In recent decades, there have been episodic financial and economic debacles which were attributable, in part, to the ethical and legal lapses by top managers of publicly traded corporations: the savings and loan crisis of the 1980s (including the failures of Lincoln Savings & Loan and Silverado Savings & Loan), the dotcom bubble of the 1990s (including the failures of Global Crossing and WorldCom), the Enron bankruptcy of 2001, and the subprime mortgage crisis that became evident in 2008 with the collapses of Lehman Brothers and Countrywide Savings. The damage inflicted on investors, employees, and taxpayers makes it apparent that “bad ethics” can lead to catastrophic outcomes and reinforced arguments that “good ethics” is also “good business.”

A number of moral leadership theories emphasizing the importance of the moral or ethical aspects of leaders’ character and behaviors emerged during this period: personal integrity, authentic leadership, ethical leadership, servant leadership, spiritual leadership, transformational leadership, and virtual leadership (Cameron 2011; Cropanzano and Walumbwa 2010). Although the conceptualizations differ, they share at least two characteristics. First, their constructs include a dimension related to integrity (Palanski and Yammarino 2007, 2009; Reed et al. 2011). Second, the exercise of those leadership qualities and behaviors, including integrity, should lead to superior performance for organizations as well as positive outcomes (satisfaction, performance, and growth) for employees (Cropanzano and Walumbwa 2010).

There is a substantial organizational level empirical research about the relationships between corporate social/environmental performance and corporate financial performance (Orlitzky et al. 2003), but relatively little work on relationships between moral leadership and organizational or employees outcomes. The current research extends prior research on perceived behavioral integrity (BI) and employee attitudes and outcomes. First, it examines relations between BI and two additional variables: work engagement and work-to-family conflict. Second, and more importantly, it proposes a model whereby BI is seen as acting indirectly on employee outcomes through a mediating construct: moral distress.

Prior Literature

There are various definitions and meanings given to the term BI in the literature (Palanski and Yammarino 2007). BI is conceptualized herein as the “perceived pattern of alignment between an actor’s words and deeds. It entails both the perceived fit between espoused and enacted values, and perceived promise-keeping. … BI is the extent to which employees believe that a manager ‘walks her talk’, and, conversely, it reflects the extent to which they see her as ‘talking her walk’” (Simons 2002, p. 19). This definition is consistent as well with Palanski and Yammarino’s (2007) conclusion that it is the consistency between words and actions which best distinguishes “integrity” from other related constructs (such as personal wholeness or virtue). As Simons emphasizes (1999, 2002), the construct does not require that the manager must espouse ethical beliefs or ascribe to any particular norms; it simply calls for there to be congruence of agreement between the stated values and the actions taken. For example, there are recordings of Enron energy traders laughing as they related to each other how they were cheating the system and misleading their companies to make outsized trading profits (McLean and Elkin 2003). Many Enron managers expressed such beliefs themselves and rewarded rather than sanctioned the unethical behaviors. Accordingly, those managers would likely be viewed as exhibiting higher BI. A manager who read the words in Enron’s Code of Ethics, “Integrity: We work with customers and prospects openly, honestly, and sincerely. When we say we will do something, we will do it; when we say we cannot or will not do something, then we won’t do it” (Lay 2000, p. 3) but rewarded the overtly unscrupulous traders should be rated as having little BI. The construct also relates to the congruency between the actor’s words and deeds from the perspective of the employee. The perceptions of congruency may not reflect the reality but it is the perceptions rather than the reality which should impact employees’ attitudes and subsequent behaviors.

Behavioral integrity (with other forms of moral leadership that encompassed integrity) is posited to be related to organizational performance as employees’ trust in their leader would positively influence employee attitudes such as job satisfaction, organizational commitment, willingness to promote and implement espoused changed and job engagement and subsequent behaviors such as organizational citizenship behaviors (OCB) and in-role job performance. There is indeed evidence that honesty/integrity is related to leader effectiveness; Hoffman et al.’s (2011) meta-analysis reports ρ = .29 (k = 11, n = 3,123) between honesty/integrity and leader effectiveness. Simons further suggests that the difference between espoused and actual values and goals may dissipate employee effort, because the employee might not feel they know what it is the manager actually wants. Low PBI might then lead to role ambiguity and ineffective goal setting. Research indicates that role ambiguity is negatively related to job satisfaction, in-role performance, and OCB (Eatough et al. 2011).

Prottas (2008) also discusses the impact of BI primarily on its effects through employee attitudes. He suggests that while BI itself did not necessitate that espoused values and goals were necessarily ethical, it would be most common for managers to espouse socially acceptable and normative values of the type embodied in the Enron Code of Ethics. Managerial statements would be more likely to avow adherence to principles of distributive and procedural justice and promise that individuals who work hard and honestly to meet organizational objectives are likely to be rewarded. The failure to award rewards commensurate with performance would, according to equity (Adams 1963) and expectancy (Vroom 1964) theories, lead to dissatisfaction with outcomes and a reduction of effort. Prottas (2008) also suggests that stakeholder theory (Jones 1995) would be applicable as employees who distrusted their manager would expend a portion of their time and energy on monitoring and attempting to enforce their manager’s compliance with their stated “contract.”

Behavioral integrity should be related to employees perceiving violations of principles of both distributive and procedural justice. Leaders with low BI are unlikely to allocate rewards and punishments within the work setting according to stated criteria while following stated procedures. There is ample evidence that violations of both forms of justice are negatively related to desirable outcomes such as job satisfaction, outcome satisfaction, evaluation of authority, trust, compliance, organizational commitment, OCB, and work performance and positively related to undesirable outcomes such as withdrawal and turnover intentions (Cohen-Charash and Spector 2001; Colquitt et al. 2001). Additionally, Robbins et al.’s (2011) meta-analysis finds that perceptions of justice were related to indicators of employees’ physical and mental health.

Finally, most employees are likely to view themselves as being trustworthy and having integrity. If such employees perceive their manager as lacking integrity, there would be incongruence with respect to person–supervisor fit. Research finds that lack of person–supervisor fit is negatively related to job satisfaction and supervisor satisfaction (Kristof-Brown et al. 2005). Further, the supervisors’ values likely influence the employees’ perceptions of the organization’s values. Person–organization value-based incongruence has been found to be negatively related to job satisfaction and organizational commitment and positively to turnover intent (Kristof-Brown et al. 2005; Verquer et al. 2003).

There is growing evidence that BI is related to employee attitudes. Davis and Rothstein (2006) conducted a meta-analysis on 12 empirical studies of the relationship between BI and employee attitudes (with a combined n of 3,026) and reported that BI is positively related to employees’ attitudes. The effect size almost reaches Cohen’s (1992) threshold for “large”: BI’s uncorrected correlation with job satisfaction is .50 (the uncorrected correlation of BI was .48 against a wider variety of positive job-related attitudes including job satisfaction, organizational, satisfaction with leader). Subsequent studies report similar results: Prottas (2008) reports a correlation of .44 between BI and job satisfaction using a US representative sample of workers (n = 2,820) and Simons et al. (2007) reports correlations of .64 and .54 between BI and job satisfaction and organizational commitment in an international sample of employees of a single hotel company (n = 1,944). Prottas (2008) further finds that BI was positively related to life satisfaction and negatively to stress and poor health.

There is more limited evidence about the relationship between BI and behavioral intentions. Prottas (2008) reports BI was significantly and negatively related to absenteeism but the effect size does not reach Cohen’s (1992) threshold for “small,” while Johnson and O’Leary-Kelley (2003) find that breach of psychological contract (i.e., breaking promises) is related to absenteeism and in-role behaviors but not to OCBs. Simons et al. (2007) report BI is negatively related to turnover intention (r = .44) and Simons (2009) finds BI is positively related to affective commitment and trust in managers (paths coefficients of .61 and .82, respectively).

There is some evidence of a relation between BI and organizational performance. Simons (2009) tests a model of the impact of BI on employee attitudes and organizational performance in a study of employees of an international hotel chain. He proposes that BI → trust in management → employee commitment → discretionary service behavior → guest satisfaction → profits (his model also includes employee commitment → turnover → profits). Using path analysis, he finds BI is directly related to hotel profitability with additional indirect and positive effects through trust in manager and affective commitment which → discretionary service behavior → guest satisfaction → hotel profitability. Affective commitment also → employee turnover which → hotel profitability. In a series of studies, Palanski and Yammarino (2011) also find evidence that the effect of BI on performance is primarily indirect, with trust as a mediator.

There is no research on BI and its relationship to job engagement and work-to-family conflict. Job engagement is an attitudinal variable whose antecedents include perceptions of procedural and distributive justice and supervisor support and which is viewed as an antecedent of job satisfaction, organizational commitment, intention to quit, job performance, and OCB (Christian et al. 2011; Saks 2006). Work-to-family conflict relates to the negative consequences of individuals attempting to actively participate in work roles as employees and in home-related roles as parents and spouses. Supervisor support appears to be negatively related to work-to-family conflict as an antecedent (Mesmer-Magnus and Viswesvaran 2006), and work-to-family conflict seems to be negatively related to outcomes such as job satisfaction and organizational commitment, and OCB and positively to turnover likelihood (Amstad et al. 2011).

In summary, there are a number of theories and ample empirical evidence that suggests employees’ perceptions of the BI of their managers will influence their attitudes toward the job as well as other outcomes. Accordingly, I hypothesize that:

Hypotheses 1a–g

BI will be related to job satisfaction, stress, job engagement, turnover likelihood, work-to-family conflict, health, and life satisfaction.

Employees have emotional and cognitive reactions to their own lack of BI. Employees who engage in actions and behaviors which contradict their stated values tend to feel anxiety, stress, and a sense of self-rejection (Festinger 1957; Kaplan 1983; Hobfoll 1989). To the extent they believe they acted against their values because of environmental or situational influences, they would likely feel dissatisfied with the environment and want to leave it. Healthcare practitioners and academics have labeled the “phenomenon in which one knows the right action to take, but is constrained from taking it” as moral distress (Jameton 1984; Schluter et al. 2008; Suhonen et al. 2011). Moral distress is associated with stress, anger, depression, reduction of effort, and turnover among nurses (Rittenmeyer and Huffman 2009). Accordingly, I hypothesize that:

Hypotheses 2a–g

Moral distress will be negatively related to job satisfaction, stress, job engagement, turnover likelihood, work-to-family conflict, health, and life satisfaction.

The attitudes and behaviors of managers also influence their subordinates’ behaviors. A manager with lower ethical standards than their employees might cause those employees to act in ways that violate the employee’s own value systems (Vardi and Weitz 2004). One function of managers is to model behaviors that are expected to be replicated by employees. Social learning theory (Bandura 1986) suggests that the extent to which subordinates view their managers as having BI influences their own levels of BI. Managers also allocate institutional rewards such as pay and promotions (Dineen et al. 2006). A manager with low BI is more likely to reward a subordinate who exhibits low BI than an employee who displays greater BI (Palanski and Yammarino 2007, 2011). A reward system influences behaviors of subordinates. Palanski and Yammarino (2011) reported that followers’ BI was related to their leaders’ BI. In other words, a person with a manager lacking integrity would be more likely to engage in behaviors that violate their own conscience. Therefore, I hypothesize that:

Hypothesis 3

Individuals’ feelings of moral distress will be related to their perceptions of the BI of their managers.

Prior research has focused on the direct relationships between managers’ PB and employees’ attitudes. Managers’ PB may also influence employees’ behaviors. Managers with low PB may influence their subordinates to engage in behaviors that cause the employees to violate their own stated values. The resulting moral distress would influence the employees’ attitudes. Accordingly, I hypothesize that:

Hypothesis 4

Moral distress will mediate the relationships between BI and the variables.

Methods

Sample

This study used data from the 2008 National Study of the Changing Workforce (2008 NSCW) (Families and Work Institute, 2008). The 2008 NSCW survey was conducted by Harris Interactive, Inc. using a questionnaire developed by the Families and Work Institute. A total of 3,502 interviews were completed with a nationwide cross-section of employed adults from November 2007 to April 2008. Telephone interviews averaged 50 minute in duration. In this research, I used unweighted data from the subsample (n = 2,679) of wage and salary workers (i.e., excluding self-employed). FWI estimates a response rate of 55 % of potentially eligible households.

Measures

Demographic information included gender, age, and education level (less than high school diploma, high school or GED, some college, no degree, associate degree, 4-year college degree). Educational level was treated as an ordinal variable in the analysis.

Perceived BI was assessed by two items: “I can trust what managers say in my organization” and “managers in my organization behave honestly and ethically when dealing with employees and clients or customers” with four Likert-type response options from strongly agree to strongly disagree. Responses were averaged and scored so that higher meant higher BI. The Cronbach alpha (α) measure of internal reliability was .83.

Moral distress was assessed by a single item: “On my job, I have to do some things that really go against my conscience” with the above four response options. Responses were scored such that higher values represented more moral distress.

Job satisfaction was assessed by three items including “All in all, how satisfied are you with your job” (with the above four response options), and one item, “Knowing what you know now, if you had to decide all over again whether to take the same job you now have, what would you decide?” with three Likert-type response options (decide without any hesitation to take the same job to decide definitely not to take the same job). Responses were standardized and averaged with higher scores indicating greater satisfaction (α = .78).

Job engagement was assessed by seven items: five items asked the extent to which participants agreed with statements such as “When I’m at work, time passes very quickly” (four response options from strongly agree to strongly disagree), one item asking “How much effort do you put into your job beyond what is required” (four response options from a lot to none) and one item asking “How often do you think about good things related to your job when you’re busy doing something else?” with five response options from very often to never). Items were standardized and scored so that higher values represented greater engagement (α = .69).

Turnover likelihood was assessed by a single item, “Taking everything into consideration, how likely is it that you will make a genuine effort to find a new job within the next year?” with three response options from very likely to not at all likely. Responses were scored so that higher values meant higher likelihood.

Stress & strain was assessed by 10 items. Seven items asked participants how often in the last month they had suffered from symptoms such as feeling nervous or having problems sleeping (five response options from never to very often); two items asked whether they had been bothered in the past month by feeling down, depressed or helpless and whether in the past month they had felt little pleasure in doing things (yes or no); one item asked “Not thinking about work, how stressful has your personal and family life been in recent months?” with five response options from extremely stressful to not stressful at all. Items were standardized and averaged with higher scores indicating greater stress (α = .83).

Health was assessed by a single item: “How would you rate your current state of health?” with four response options (excellent to poor). Responses were scored so that higher values indicated better health.

Work-to-family conflict was assessed by five items such as “How often have you not had the energy to do things with your family or other important people in your life because of your job?” with five response options from very often to never. Responses were averaged and scored such that higher values indicated more conflict (α = .86).

Absenteeism was assessed by a single item: “thinking back over the past three months, how many times have you missed any work – either part days or full days – that were NOT scheduled in advance with the approval of your supervisor or manager?”

Results

A slight majority of the 2,679 participants were female (54.7 %) and almost half (44.5 %) had an associate or a 4-year college degree. Their average age was 45.9 years (SD = 12.3). Basic statistics and correlations are shown in Table 1. The large sample size provided for statistical power to detect effect sizes that were statistically but not practically significant. Accordingly, this research will discuss only relationships that were statistically significant and exceeded Cohen’s (1992) thresholds for small: .10 for correlations (.30 for medium and .50 for large) and .02 for f 2 for regression (.15 for medium and .35 for large). As hierarchical regression is being used in the analysis, the effect size for any variable will be ∆f 2 (=∆R 2/(1 − ∆R 2). Using these criteria, the demographic variables were unrelated to either BI or moral distress. This is consistent to prior studies which showed no relationship between gender and BI (Davis and Rothstein 2006; Prottas 2008; Simons et al. 2007) and no relationship between age and BI (Prottas 2008; Simons et al. 2007).

Table 1 Basic statistics and correlations
Full size table

The zero-order correlations between BI, moral distress, and each of the dependent variables are shown in Table 1. BI and moral distress were similarly related to the other variables but the directions of relationships were reversed. For example, job satisfaction was positively related to BI and negatively to moral distress, and stress was negatively related to BI and positively to moral distress. I tested for differences with respect to the strengths of the relationship. As the correlations were dependent rather than independent, I calculated Steiger’s Z to determine if the differences in the absolute values were statistically significant. As shown in Table 2, the correlations between BI and three of the variables were stronger than their corresponding correlations with moral distress: job satisfaction, work engagement, and life satisfaction (although the difference between the correlation coefficients with respect to life satisfaction did not quite reach the .10 threshold for small).

Table 2 Differences between correlations with behavioral integrity and moral distress
Full size table

To test my hypotheses, I performed a series of hierarchical regressions. I regressed each of the dependent variables on BI, after controlling for age, gender, and education level (Model 1a). I also regressed each of the dependent variables on moral distress, after controlling for age, gender, and educational level (Model 1b). I then conducted a three-step regression of each of the dependent variables with step 1 entering the demographic variables, step 2 entering BI, and step 3 entering moral distress (Model 2). The results of the regressions are shown in Tables 3, 4, and 5. In order to save space, in Tables 3, 4, and 5, I present the results of second and final step for Models 1a and Models 1b and the results of third and final steps only for Models 2.

Table 3 Hierarchical regressions and results of Sobel test of mediation
Full size table
Table 4 Hierarchical regressions and results of Sobel test of mediation (cont.)
Full size table
Table 5 Hierarchical regressions and results of Sobel test of mediation (cont.)
Full size table

To test for mediation I first regressed the proposed mediator (moral distress) on BI after controlling for age, gender, and education. I then regressed each of the dependent variables on BI and the proposed mediator (moral distress). I calculated the Sobel test statistics using an on-line interactive calculator that required us to input betas and standard errors from the above regressions (Preacher and Leonardelli 2010). The results of those analyses are also shown in Tables 3, 4, and 5.

My hypotheses 2a–g were supported as the beta coefficients for BI (Tables 3, 4, 5—Models 1a) were significant when BI was entered as a second step (after control variables of age, gender, and education) against dependent variables of job satisfaction, stress, work engagement, work-to-family conflict, overall health, and life satisfaction. The effect sizes of the relationships with the work-related attitudes (∆f 2 = .37 and .27 for job satisfaction and job engagement) were larger than those with work-oriented behavioral intention or inter-domain conflict (∆f 2 = .17 and .10 for turnover likelihood and work-to-family conflict) or non-work attitudes (∆f 2 = .05, .02, and .05 stress, overall health, and life satisfaction). Given Prottas’ (2008) prior findings using data from the 2002 NSCW, I made no hypothesis with respect to absenteeism and found no relationships.

My hypotheses 3a–g were largely supported as the beta coefficients for moral distress (Tables 3, 4, 5—Models 1b) were significant when moral distress was entered as a second step (after control variables of age, gender, and education) against dependent variables of job satisfaction, stress, job engagement, work-to-family conflict, overall health, and life satisfaction. The effect size of the relationship with job satisfaction (∆f 2 = .10) was somewhat larger than those for the other dependent variables: job engagement, .04; stress & strain, .03; work-to-family conflict, .07; turnover likelihood, .03; and life satisfaction, .02. The effect size with respect to health did not reach the threshold for small, and moral distress was not related to absenteeism.

Hypothesis 3 was supported as BI, and moral distress were negatively correlated (r = .35, p < .001); when BI was entered in the second step of a hierarchical regression (after age, gender, and educational level) with moral distress, its beta coefficient was negative and significant (−.33, p < .001) with ∆R 2 of .11 (f 2 = .12).

I conducted the three-step hierarchical regressions to determine whether moral distress explained additional variance after the control variables and BI. The results are shown in Model 2 in Tables 3, 4, and 5. Moral distress explained additional variance only with respect to job satisfaction and work-to-family conflict.

My hypotheses with respect to mediation were largely supported as the Sobel statistic for mediation was significant with respect to job satisfaction, turnover likelihood, stress & strain, work-to-family conflict, health and life satisfaction. The statistic for job engagement was not quite statistically significant (p = .057).

Discussion

A number of leadership theories assert that integrity is a crucial characteristic of the effective leader who creates an environment which both benefits the employees and enhances organizational performance. The theory and research to date have emphasized the effects of the leader’s behaviors, as perceived by the employees, on the employees’ attitudes. To the extent that the leader was trusted, employees were viewed as more likely to engage in constructive and productive behaviors. Some of the findings of this study using the 2008 NSCW study merely replicate what Prottas (2008) reported in his study of the 2002 data: BI’s correlation with job satisfaction (.53 vs. .44), life stress (−.24 vs. −.21), health (−16 vs. −.12) life satisfaction (.24 vs. .19), and absenteeism (−.06 vs. −.06). As the measures were very similar, the similarities of relationships are to be expected; however, the consistency provides confidence as to the robustness of the relationships. The findings that BI is related to other outcomes such as work engagement (positively) and work-to-family conflict (negatively) represent a minor expansion of our knowledge of BI.

The findings with respect to the relationships between moral distress and the other outcomes suggest that relationships found within the healthcare profession may be generalizable to people working in less morally and ethically intense and challenging professions. However, the principle contribution of this research relates to expanding the model of relationships to allow for the influence of PB to be related to the behaviors of the employees and not only to their attitudes. The current research finds support for a model whereby leaders’ behaviors are related to followers’ behaviors which then influence employees’ attitudes.

While these findings are contributory, there are limitations to this study. First, the cross-sectional nature of the research design cannot provide support for the causality of the relations. It is possible to argue that individuals who experience moral distress because they engage in acts that violate their conscience may wish to preserve their self-image by ascribing low BI to their superiors. As Festinger (1957) recognized, one form of eliminating dissonance may be by distorting perceptions of reality.

Second, all of the data came from the same source with the potential for the biases that could either attenuate or accentuate the relationships (Podsakoff et al. 2003). The length of the telephone survey (50 minutes) and the different formats of the questions likely reduce the potential for some forms of bias (e.g., consistency, acquiescence, and self-presentation). Most of the key constructs such as perceived rather than actual BI, moral distress, and job satisfaction relate to internal states which could not be evaluated by others.

Third, the use of a single item to assess moral distress has to be recognized as a significant limitation. Single item measures are likely to be less reliable than a scale consisting of multiple items and low reliability of measures attenuates the relationships among variables. Future research should use a reliable and construct validated measure appropriate for the sample population (such as Craig and Gustafson 1998).

Fourth, Simons’ (1999, 2002) and Palanski and Yammarino’s (2007, 2009) conceptualizations of integrity emphasize the congruency between an actor’s espoused values and behaviors and did not require that the espoused values themselves satisfy any normative ethical standards. Someone with high BI could be very unethical in both words and deeds. For example, in relating the Enron tale, McLean and Elkin (2003) describe Tim Belden, the head of the energy traders, as defending his actions by asserting that lying about one’s position was common practice and then describing an ethical climate where the managers would encourage and reward traders for acting in “practices which made a mockery of Ken Lay’s exhortations about Enron’s high ethical standards” (p. 275). It is likely that Belden’s talk was consistent with his walk: the Enron traders were indeed handsomely rewarded for coming up with trading practices that depended on deception, coercion, violation of the spirit and perhaps letter of the law, and caused financial and other harm to millions of people. One would hope that most instances of consistence between stated values and actual behaviors occur when both are at a higher ethical level, but some, hopefully a minor portion, may occur when both are at a very low ethical standard. The analysis in this research assumes that discrepancies between the “talk” and the “walk” generally arose from the “talk” being of normative ethical type found in Enron’s Ethics policy and with the “walk” being less ethical. Naturally, it is possible that some incongruency occurs when managers state that they conduct business through cheating and deception but actually reward honesty and transparency, but that seems unlikely.

Fifth, and perhaps most importantly, the conceptualization of BI is essentially a difference score. Edwards and others (Edwards and Parry 1993; Edwards 1993) have criticized difference scores on conceptual and methodological grounds. It is probable that effects on employees would be different if the manager’s “talk” was ethical but the “walk” was unethical than it would be if the manager’s “talk” was unethical but the “walk” was ethical. Additionally, the effect of a discrepancy between strongly voiced ethical standards and weakly voiced standards might differ, even if the magnitudes of the discrepancies between the talks and the walks were the same.

Future research should collect and analyze longitudinal data to tests models in which both trust and moral distress mediate BI’s relationships on employee attitudes and performance (in-role and OCB) and in which these behaviors are related to organizational performance and outcomes. Future research might also develop measures which address some of the limitations related to difference scores.