Introduction

Over the last decade, high-profile incidents of leaders’ ethical failure in organizations such as Enron, WorldCom, and Tyco have increased both scholars’ and practitioners’ attention to the ethical aspects of leadership. Despite this attention, and the increasing pressure on corporate leaders to behave ethically, new incidents of ethical failure continue to emerge, such as the recent Volkswagen emission test scandal. Brown et al. (2005) define ethical leadership (EL) as “the demonstration of normatively appropriate conduct…and the promotion of such conduct to followers through two-way communication, reinforcement, and decision-making” (p. 120). Scholars studying ethical leadership behavior in organizations have empirically demonstrated its positive links with important work outcomes such as employee ethical conduct (Mayer et al. 2012), organizational citizenship behavior (Babalola et al. 2017; Newman et al. 2014; Ogunfowora 2014a), voice and employee cynicism (Pelletier and Bligh 2008; Walumbwa and Schaubroeck 2009), reduced workplace conflicts (Babalola et al. 2016), performance (Zhu et al. 2015), and (ethical) job applicant attraction (Ogunfowora 2014b).

Although the majority of this work has focused on the consequences of ethical leadership, comparatively less research has addressed its antecedents (e.g., Kalshoven et al. 2011; Mayer et al. 2012; Walumbwa and Schaubroeck 2009; Zhu et al. 2016). For instance, Mayer et al. (2012) and Zhu et al. (2016) have shown that morally oriented individual differences are associated with perceptions of ethical leadership. Specifically, leaders with high moral identity (i.e., those who self-identify and believe themselves as ethical; Mayer et al. 2012) and high moral attentiveness (i.e., those who chronically perceive and consider morality and moral elements in their experiences; Zhu et al. 2016) are more likely to be perceived as ethical leaders. While interesting insights have emerged from these few studies, they still remain limited in that they do not provide deeper knowledge of the psychological mechanisms and boundary conditions associated with the emergence of ethical leadership. Thus, extant research studies insufficiently address the question of why and when some leaders engage in ethical leadership behaviors. This is the primary goal of the current study.

A limited but growing stream of research on leader personality traits suggests that leader conscientiousness and agreeableness are important antecedents of ethical leadership behavior (Kalshoven et al. 2011; Walumbwa and Schaubroeck 2009). Scholars (e.g., Den Hartog 2015) have however argued that, of the Big Five personality traits, conscientiousness seems most crucial for the explicit ethical focus that distinguishes ethical leadership from related leadership behaviors, primarily due to its moral cognitive foundations (Costa and McCrae 1992). While establishing and explicitly focusing on conscientiousness as an antecedent of ethical leadership is an important first step, this approach is likely to be limited in utility and overly simplistic for two major reasons. First, whereas Walumbwa and Schaubroeck (2009) found a moderate association between conscientiousness and ethical leadership, Kalshoven et al. (2011) reported evidence of a relatively weaker relationship. This suggests that the strength of this relationship may vary depending on situational or other moderating factors. Second, scholars have debated for several decades whether personality traits such as conscientiousness hold any value for reliably predicting leadership behavior (Barrick and Mount 2005; Judge et al. 2002). In light of varying effects and entrenched debates, Barrick and Mount (2005) emphasized that “systematically and carefully studying mediating and moderating effects is precisely where we need to go in personality research” (p. 369). Thus, understanding the processes and contexts that determine when and how employees are more likely to recognize conscientious leaders as ethical is crucial to developing our theoretical understanding of ethical leadership, as well as to aid practical efforts to recruit and develop ethical leaders in organizations.

Accordingly, our goal in the present research is to shed light on the relationship between leader conscientiousness and ethical leadership in light of existing debates in the personality trait approach to leadership. To do so, we draw on social cognitive theory (SCT; Bandura 1986, 1991), which offers a promising theoretical framework for deepening our knowledge about this relationship. A core premise of SCT is that personality shapes an individual’s behavior through proximal reflective mechanisms. Since conscientiousness taps into aspects of individuals’ thoughts and behaviors that are deemed morally oriented (Horn et al. 2004), we use SCT as a theoretical basis to argue that leader conscientiousness is likely to lead to ethical leadership behaviors by stimulating moral reflectiveness, which refers to the extent to which leaders contemplate moral matters in their daily experiences and decisions (Reynolds 2008).

Furthermore, we propose that the leader’s specific work context may facilitate or hinder the extent to which his or her moral reflectiveness ultimately results in observable ethical leadership behaviors. SCT suggests that the environment or situational context in which an individual finds him/herself is a crucial factor that either strengthens or constrains the extent to which domain-specific cognitive reflections invoke relevant behavioral actions (Bandura 1991). From this perspective, the impact of a leader’s moral cognitive reflections may depend on situational contexts that encourage or inhibit the behavioral expression of moral reflectiveness. We propose that decision-making autonomy is one such contextual factor that may hinder or foster the extent to which leader moral reflectiveness translates into observable ethical leadership behaviors. According to Brown et al. (2005), ethical leadership involves more than simply possessing ethical personal qualities (i.e., ethically oriented traits and cognitions); leaders must also engage in “moral manager” behaviors such as setting high ethical standards and establishing reinforcements to uphold those standards. However, a leader’s capacity to engage in moral management behaviors is significantly limited if he/she lacks autonomy to make relevant decisions such as determining the work unit’s ethical standards or the rewards (punishments) for upholding (violating) them. It is therefore surprising that leader decision-making autonomy has received limited theoretical and empirical attention to date in the ethical leadership literature. As such, the second goal of our research is to explore the moderating role of leader decision-making autonomy in understanding when leader conscientiousness (and moral reflectiveness) translates to ethical leadership behavior.

Our research offers at least two major contributions to research on ethical leadership. First, in contrast to past research focusing on the direct relationship between conscientiousness and ethical leadership, we examine moral reflectiveness as a potential mediatory mechanism that underlies this relationship. Second, our research answers a recent call to identify potential boundary conditions of ethical leadership antecedents (Den Hartog 2015) by investigating the moderating role of leader decision-making autonomy. In doing so, we contribute to the literature by illuminating circumstances in which morally reflective leaders may not necessarily demonstrate or be seen as displaying ethical leadership behaviors. Taken together, our theoretical model (see Fig. 1) addresses Barrick and Mount’s (2005) call to build and test theory about the mechanisms and contexts that explain why and when personality may foster specific leadership behaviors.

Fig. 1
figure 1

Theoretical model

Full size image

Theoretical Background and Hypotheses

Overview of Social Cognitive Theory

“Most human behavior, being purposive, is regulated by forethought” (Bandura 1991, p. 248). According to SCT (Bandura 1986, 1991), an individual’s specific cognitive reflection in any given domain is central to predicting his or her behavioral course of action in that domain. SCT suggests that domain-relevant behaviors depend first on individual personality, but also on the cognitive reflection that accompanies the specific trait. In this way, SCT conceptualizes domain-relevant behaviors as a process that develops through reflectiveness in that domain, starting as a result of relevant personality traits and ending with behavioral actions. We draw on SCT to suggest that ethical leadership behaviors are rooted in a similar process of personality–cognitive reflection–behavioral action. Given that the Big Five trait most often seen as the standard for moral evaluation is conscientiousness (Horn et al. 2004), we argue that leader conscientiousness (a personality trait) is likely to stimulate leader moral reflectiveness (a morally inclined cognitive reflection), which subsequently induces ethical leadership behavior (a morally informed behavioral action), with the strength of this indirect relationship depending on a situational factor that constrains the expression of ethical leadership behavior (i.e., decision-making autonomy).

In building conceptual support for our model, we draw on Reynolds’ (2008) work on moral attentiveness, a sociocognitive construct that captures the extent to which people habitually attend to moral issues in their social interactions. Reynolds proposes that there are two aspects to moral attentiveness: perceptual (the extent to which an individual chronically searches for and focus on the moral aspects of social information) and reflective (a more nuanced cognitive process that emphasizes the use of morality in examining social information). We focus on moral reflectiveness in our model because, while perceptual moral attentiveness is considered an automatic response and primarily entails information coding, moral reflectiveness is more intentional and emphasizes behavioral action (Reynolds 2008). This is important because ethical leadership entails more behavioral action than information coding (Brown et al. 2005), which is an important but insufficient driver of behavioral action. Moreover, our focus on moral reflectiveness builds upon past work on the link between ethical leadership and the broader moral attentiveness construct (e.g., Zhu et al. 2015). We propose that a more nuanced examination of this construct should enrich our understanding of why conscientious leaders are more likely to demonstrate ethical leadership behaviors at work. We now turn to this issue in more detail below.

Leader Conscientiousness and Moral Reflectiveness

To be “conscientious” is to be governed by one’s conscience, a foundation for moral evaluation (Horn et al. 2004). Individuals low in conscientiousness are seen as undependable, careless, and thoughtless. In contrast, conscientious individuals are reliable, self-disciplined, careful, and thorough (McCrae and Costa 1987). Conscientious individuals pay careful attention to detail, are deliberate rather than haphazard in their decision-making, have a strong sense of moral obligation, and act based on their conscience (Judge et al. 2009). These attributes suggest a possible link between conscientiousness and moral reflectiveness, which Reynolds (2008) defines as the extent to which an individual is guided by moral consideration in his or her daily experiences and decisions. Given that moral reflectiveness is a consciously informed, self-controlled process (Haidt 2001; Reynolds 2008), we propose that leaders with high conscientiousness routinely reflect on the extent to which their behavior is morally appropriate. In addition, they are likely to subsequently use such reflection as a guide for their behavioral actions since they tend to value honesty and think carefully before acting.

Accordingly, as conscientious leaders go about their daily lives, they likely become more conscious and thoughtful regarding the normative aspects of their actions. Conscientious leaders will therefore pursue morality, reflect moral values (Collins and Schmidt 1993), and more regularly reflect on morality in their daily experiences (Kim et al. 2014). Therefore, based on the above theoretical arguments and empirical evidence, we expect that for people occupying leadership roles, conscientiousness will be associated with higher levels of moral reflectiveness:

Hypothesis 1

Leader conscientiousness is positively related to leader moral reflectiveness.

Leader Moral Reflectiveness and Ethical Leadership

At the heart of SCT (Bandura 1986, 1991) is the idea that reflection and forethought in a specific domain induces actionable behaviors in that particular targeted domain (Bandura 1986, 1991). As a result, moral reflectiveness should prompt leaders’ demonstration and encouragement of ethical behaviors as they fulfill their leadership roles and responsibilities. Examples of such morally informed leadership behaviors include acting in the best interests of employees, actively listening to what they have to say, discussing the implications of unethical behaviors with them, and generally being a role model for ethical conduct (Brown et al. 2005). These ethical leader behaviors are more likely to “shine through” to others when leaders consider moral matters on a regular basis. Indeed, previous research has generally supported the notion that individual’s moral reflectiveness is an important basis for the internalization of morally oriented values and behaviors (Kohlberg 1981; Reynolds 2008). Thus, consistent with prior research and SCT, we argue that leaders’ chronic attention to and consideration of morality and moral matters in daily work interactions should lead to the demonstration of normatively appropriate conduct and the promotion of ethics to employees as reflected in their decision-making (i.e., ethical leadership).

In line with the above argument, a recent study by Kim et al. (2014) found that leaders who reflected on moral matters in their experiences pursued morally informed behaviors to protect and save their organization’s resources, as well as engaged in actions that contribute to environmental sustainability. Zhu et al. (2016) also argued and found that individuals who are high on overall moral attentiveness (including both perceptual and reflective moral attentiveness) were more likely to be perceived as ethical leaders. Therefore, we argue that enhanced moral reflectiveness associated with leaders’ conscientiousness should guide and motivate leaders to engage in ethical leadership behaviors. Taken together, since ethical leadership is regarded as an explicit demonstration and expression of a leader’s moral value of being a moral person and moral manager, moral reflectiveness should serve as an internal reflective process that not only guides the leader to display ethical conducts but also motivates him or her to encourage and promote ethical behavior in the workplace.

Hypothesis 2

Leader moral reflectiveness is positively related to perceptions of ethical leadership.

A key proposition in our model is the mediating role of leader moral reflectiveness in expanding our understanding of the relationship between leader conscientiousness and ethical leadership. To this point, we have argued that leader conscientiousness is positively linked to moral reflectiveness and that leader moral reflectiveness in turn is positively associated with ethical leadership. That is, moral reflectiveness serves as the proximal mechanism that underlines the distal relationship between leader conscientiousness and ethical leadership behavior.

Given the morally laden definition of conscientiousness (Becker 1998; Costa and McCrae 1992) and the social cognitive origin of moral reflectiveness (Reynolds 2008), the application of SCT in understanding the relations between conscientiousness and ethical leadership suggests that the effect of leader conscientiousness (a distal individual personality predictor) on moral reflectiveness (a moral-specific forethought) shapes ethical leadership (a leader’s morally informed behavioral action). This is because a leader’s concern about morality underscores his or her motivation to demonstrate normatively appropriate behaviors in the workplace (Brown et al. 2005). Therefore, we propose that leader conscientiousness stimulates ethical leadership behavior through its linkage with moral reflectiveness.

Hypothesis 3

Leader moral reflectiveness mediates the relationship between leader conscientiousness and ethical leadership

The Moderating Role of Decision-Making Autonomy

We further propose that the strength of the mediatory effect of moral reflectiveness depends on the leader’s decision-making autonomy. As earlier noted, one of the basic tenets of SCT (Bandura 1986, 1991) is that the “situation” an individual finds him/herself is an important conditional factor that either strengthens or constrains the behavioral manifestation of one’s cognitive reflection or forethought. Reynolds (2008) has similarly proposed that moral reflectiveness should be more influential in contexts where an individual must recognize a moral issue, consider its implications, deliberate alternate options, and ultimately announce a course of action. In this paper, we propose that the behavioral expression of moral reflectiveness is likely to be either strengthened or constrained by the leader’s decision-making autonomy—i.e., the extent to which the leader has freedom to make decisions about job-related issues (Karasek 1979). Specifically, ethical leadership behaviors are more likely to occur if a morally reflective leader, having given much thought and consideration to ethical issues, has greater autonomy to make pertinent decisions on the job.

The extant literature suggests that beyond being a moral person, the unique strength of an ethical leader is to actively manage morality on a day-to-day basis by being a moral manager (Brown et al. 2005; Mayer et al. 2012). This entails behavioral actions such as defining success not just by the results but also the way they are obtained, disciplining employees who violate ethical standards, asking “what is the right thing to do?” when making decisions, and setting an example of how to do things the right way in terms of ethics. We argue that these ethical leadership behaviors are more likely to occur when morally reflective leaders have high decision-making autonomy. For instance, leaders, in order for ethical leaders to address the core question of “what is the right thing to do” when faced with a critical decision, it is important for them to possess decision-making latitude (Brown et al. 2005). As such, having decision-making autonomy should enhance the extent to which morally reflective leaders can better take others’ considerations into account since they are able to act upon such inputs. As a result, decision-making autonomy should enhance the extent to which morally reflective leaders engage in observable ethical leadership behaviors.

In contrast, low levels of decision-making autonomy suggest that the leader has little discretion or control in terms of his or her responsibilities and behavioral strategies for fulfilling those responsibilities (Hackman and Oldman 1976). Under this condition, a leader’s moral reflectiveness may not always lead to observable ethical leadership behavior because he/she is unable to take meaningful actions, including those with ethical implications. In organizations, leaders are often faced with situations that may question their stance of being fair and moral (Gino and Mogilner 2014). Even though conscientious leaders value honesty in their personal and work life (Costa and McCrae 1992), and may reflect more frequently on moral issues in their daily decisions and experiences, being in a work situation where they do not have sufficient freedom to make decisions is likely to create less opportunities for them to behave in ways consistent with their morally inclined reflection. As such, the relationship between leader moral reflectiveness and ethical leadership should be weaker in these contexts. In other words, low decision-making autonomy is likely to weaken the extent to which a conscientious leader’s moral reflectiveness leads to the successful demonstration ethical leadership behavior as perceived by subordinates. Lastly, we expect that decision-making autonomy will have no significant influence on the ethical leadership conduct of leaders who are low on conscientiousness. Generally, these leaders are unlikely to engage in ethical leadership behaviors because they are less likely to habitually reflect on issues of morality in their day-to-day work life. For these individuals, therefore, having greater or less decision-making autonomy should have minimal impact on the extent to which they engage in ethical leadership behaviors.

Hypothesis 4

The indirect relationship between leader conscientiousness and ethical leadership through moral reflectiveness is moderated by decision-making autonomy, such that moral reflectiveness engenders ethical leadership when decision-making autonomy is high, but not when decision-making autonomy is low.

Research Overview

We test our hypothesized model in two samples using two different populations and methodologies (i.e., multisource and three-wave field studies). In Sample 1, we examine the proposed model using a cross-sectional sample of leaders and their direct reports in Nigerian organizations. We then replicate and test this model in Sample 2 using a three-wave multisource sample of leaders and their direct reports in Chinese organizations. Given that both Nigeria and China are situated in a cultural context where decision-making tends to be hierarchically structured (Hofstede, 2001), they provide two ideal samples for testing our model (Fig. 1).

Method

Sample 1

Participants and Procedure

Sample 1 consisted of workgroup leaders and their direct reports from three companies in the Nigerian consulting, hospitality, and financial industries. Prior to survey administration, human resource personnel in each of the participating firms, who also served as our internal contacts, provided a list of workgroup leaders and employees who worked directly under them from which we randomly selected potential participants. Participants were informed through a cover letter that the purpose of the survey was to examine effective leadership development, and that their responses would be confidential. We also emphasized the voluntary nature of participation. We administered two separate questionnaires to workgroup leaders and their direct reports. The questionnaires included a customized personal code based on the list received from the HR officer in order to enable us link the two surveys. With the help of our contacts, surveys were distributed on-site and directly returned to the research team after completion. To thank participants for their participation, they were entered into a drawing to win one of four online shopping vouchers priced at approximately US$15 each.

Of the 50 workgroup leader and 200 direct report surveys distributed, 34 leaders and 133 direct reports returned completed questionnaires (approximately 68% response rate). Six leader surveys were discarded because of either unmatchable or unavailable corresponding direct reports, thereby yielding an effective overall response rate of 56%. As such, the final sample consisted of 28 workgroup leaders and 115 direct reports. The size of each workgroup ranged from three to seven members, with an average group size of 4.5 workgroup members. The average organizational tenure was four years, and employees reported spending an average of 2.6 years with their immediate leader.

Measures

Surveys were administered in English, the official language in Nigeria. All items were measured on a 5-point scale (1 = strongly disagree, 5 = strongly agree).

Leader Conscientiousness

We measured leader conscientiousness using a four-item scale in the Mini International Personality Item Pool (Mini-IPIP; Donnellan et al. 2006). Work group leaders rated the following items: “I get chores done right away”, “I like order”, “I make mess of things” (Reverse-coded), and “I often forget to put things back in their proper place” (Reverse-coded). The internal consistency reliability was .77. Past empirical research (Kim et al. 2014) has demonstrated the convergent validity of the Mini-IPIP with Goldberg’s (1999) conscientiousness scale.

Moral Reflectiveness

We measured leader moral reflectiveness using a 5-item scale developed by Reynolds (2008). The sample items include, “I often reflect on the moral aspects of my decisions”, and “I regularly think about the ethical implications of my decisions”. The internal consistency reliability was .75.

Decision-Making Autonomy

We assessed decision-making autonomy using Morgeson and Humphrey’s (2006) 3-item scale. Sample items include, “This job gives me a chance to use my personal initiative or judgment in carrying out the work”, and “This job provides me with significant autonomy in making decisions”. The internal reliability of the scale was .88.

Ethical Leadership

Employees provided ratings of ethical leadership using Brown et al.’s (2005) 10-item scale. Sample items include: my manager/supervisor… “makes fair and balanced decisions”, and “defines success not just by results but also the way that they are obtained”. The internal reliability of the scale was .89.

Analytic Strategy and Levels of Analysis

First, we sought to establish our measurement model. We carried out multilevel confirmatory factor analyses in Mplus (MCFA; Muthén and Asparouhov 2009) to ensure that the scales loaded on their intended constructs. We used MCFA because employee responses to the ethical leadership scale are not independent, given that our participants were nested in teams. In our hypothesized factor structure, we modeled within- and between-group variances in ethical leadership, while other constructs completed by the supervisor were modeled as group-level constructs. To minimize potential estimation issues, we created three randomly distributed parcels for the ethical leadership scale (Landis et al. 2000). We compared our hypothesized four-factor model (Model 1) with two alternate models: Model 2—a three-factor model that combined ethical leadership and moral reflectiveness into one construct; and Model 3—a one-factor model where all items loaded on a single construct.

Lastly, we tested our theoretical model using linear and hierarchical regression analyses. We further specified a multilevel moderated mediation model in Mplus (Preacher et al. 2010), which allowed us to simultaneously test our theoretical model. This approach is preferable to OLS moderated mediation regression analyses (e.g., using the PROCESS macro) because it does not assume independence of observation in the data. Moreover, Preacher et al. (2010) have discussed the advantages of Mplus over traditional multilevel modeling paradigms (e.g., HLM) for testing mediation in a multilevel context. For instance, unlike HLM, Mplus allows one to directly test relationships among “Level 2” variables. We tested for indirect and conditional indirect effects using Bayesian estimation in Mplus. This procedure uses the Markov Chain Monte Carlo (MCMC) process to iteratively estimate the indirect and conditional indirect parameters (Zyphur and Oswald 2015). This process (iterations = 10,000 in the present study) is similar in principle to nonparametric bootstrapping procedures (Tucker et al. 2016; Zyphur and Oswald 2015). Mplus provides 95% confidence intervals around the estimated parameters.

Results

Measurement Model

The results of the multilevel CFA showed that our hypothesized factor structure was a good fit to the data, χ2 (86) = 89.95, p > .05, CFI = .99, TLI = .98, RMSEA = .02, SRMRwithin = .02, and SRMRbetween = .12. All fit indices are within acceptable ranges, with the exception of SRMRbetween, which captures the fit at the group level. We suspect that this is likely due to the relatively small sample size at the group level. For instance, although the model estimation terminated normally, Mplus suggested caution due to the fewer number of groups (relative to the number of parameters estimated). Nonetheless, the hypothesized model was significantly better than the alternate models, including Model 2 (Δχ2 (3) = 11.19, p < .05 [χ2 (89) = 106.87, p > .05, CFI = .94, TLI = .93, RMSEA = .04, SRMRwithin = .01, and SRMRbetween = .13]), and Model 3 (Δχ2 (6) = 112.66, p < .05 [χ2 (92) = 227.67, p < .05, CFI = .54, TLI = .46, RMSEA = .11, SRMRwithin = .15, and SRMRbetween = .31]).

Test of Hypotheses

Table 1 presents the correlations, means, and standard deviations of the variables in this sample. We carried out a series of regression analyses in Mplus to test our hypotheses. The results showed that leader conscientiousness was positively related to moral reflectiveness, B = .39, p < .01, 95% CI [.10, .68], providing support for Hypothesis 1. In support of Hypothesis 2, leader moral reflectiveness was positively associated with employee ratings of ethical leadership, B = .34, p < .05, 95% CI [.06, .63]. Hypothesis 3, which predicted that moral reflectiveness mediates the link between leader conscientiousness and ethical leadership, was not supported, B = .07, p = .09, 95% CI [−.05, .24].

Table 1 Descriptive statistics and correlations among study variables (Sample 1)
Full size table

Next, we tested our proposed moderated mediation model. The results showed that decision-making autonomy did moderate the indirect effect of leader conscientiousness on ethical leadership perceptions through moral reflectiveness, B = .23, p < .01, 95% CI [.07, .40]. This provides support for Hypothesis 4 (Table 2). Tests of simple slopes showed that moral reflectiveness significantly mediated the relationship between leader conscientiousness and subordinate perceptions of ethical leadership when leader decision-making autonomy was high (B = .18, p < .01, 95% CI [.01, .39] but not when it was low (B = −.04, p > .05, 95% CI [−.22, .10]. A graphical depiction of this interaction is shown in Fig. 2. The graph shows that when leader decision-making autonomy is high, the slope of the relationship between leader conscientiousness and ethical leadership (through moral reflectiveness) is positive and strong, and relatively weak when leader decision-making autonomy is low.

Table 2 Results of multilevel moderated mediation analyses using Bayesian estimation procedures (Sample 1)
Full size table
Fig. 2
figure 2

Indirect effects of leader conscientiousness on ethical leadership perceptions (through moral reflectiveness) at different levels of leader decision-making autonomy (Sample 1)

Full size image

Although we used two sources of data collection in our first sample—leader and their direct reports—the sample is relatively small and cross-sectional in nature. Therefore, we sought to test our model with a dyadic dataset of leaders and their direct reports, while separating measurements in time in order to limit common method bias and strengthen the order of occurrence of the tested relationships in using an additional sample (Podsakoff et al. 2012).

Sample 2

In this sample, we addressed two critical limitations of Sample 1. First, we recruited a significantly larger sample of workgroups. This increases our statistical power to detect effects and also addresses the MCFA estimation issues associated with the small number of workgroup in Sample 1. In addition, we employed a time-lagged design in Sample 2 to address the potential effects of common source bias associated with the cross-sectional design of Sample 1.

Participants and Procedure

Sample 2 consisted of 163 workgroups from seven ICT companies located in Beijing (Northern China), Chengdu (Western China), and Guangzhou (Southern China), three major Chinese high-tech metropolitan clusters. We surveyed both work group managers and members of work groups in functional areas such as IT operations, finance, accounting, sales and marketing, and human resources. Prior to the survey administration, we invited human resource personnel to provide a list of group managers as well as employees who work directly under them. A cover letter was used to explain the purpose of the study and emphasize the voluntary nature of participation and the strict confidentiality of their responses. We administered two separate questionnaires to workgroup managers and their direct reports in three phases. Whereas workgroup managers were surveyed at Time 1 and Time 2 (three weeks after Time 1), their direct reports were surveyed at Time 3 (three weeks after workgroup managers completed the time 2 surveys). The questionnaires included a personal ID based on the list received from the HR personnel of each firm in order to enable us link the surveys. With the help of our contacts, surveys were distributed on-site and directly returned to the research team after completion. To thank those who participated across the three different phases, they were entered into a drawing to win one of twenty restaurant vouchers priced at 200 RMB (approximately US$31) each.

Of the 259 workgroup managers who were invited at Time 1 to provide ratings of their conscientiousness and demographic information, we received responses from 182 (a response rate of 70.27%). These individuals were then sent another survey to rate their moral reflectiveness and decision-making autonomy at Time 2, of which only 178 of them responded (overall response rate of 68.73%). At time 3, employees were invited to rate the ethical leadership behavior of their workgroup manager. Of the 1,316 direct report surveys distributed, we received 723 (approximately a 55% response rate). Of the returned surveys, six uncompleted leader surveys were discarded. We also eliminated nine workgroups in which group members did not fully complete the questionnaires. Therefore, our final sample size consists of 163 workgroup managers and 714 group members.

The size of each workgroup ranges from 3 to 23 individuals, with an average group size of 4.38 (SD = 3.84). The average age of group members and managers is 28.57 and 36.51, respectively. The average organizational tenure is around 4.10 years, with an average of 3.23 years with their immediate leader. Both group members and managers work relatively long hours per week (mean = 54.21 and 62.74 h, respectively). Finally, 94% of group members completed at least post-secondary studies and 37.1% of group managers hold postgraduate degrees.

Measures

We used the same measures from Sample 1. All items were measured on a 7-point scale (1 = strongly disagree, 7 = strongly agree) because prior studies report that in Asian cultures, people are more likely to select the midpoints and avoid extreme responses on 5-point Likert scales (Chen et al. 1995; Hamamura et al. 2008). Higher scale granularity can also reduce response bias.

We followed the back-translation procedure (Brislin 1980) to translate the scales from English to Chinese: a professional translator first translated the questionnaires from English to Chinese. Then, a different translator proceeded with the reverse translation (i.e., translated the scales from Chinese back to English). Finally, the first translator compared the two scales to check whether the back-translation scales were similar to the original ones. We also pretested the questionnaires with ten individuals to make sure all items were clearly understood.

The alpha values of leader conscientiousness at Time 1, moral reflectiveness and decision-making autonomy at Time 2, and ethical leadership at Time 3 were .78, .92, .89, and .83, respectively.

Results

Measurement Model

Multilevel CFA showed that our hypothesized factor structure was an excellent fit to the data, χ2 (129) = 138.83, p > .05, CFI = .99, TLI = .99, RMSEA = .01, SRMRwithin = .001, and SRMRbetween = .04. The hypothesized model was significantly better than the alternate models, including Model 2 (Δχ2 (3) = 250.22, p < .01 [χ2 (132) = 299.21, p < .01, CFI = .89, TLI = .87, RMSEA = .04, SRMRwithin = .01, and SRMRbetween = .09]), and Model 3 (Δχ2 (5) = 410.18, p < .01 [χ2 (134) = 513.00, p < .01, CFI = .75, TLI = .71, RMSEA = .06, SRMRwithin = .01, and SRMRbetween = .17]). These results provide strong evidence in support of the proposed measurement model in this sample.

Test of Hypotheses

Table 3 presents the correlations, means, and standard deviations of the variables in this sample. The results showed that leader conscientiousness was positively related to moral reflectiveness, B = .68, p < .01, 95% CI [.46, .90], providing support for Hypothesis 1. In support of Hypothesis 2, leader moral reflectiveness was positively associated with employee ratings of ethical leadership, B = .32, p < .01, 95% CI [.21, .44]. Hypothesis 3, which predicted that moral reflectiveness mediates the link between leader conscientiousness and ethical leadership, was supported in this sample, B = .19, p < .01, 95% CI [.09, .29]. Next, we tested the proposed moderated mediation model. We found once again that leader decision-making autonomy moderated the indirect effect of leader conscientiousness on ethical leadership perceptions through moral reflectiveness, B = .10, p < .01, 95% CI [.02, .18].

Table 3 Descriptive statistics and inter-correlations among study variables (Sample 2)
Full size table

This provides support for Hypothesis 4. Tests of simple slopes showed that moral reflectiveness significantly mediated the relationship between leader conscientiousness and subordinate perceptions of ethical leadership when leader decision-making autonomy was high (B = .30, p < .01, 95% CI [.16, .46], but not when it was low (B = .08, p > .05, 95% CI [−.04, .21]. Similar to Sample 1, the graphical depiction of the relationship (Fig. 3) shows that when leader decision-making autonomy is high, the slope of the relationship between leader conscientiousness and ethical leadership (through moral reflectiveness) is positive and strong, and relatively weak when leader decision-making autonomy is low (Table 4).

Fig. 3
figure 3

Indirect effects of leader conscientiousness on ethical leadership perceptions (through moral reflectiveness) at different levels of leader decision-making autonomy (Sample 2)

Full size image
Table 4 Results of multilevel moderated mediation analyses using Bayesian estimation procedures (Sample 2)
Full size table

Discussion

In the present research, we attempt to reevaluate the relationship between leader conscientiousness and ethical leadership in light of the existing debates in the literature regarding the extent to which personality may predict observed leadership behavior (Judge et al. 2002; Barrick and Mount 2005). In doing so, we delve more deeply into why and when conscientious leaders may influence employees’ perceptions of ethical leadership. Specifically, we drew on SCT (Bandura 1986, 1991) to develop and test an integrated moderated mediation model of how and under what conditions leader conscientiousness may lead to observable demonstrations of ethical leadership behaviors. As predicted, our research findings suggest that leader conscientiousness is positively related to moral reflectiveness, which subsequently engenders ethical leadership as perceived by employees. However, the indirect effect of leader conscientiousness on ethical leadership was conditional on the leader’s decision-making autonomy at work. Decision-making autonomy enhanced the strength of the indirect effect of leader conscientiousness on ethical leadership through moral reflectiveness, such that the relationship was stronger for leaders with high decision-making autonomy rather than those with low decision-making autonomy. These findings suggest a number of meaningful theoretical and practical implications.

Theoretical Implications

Our study contributes to theory and emerging research on ethical leadership antecedents in several ways. First, as previously discussed, although prior research has linked leader conscientiousness with ethical leadership behaviors (Kalshoven et al. 2011; Walumbwa and Schaubroeck 2009), research on ethical leadership has largely focused on examining the consequences of such leadership behavior. Despite promising initial evidence, this stream of research still lacks a strong theoretical framework that addresses the debates in the leadership literature as to when and precisely why conscientiousness may predicts leadership behaviors (Judge et al. 2002). Accordingly, we developed a social cognitive perspective to advance this emerging stream of research on ethical leadership by exploring both distal (leader conscientiousness) and proximal (moral reflectiveness) antecedents and consequently, illustrating how this particular leader personality trait and related cognitive reflection may lead to perceptions of ethical leadership. Our research findings in this regard thus help to demonstrate conscientiousness as a potentially beneficial personality trait for predicting ethical leadership. By applying SCT (Bandura 1986, 1991), we explicate the psychological process that makes conscientious leaders more likely to exhibit ethical leader behaviors by highlighting the critical role that moral reflectiveness plays in this process. Our findings are consistent with the hypothesized model, in that leaders who are higher in conscientiousness are inclined to be more morally reflective, and in turn, demonstrate more leadership behaviors that employees perceive as ethical. As such, we extend SCT by specifically measuring a domain-specific cognitive forethought/reflection and its relationship to morally specific leadership behaviors of great concern to both scholars and practitioners. In sum, drawing on SCT, we highlight that ethical leadership stems from leaders’ reflection on morality in their daily experiences. This finding underscores the need for organizational processes, training activities, and both leader- and follower-initiated conversations that foster closer attention to the moral implications of even seemingly mundane management decisions.

The present research also contributes to the literature by examining leader decision-making autonomy as a situational moderator of ethical leadership antecedents. The inclusion of this important leadership constraint as a moderator in our model was crucial given that the literature on ethical leadership has generally overlooked the role that situational factors may play (Den Hartog 2015). As Den Hartog suggested, models proposing trait antecedents of ethical leadership may be insufficient and may limit insights that could be gained from such research without accounting for the role of a leader’s job context. Accordingly, and in line with the SCT framework (Bandura 1986, 1991), our findings suggest that high decision-making autonomy strengthens the extent to which a leader’s proximal cognitive reflection (i.e., moral reflectiveness) translates into ethical leadership, while low decision-making autonomy constrains transmission of leader moral reflectiveness into ethical leadership. This suggests that future research may also examine other potential situational variables, such as work demands (e.g., time pressure, organizational bottom line mentality), perceived organizational support for ethical leadership, and perceptions of the political risk or consequences of a decision, among others.

More generally, our research also contributes to SCT (Bandura 1986, 1991) by extending and empirically testing the basic tenets of Reynolds’ (2008) moral reflectiveness model. Reynolds provided a framework for understanding how reflecting on morality and moral matters (i.e., moral reflectiveness) can lead to moral behavior. Extending this model, we proposed and empirically confirmed the importance of moral reflectiveness in the domain of workplace ethical leadership. Specifically, moral reflectiveness facilitates ethical leadership consistently in both of our samples, yet only when leaders have high levels of decision-making autonomy. Hence, moral reflectiveness may be a necessary but insufficient precursor of moral behavior. This finding suggests that future research investigating how individual moral reflectiveness relates to moral behaviors may benefit from incorporating situational factors such as decision-making autonomy into their models.

Practical Implications

This research also offers important practical knowledge that can facilitate and guide the development of ethical leadership in organizations. Given the complex linkages among leader conscientiousness, moral reflectiveness, and ethical leadership behavior, our findings suggest that it is not sufficient to only hire conscientious leaders who are cognitively disposed to showing ethical behaviors. Rather, encouraging them to pay attention to moral issues in their experience is also an important consideration. By doing so, such leaders can benefit from their cognitive reflectiveness on morality and moral matters as it can importantly shape observable ethical behavior. As such, practical investigations aimed at developing ethical leadership could devote adequate sessions to getting participants to role-play, discuss, and reflect on moral issues around them.

Our findings also indicate that giving leaders the freedom to make decisions also strengthens the proximal motivational mechanism of a leader’s increased moral reflectiveness. Hence, the degree of decision-making autonomy is an important consideration both for organizations seeking to recruit conscientious leaders and for programs designed to facilitate ethical leadership. Without such consideration, the potential benefit of recruiting and selecting conscientious leaders could be attenuated, as such leaders require relevant job contexts to maximize their potential and further develop and role model their ethical leadership.

Limitations and Directions for Future Research

Despite these contributions, our study is not without limitations. First, we did not receive permission from the organizations to collect data longitudinally in Sample 1, which would have been preferable given the causal direction implied in our model. Hence, because of the cross-sectional nature of our research design, we cannot draw strong causal conclusions. We addressed this limitation in Sample 2 by temporally separating the variables in the causal chain using a time lag of three weeks between each phase. In doing so, our research design also helps to reduce common method bias (Podsakoff et al. 2012). Moreover, where inferred, the relationship directionality of our model was grounded in theory and empirical evidence. Future research may however utilize a longitudinal research design where all variables are measured repeatedly or an experimental design to strengthen causal claims.

Second, although our focus on the mediating role of moral reflectiveness rather than perceptual moral attentiveness in the relationship between leader conscientiousness and ethical leadership was theoretically motivated, it would have been helpful to also include perceptual moral attentiveness in order to test whether (or not) a similar pattern may hold. For instance, compared to moral reflectiveness, perceptual moral attentiveness is less likely to fluctuate (i.e., it is more likely to be static; Kim et al. 2014; Sturm 2015). Moreover, as previous research demonstrates, although perceptual moral attentiveness helps individuals recognize moral dilemmas and recall morality-related behaviors in others, unlike moral reflectiveness it does not necessarily translate into moral behaviors (Reynolds 2008; Sturm 2015). Thus, it seems quite unlikely that perceptual moral attentiveness will be more relevant than moral reflectiveness in our model since ethical leadership focuses more on the behavioral actions of leaders (Brown et al. 2005). Nevertheless, we encourage future research exploring the link between leader conscientiousness and ethical leadership to include perceptual moral reflectiveness in order to further substantiate this claim. Relatedly, our findings regarding the significant relationship between conscientiousness and moral reflectiveness seem to be inconsistent with Reynolds’ (2008) study that found a nonsignificant relationship. This suggests the possibility of a moderating variable, which points to an interesting direction for future research. For instance, trait activation theory (Tett and Burnett 2003) suggests that personality traits require trait-relevant situations for their expressions. Research has shown that low time pressure, for example, is an important cue that conscientious leaders may require to be effective (Ng et al. 2008). As such, it is possible that conscientious leaders also need less time pressure to enable them become moral reflective, thus offering a potential resolution for the nonsignificant finding in Reynolds (2008). We encourage future research to investigate this possibility.

Third, we found somewhat inconsistent support for our mediation hypothesis across the two samples. Specifically, we found that moral reflectiveness mediated the link between leader conscientiousness and ethical leadership in Sample 2, but this mediation effect was marginally significant (p = .09) in Sample 1. We believe that the relatively small group-level sample size (n = 28) in Study 1 may have contributed to this discrepancy. In our model, the proposed mediation effect primarily occurs at the group or “between” level of analysis (although ethical leadership varies both within and between groups, our goal was to predict the between-level variance component of this construct). Previous simulation studies generally indicate that increasing the number of groups, rather than number of individuals in each group, is necessary to ensure sufficient power to detect between-group effects (Kreft 1996). Hofmann (1997) however notes that sample size requirements should be less for detecting between-group main effects, compared to “cross-level” interaction effects. Thus, we believe that our small sample size in Sample 1 is a clear limitation, although it may not be as significant given the simplicity of our mediation model (between-group effects involving three constructs). Indeed, past multilevel studies with similar or more complex models have published results based on similar sample sizes (e.g., Eisenbeiss et al. 2008; Hofmann et al. 2003; Hirst et al. 2009). Nonetheless, we addressed this limitation by recruiting a much larger sample of workgroups in Sample 2 (n = 163). With greater statistical power, the results from Sample 2 provide full support for mediation. Moreover, results from both samples showed that the indirect effect of leader conscientiousness on ethical leadership through leader moral reflectiveness was strongest under conditions of high (versus low) leader decision-making autonomy. The replication of this moderated mediation relationship across both samples is encouraging and serves to further allay our sample size and statistical power concerns.

Finally, although our findings provide support for the generalizability of leader conscientiousness in predicting EL in non-Western cultures (in this case, Nigeria and China), future research should replicate and examine the generalizability of our findings in other cultures. Similarly, such research could also consider cross-cultural variation in EL emergence. Insights from the GLOBE studies suggest that the behavioral enactment of EL may vary by cultural context, yet the importance of EL as a critical aspect of leadership does not (Chhokar et al. 2007; House et al. 2004). In other words, existing evidence suggests that EL is universally desirable, and we would expect that our theoretical arguments and findings are generalizable to other contexts such as Western cultures, especially in organizations where decision-making is hierarchically structured. In such contexts, our findings suggest that the leader’s level of autonomy will be a critical aspect to consider; however, the nature and range of that autonomy will likely differ significantly based on both the organizational and cultural context.

Generally speaking, the research on the extent to which conscientiousness predicts ethical leadership is still developing in the ethical leadership literature and seems promising based on our findings. However, much can still be learned in this area. For instance, our research findings showing that morally reflective leaders are less likely to be perceived as ethical leaders when their decision-making autonomy is low raises an important question for future research. Given that the ethical leadership literature largely relies on subordinates’ perceptions of such leadership behavior, it would be interesting to further examine when leaders fail to behave ethically under conditions of low decision-making autonomy (reflecting on “actual” leader ethical behavior rather than “perceptions” of ethical leadership). For instance, it may be that such situations may constrain the perception of ethical leadership and perhaps not necessarily constrain actual leader ethical behavior. Clearly, exploring this line of research would be an interesting next step. For example, is it that morally reflective leaders do not behave ethically in low decision-making autonomy contexts, or that they are just not “perceived” by employees as demonstrating ethical leadership under such conditions? Although “actual” ethical behaviors demonstrated by leaders are very important, research has shown that individuals are more strongly influenced by their own “perceptions” of others’ behavior than its actual “objective” nature (Lewin, 1951). Regardless, future investigation may refine our outlook on ethical leadership in organizations and further open scholarly discussions regarding the extent to which we need to study “actual” vs. “perceived” ethical leadership. Importantly, such research should also specifically test whether there is a differential effect between actual and perceived ethical leadership.

Conclusion

In light of the long lasting debates in the personality approach to leadership literature, our research offers important contributions to the literature. We shed new light on the mechanisms and boundary conditions that lead conscientious leaders to be perceived as ethical leaders, thus demonstrating that the link between leader conscientiousness and ethical leadership is more complex than assumed by previous research. In line with SCT, we illustrate the importance of leader moral reflectiveness as a mediator, and decision-making autonomy as a moderator, in the relationship between conscientiousness and ethical leadership. Further, our research suggests that, to aid the effectiveness of practical interventions that can potentially support ethical leadership, there is a need for morally reflective leaders to have autonomy in making decisions at work. We hope this research reinforces the need for researchers to devote more attention to the antecedent conditions associated with ethical leadership, ultimately to help avoid the potentially disastrous impacts of its absence for both organizations and employees.