider a wider range of contingencies (e.g., religiosity) to identify the best balance for helping our constituents do the right things well while questioning and improving not only what those things are, and how to do them, but also what is ‘right’ in the first place.