Research concerning transformational leadership (TL; Bass 1985; Bass and Avolio 2000; Bass and Riggio 2006) has shown it to be consistently related to a wide range of valued organizational outcomes at both the individual and group levels (Hoch et al. 2018; Wang et al. 2011). Despite its dominance in the literature, the transformational model with its emphasis on the role of charisma, has been criticized as lacking an ethical dimension (e.g., Sinnicks 2018). Concerns of this nature have grown driven by high-profile scandals in which leader behavior has been explicitly linked to employee suicides (Kostov 2019), and has resulted in the demise or near demise, of entire companies, such as WorldCom (Krouse 2020) and Theranos (Carreyrou 2018). Even when the survival of the company is not in question, multi-billion-dollar fines can be involved, and tremendous reputational damage can be caused, as with the scandal involving fraudulent customer accounts at Wells Fargo (Ensign and Eisen 2020) and emissions test cheating at various automobile manufacturers (Boston 2020). Abuses of power by business leaders also continue, as reflected by the harassment-related resignations of Harvey Weinstein (Paul 2020) and Travis Kalanick, the founder and former CEO of Uber Technologies Inc. (Bensigner and Farrell 2017). Indeed, in North America, business executives are among the most distrusted occupational groups; only lawyers are distrusted to a greater degree (Pew Research Center 2013).

Several alternative models to TL have been advanced that place greater emphasis on ethics as a component of effective leadership. One widely researched example (Bedi et al. 2016), is Brown et al.’s (2005) concept of ethical leadership, assessed by the Ethical Leadership Questionnaire (ELS). Another approach to infusing morality into business leadership involves the study of leader virtuousness (Crossan et al. 2013; Hackett and Wang 2012; Wang and Hackett 2016). For example, the Virtuous Leadership Questionnaire (VLQ; Wang and Hackett 2016) measures courage, temperance, justice, prudence, and humanity, as reflections of overall leader excellence.

As relatively new additions to the field, it is important to ensure that ethical and virtuous leadership are conceptually and empirically distinct from the well-established transformational model. To date, as illustrated by Bedi et al. (2016) among others, the empirical distinctiveness of leadership models has typically been examined using meta-analysis, in which estimates of effect sizes are derived and compared by combining findings across studies. Based on their meta-analytic review, Banks et al. (2018) concluded that the newer moral approaches to leadership are highly correlated with more traditional approaches, adding little in the way of incremental validity. They suggested that future research be more critical of both new and existing theory to ensure that redundant leadership constructs are pruned from the literature. Nonetheless, these meta-analytically derived conclusions might be premature because there tends to be significant variance in the estimates meta-analysis generates due to unidentified differences across studies. Thus, it is helpful to compare alternative leadership models within the same study, but research of this type is rare (DeRue et al. 2011). As further noted by Den Hartog (2015, p. 416), among others, “[m]any studies do not incorporate both ethical leadership scales and scales to measure related leadership styles; however, those that do suggest that correlations with related scales are relatively high and that more work on discriminant and incremental validity of ethical leadership scales is needed”. Relatedly, Bedi et al. (2016) specifically highlighted the need to further examine the relationship between ethical leadership and TL. Accordingly, for the first time, and in the same study, we simultaneously examine ethical leadership (Brown et al. 2005), virtuous leadership (Wang and Hackett 2016) and the major components of the long-standing TL model (see the Multifactor Leadership Questionnaire; MLQ; Bass and Avolio 2000), for the relative strength of their associations with a range of organizationally and personally valued outcomes. We offer a theoretical rationale for expecting the three leadership approaches to predict important work-related and personal outcomes at different levels of magnitude. Such differences would support the unique value of each of the three leadership approaches we examine, responding to recent calls for such comparisons (e.g., Den Hartog 2015). Thus, our research design and the data we present address a need to directly and simultaneously compare, within a single study, three leadership approaches in their prediction of organizationally and personally valued variables.

In choosing the leadership approaches in question, our aim was to compare two of the relatively new leadership approaches that feature an ethical component both to one another, and to TL, without burdening our respondents with an unduly large number of leadership-related items. Specifically, in this investigation involving two samples, we examined the three approaches for their association with outcomes of importance to businesses: i.e., leader effectiveness, follower in-role and extra-role performance, follower ethicality, as well as happiness and life satisfaction among leaders and followers. Importantly, in choosing these outcomes, as noted above, there are cases in which theory suggests that there will be significant differences in the strength of the relationships observed as a function of the leadership model.

Below, we begin by describing the central concepts associated with ethical leadership, virtuous leadership and targeted components of TL, as reflected by socialized charismatic leadership (Bass and Avolio 2000; Galvin et al. 2010; Howell 1988). We then present a brief review of what is already known about their interrelationships. Subsequently we present hypotheses concerning the respective ties of each model to our outcomes of interest.

Ethical Leadership, Virtuous Leadership, and TL: Central Concepts

Ethical Leadership

There are many alternative definitions of ethical leadership, but Brown et al.'s (2005) perspective reflected by the ELS dominates the literature (Bedi et al. 2016; Brown and Treviño 2006; Hoch et al. 2018). For Brown et al. (2005, p. 120), ethical leadership consists of “the demonstration of normatively appropriate conduct through personal actions and interpersonal relationships, and the promotion of such conduct to followers through two-way communication, reinforcement, and decision-making”. The positive influence of ethical leadership is thought to originate, for example, from the communication of ethical standards and expectations, intentional role modeling (see social learning theory Bandura 1976) by the leader, and through the explicit use of rewards and punishments. Indeed, perceptions of leader ethical behavior as assessed by the ELS associate positively with a range of valued organizational outcomes (Bedi et al. 2016; Hoch et al. 2018).

Virtuous Leadership

While there are a variety of alternative perspectives involving the concept of leader virtue (e.g., Crossan et al. 2013; Riggio et al. 2010; Thun and Kelloway 2011), our focus is on Wang and Hackett's (2016) model, because it has strong grounding in virtue ethics, and is reflective of both Confucian and Aristotelian thinking (Hackett and Wang 2012). For Wang and Hackett (2016: 326), virtuous leadership is “a leader–follower relationship wherein a leader’s situational appropriate expression of virtues triggers follower perceptions of leader virtuousness, worthy of emulation”. Here, the interest is on the subset of virtues directly related to morality, i.e., moral virtues that collectively comprise the “good” component of character (Hartman 1998).

As noted earlier, the VLQ (Wang and Hackett 2016), assesses the leader virtues of courage, temperance, justice, prudence, and humanity, which collectively, are intended to reflect excellence in leadership. Specifically, courage enables people to do good in the face of danger or at the risk of negative consequences, enabling worthwhile long-term risk-taking even if one’s personal reputation suffers. Temperance enables the control of desires for instant gratification even when there are no externally imposed restraints. Justice enables leaders to be respectful and fair in the treatment of others while allocating valued resources, even when conflicting interests are involved. Prudence enables leaders to make a right assessment and/or decision even in the face of the needs of a plurality of stakeholders as in complex business environments, so that resources are used in an efficient, effective manner. Finally, humanity enables leaders to show care and concern for the needs and interests of others across contexts. In practice, these virtues are intimately related in that they are often demonstrated simultaneously (Wang and Hackett 2016). For example, acting in a just manner often requires courage, and prudence frequently demands temperance.

Positive outcomes are anticipated when a leader is seen by followers as behaving in a virtuous (excellent) manner. As with ethical leadership (Brown et al. 2005), Wang and Hackett (2016) argue that social learning theory (Bandura 1976) underlies the practice of virtuous leadership, as it becomes a source of influence when followers begin to experience the intrinsic rewards associated with engaging in virtuous behavior. These rewards are grounded in internalization, wherein followers accept influence because the modeled behaviors are (or gradually become) congruent with followers’ values system. This value congruence is likely to be strengthened by co-workers, since virtuous behavior is widely accepted in both Western and Eastern traditions as reflective of desirable personal qualities (Sison 2003), and is “good” for everyone (MacIntyre 1984). In line with this theory, VLQ-based ratings of leader virtue provided by followers positively predict, for example, leaders' ratings of follower in-role and extra-role performance, followers' views of leader ethicality, as well as self-reported happiness and life satisfaction of both leaders and followers (Wang and Hackett 2016).

TL

In its current form (e.g., Bass and Avolio 2000; Bass and Riggio 2006), transformational leaders are charismatic; they articulate a vision (inspirational motivation) for the work group, model how to realize it (idealized influence), urge followers to challenge the status quo (intellectual stimulation), and offer personal support (individualized consideration) to subordinates. Followers in response, for example, tend to adopt the values and interests of their employer, reflective of a cohering of self- and organizational-identities (Kark et al. 2003) that fosters a wide range of positive outcomes, including performance at the individual, team, and unit levels (Hoch et al. 2018; Wang et al. 2011).

As noted by others (e.g., Hoch et al. 2018), the original conceptualization of TL (see Bass 1985) did not especially emphasize the importance of the ethical implications of leader behaviors. Nonetheless, these considerations have received greater attention over time such that it has been argued that certain dimensions of TL should be related to ethical leadership (Bedi et al. 2016). For example, the idealized influence and inspirational motivation components together reflect socialized charismatic leadership (Bass and Avolio 2000; Galvin et al. 2010; Howell 1988), which makes specific reference to the ethical consequences of decisions. As noted from the outset, our primary interest is in the extent to which the more recent leadership models—that emphasize leader ethicality and leader virtue – are conceptually and empirically distinct from key components of TL as reflected, for example, by socialized charismatic leadership. As explained earlier, some have used meta-analytic findings based upon the combining of disparate studies to conclude that the newer moral approaches offer little in the way of incremental validity (e.g., Banks et al. 2018).

Ethical Leadership, Virtuous Leadership, and TL: Construct Distinctiveness

To effectively address concerns of construct redundancy, evidence of both conceptual and empirical distinctiveness is required. Below, we briefly address the current state of research involving our three targeted leadership models.

Conceptual Distinctiveness

As reflected in the previous section, conceptual differences across the three models are apparent in that, for example, the TL components are dominated by the importance of a compelling workplace vision, and do not address leader ethicality and morality to the same extent as ethical and virtuous leadership. There are also important conceptual differences between ethical leadership (Brown et al. 2005) and Wang and Hackett's (2016) virtuous leadership. For example, Brown et al. (2005) targeted ethical leadership per se, while Wang and Hackett (2016) addressed the broader concept of leader excellence. As such, the construct of virtuous leadership has content beyond ethics, to, for example, encompass good business judgment. Moreover, although both emphasize the importance of the leader as a role model, Brown et al. (2005) addressed obligations and consequences reflected in using rewards and punishments to foster normative ethical behavior; whereas Wang and Hackett (2016) argued that the virtuous leader engages in virtuous leadership because it is virtuous, and followers are seen as adopting virtuous behaviors on their own for intrinsic reasons, especially as it becomes apparent that virtuousness is socially desirable.

Empirical Distinctiveness

Regarding empirical distinctiveness among our three targeted models, a large data base has enabled important comparisons of ethical leadership and TL (Bedi et al. 2016; Hoch et al. 2018). For example, using a combination of meta-analysis and hierarchical regression, Hoch et al. (2018) showed that ethical leadership typically accounted for variance in a variety of targeted outcomes beyond TL alone, but the magnitude of increase varied depending on the variables involved. For example, ethical leadership accounted for only a one percent increase in the variance of subordinates’ job performance relative to TL alone, whereas the gain was more substantial for employee deviance (13% of increase), job satisfaction (8% increase), and affective commitment (6% of increase).

In comparison to the ethical versus TL pairing above, empirical comparisons of ELS-based ethical leadership (Brown et al. 2005) and VLQ-sourced virtuous leadership (Wang and Hackett 2016) are lacking. This is an important gap given that both models were formulated, in part, to address the relative lack of attention to ethics in the leadership literature. As for the empirical distinctiveness of virtuous leadership relative to TL, Wang and Hackett (2016) showed that VLQ-sourced virtuous leadership accounted for significant variance in leader effectiveness and leader ethicality, among other outcomes, beyond that of idealized influence and inspirational motivation (Bass and Avolio 2000).

Three Leadership Models: Possible Differential Associations with Valued Outcomes

Having briefly reviewed the research concerning the conceptual and empirical distinctiveness among ethical leadership (Brown et al. 2005), virtuous leadership (Wang and Hackett 2016) and TL (Bass 1985; Bass and Avolio 2000; Bass and Riggio 2006), we now seek to advance the literature by proposing and empirically evaluating a series of hypotheses concerning the possible differential ties of these three leadership models to a series of business-related outcomes: leader effectiveness, follower in-role and extra-role performance, follower ethicality, as well as happiness and life satisfaction among leaders and followers.

Leader Effectiveness

Brown et al. (2005), Wang et al. (2011), and Wang and Hackett (2016), among others, offered detailed theoretical discussions to justify the positive associations observed between ethical leadership, virtuous leadership, and TL respectively, and subordinates’ perceptions of leader effectiveness. Briefly, for Wang and Hackett (2016) a positive association was anticipated because virtuous persons are expected to strive for excellence in life generally, including job performance. Virtuous leaders are also expected to be effective due to the referent-based influence they gain through the prudent, judicious, and humane use of power. In comparison, ELS-based perceptions of leader effectiveness are rooted in the fairness, openness, honesty, integrity and consideration displayed by ethical leaders (Brown et al. 2005), which are thought to build legitimacy, and foster followers' compliance, trust, and support. Finally, the association between TL and leader effectiveness is based, in part, on the ability of the leader to use inspirational motivation to articulate an exciting, challenging vision for subordinates, and in turn employ idealized influence to model its realization. In all, the mechanisms thought to underlie leader effectiveness are quite different in each case. Accordingly, we expect all three perspectives to offer significant, unique prediction of leader effectiveness.

Hypothesis 1 (H1)

Follower ratings of virtuous leadership, ethical leadership, and the inspirational motivation and idealized influence components of TL, will each be significantly and positively associated with follower judgments of leader effectiveness.

Follower In-Role and Extra-Role Performance

It is noteworthy that follower performance was not a matter of primary interest in the initial development of the ELS (cf. Brown et al. 2005). Nonetheless, positive associations between ethical leadership and both in-role and extra-role performance can be anticipated due to social exchange-based follower obligations in response to fair and caring treatment from the leader (Brown and Treviño 2006; Ruiz-Palomino et al. 2011). Virtuous leaders, as models of excellence in general (Wang and Hackett 2016) should also foster both in-role and extra-role performance among followers, who for intrinsic reasons, will want to perform their jobs well. Virtuous leaders also convey concerns for follower well-being and the community, thus building trust, another positive for both aspects of performance (Dirks and Ferrin 2002). As for TL, there are several reasons for its positive links to follower in-role performance. One is that the compelling vision associated with inspirational motivation should enhance both meaningfulness at work and intrinsic motivation (Wang et al. 2011). Also, a leader's idealized influence should strengthen followers' beliefs that they can achieve challenging goals. Inspirational motivation and idealized influence should enhance follower extra-role performance as well, because they emphasize the needs of the collective over those of individuals, thus urging followers to help their colleagues in the quest to meet organizational goals. Finally, transformational leaders also serve as role models of good performance (Wang et al. 2011).

In all, the discussion above suggests that each of our three approaches to leadership offers somewhat different reasons for anticipating links to follower performance. As such (analogous to leader effectiveness as covered earlier), each should be significantly and uniquely related to both follower in-role and extra-role performance. Nonetheless, in reviewing the differences in both theory and item content, of the three approaches, the transformational components represent the most direct attempts to influence employee performance. Specifically, inspirational motivation and idealized influence explicitly encourage employees to focus on the current and future goals of the organization, through both in-role and extra-role efforts (Wang et al. 2011). In comparison, as described earlier, employee performance per se is not a focus of ELS-based ethical leadership (Brown et al. 2005). Instead, the association to performance is indirect, grounded in employees’ reciprocation for fair treatment (Brown and Treviño 2006; Ruiz-Palomino et al. 2011). Similarly, regarding VLQ-based (Wang and Hackett 2016) virtuous leadership, few (if any) of the behaviors involved are direct attempts to enhance performance. Rather, improvements in follower performance are expected because they are intrinsically gratifying (Wang and Hackett 2016). In all, the transformational components make the most direct reference to the importance of accomplishing the mission, vision and goals of the organization. Accordingly,

Hypothesis 2a (H2a)

Follower-rated virtuous leadership, ethical leadership, and the inspirational motivation and idealized influence components of TL, will each be significantly and positively associated with leader judgments of followers’ in-role and extra-role performance.

Hypothesis 2b (H2b)

The inspirational motivation and idealized influence components of TL will have a significantly stronger relationship with followers’ in-role and extra-role performance, relative to ethical leadership, and virtuous leadership.

Follower Ethicality

As explained earlier, the ELS (Brown et al. 2005) was especially intended to highlight the importance of ethicality in the context of one’s leadership responsibilities. Thus, for example, the practice of ethical leadership is thought to promote analogous follower behaviors through both leader role modeling, and the explicit use of targeted rewards and punishments. The leader as a role model is an important aspect of the anticipated link between VLQ-based virtuous leadership and follower ethicality as well; subordinates observe and imitate their leader, as virtuousness becomes both intrinsically self-reinforcing, and supported by co-workers (Wang and Hackett 2016). Nonetheless, the VLQ does not make specific reference to either ethical standards or to the use of discipline for violations. Thus, increases in follower ethicality are thought to occur primarily for intrinsic reasons. Moreover, as explained earlier, unlike the ELS, leader ethics is only one aspect of overall leader excellence targeted by the VLQ (Wang and Hackett 2016). Regarding TL, as noted at the outset, the early conceptualizations especially, lacked a strong moral component. Importantly, contemporary TL does include leaders’ consideration of the moral and ethical consequences of decisions as an aspect of idealized influence (Bass and Avolio 2000). As such, socialized charismatic leaders model a concern for ethical issues. Nonetheless, in comparison to the ELS, the predominant focus of TL is on the enthusiastic pursuit of a compelling and shared mission and vision; matters concerning ethical standards and discipline for violations are not addressed. In all:

Hypothesis 3a (H3a)

Follower-rated ethical leadership, virtuous leadership, and the inspirational motivation and idealized influence components of TL will each be significantly and positively associated with leader judgments of followers’ ethicality.

Hypothesis 3b (H3b)

Of the three leadership models, ethical leadership reflected by the ELS, will have the strongest positive association with leader judgments of followers’ ethicality.

Happiness and Life Satisfaction Among Leaders and Followers

As was the case with follower performance, outcomes related to happiness and life satisfaction were not a primary focus in the initial development of the ELS (cf. Brown et al. 2005). Nonetheless, there have been recent calls for leadership researchers to treat employee well-being as an intrinsically important criterion (e.g., Inceoglu et al. 2018; Tziner et al. 2014). As aspects of the multi-dimensional construct of employee well-being (cf. Inceoglu et al. 2018), happiness is of interest as a non-cognitive appraisal of life circumstances, reflective of moods and emotions, whereas life satisfaction is primarily a cognitive evaluation of the totality of ones’ life experiences (Hackett and Wang 2012). Thus, for example, ethical leaders could contribute to subordinates’ life satisfaction because they are perceived as being supportive of employees, creating an atmosphere of trust and two-way communication that fosters an overall positive work experience. Consistent with this view, Avey et al. (2012) found the ELS to benefit subordinates’ psychological well-being, mediated by employee voice. Relatedly, Yang (2014) found ethical leadership (assessed largely by the ELS) to associate positively with both well-being and life satisfaction among followers, as mediated through job satisfaction. Indeed, the link between the ELS and job satisfaction is well supported (Bedi et al. 2016; Hoch et al. 2018), which in turn has the potential to positively impact employee life satisfaction (Judge and Watanabe 1993; Rode 2004). Although the literature concerning leadership and well-being is dominated by a focus on followers’ outcomes (cf. Inceoglu et al. 2018), it is also of interest to consider the impact that the practice of ethical leadership has on leaders themselves. It may be, for example, that in both living up to and enforcing normative standards of the workplace, leaders also experience enhanced happiness and life satisfaction. Such increases could occur for intrinsic reasons associated with a knowledge of a job well-done. Moreover, ethical leaders should benefit from being part of the positive supportive work environment they help create.

Although outcomes related to happiness and life satisfaction do not immediately come to mind as priorities associated with TL, several studies link it to various aspects of subordinates’ well- being (Inceoglu et al. 2018). Thus, it may be that TL enhances happiness and life satisfaction for reasons comparable to those involving ethical leadership. Specifically, as with ethical leadership, TL is positively related to job satisfaction (Braun et al. 2013; Hoch et al. 2018), at least partly because inspirational motivation and idealized influence can be tailored to the specific goals and interests of individual followers (Chun et al. 2009), which may in turn, ultimately enhance their life satisfaction (Judge and Watanabe 1993; Rode 2004). Consistent with this view, TL associates positively with followers’ psychological well-being, mediated by the experience of meaningful work (Arnold et al. 2007; Nielsen et al. 2008), increases in self-efficacy (Nielsen et al. 2009), as well as trust in the leader (Kelloway et al. 2012). Finally, to a greater degree than other leadership perspectives, those practicing TL may consciously use displays of emotion (e.g., optimism) to positively impact the well-being of their followers (Ashkanasy and Tse 2000).

In a manner analogous to followers, it may be that leaders themselves also experience enhanced happiness and life satisfaction tied to their work role(s) of formulating and supporting a compelling, meaningful vision. Moreover, practicing TL is likely to result in strong ties between ones’ self-identity and organization identity (Shamir et al. 1993), such that happiness in one realm could significantly impact the other.

Relative to ethical leadership and TL it is notable that happiness and life satisfaction among both followers and leaders is of direct relevance to virtuous leadership. This is because, as explained earlier, virtue refers to the broad concept of excellence. Behaving virtuously is thought to be intrinsically motivating, and a contributor to a life of happiness (Hackett and Wang 2012; Wang and Hackett 2016). Specifically, virtue is likely to contribute to both happiness and life satisfaction by helping to fulfill personal needs associated with self-understanding, and the development of moral capacities, useful also in the successful pursuit of extrinsic goals (Hackett and Wang 2012). Thus, as a reflection of overall leader excellence, the VLQ has the content that goes beyond normatively appropriate conduct, which is a primary focus for Brown et al. (2005, p. 120), and the vision-dominated components of TL. In all, the practice of courage, temperance, justice, prudence, and humanity, as a part of a virtuous life, is anticipated to produce the greatest happiness (Flynn 2008; Wang and Hackett 2016). This should apply both to followers as they are exposed to the leader as a role model, and to the leaders themselves, as they experience the intrinsic rewards of behaving virtuously. Accordingly:

Hypothesis 4a (H4a)

Follower-rated virtuous leadership will be more strongly related to self-rated leader happiness and self-rated leader life satisfaction, than either ethical leadership, or the inspirational motivation and idealized influence components of TL.

Hypothesis 4b (H4b)

Follower-rated virtuous leadership will be more strongly related to self-rated follower happiness and self-rated follower life satisfaction, than either ethical leadership, or the inspirational motivation and idealized influence components of TL.

Research Design

A two-sample investigation was undertaken to compare the associations among the three leadership approaches (i.e., ethical leadership, virtuous leadership, and socialized charismatic leadership) with several leader and follower outcomes (i.e., leader effectiveness, follower in-role and extra-role performance, follower ethicality, as well as happiness and life satisfaction among leaders and followers). Our within-study comparisons directly assess the extent to which there is undue overlap among some of the constructs in the leadership literature (Bedi et al. 2016; Den Hartog 2015). As detailed below, Sample 1, which reflects the Study 2 data from Wang and Hackett (2016), consisted of leaders and followers from a variety of industries in the US, Canada, and the UK. Sample 2, which allowed us to examine the extent to which our initial findings replicated, was comprised of leaders and followers working across a range of industries in the US.

Method

Sample and Procedure

Sample 1

Data were collected using two surveys, both of which were made available on the FluidSurveys website. The first survey was targeted to individuals in paid supervisory or managerial roles while the second was designed for their direct subordinates. A total of 381 leader–follower pairs were recruited with the assistance of The Study Response Center for Online Research (Syracuse University). The surveys were completed by 286 supervisors/managers and 300 subordinates, resulting in 230 dyads, a 60% response rate. The 230 dyads consisted of 131 male and 99 female supervisors/managers (57% and 43%, respectively) and 129 male and 101 female subordinates (56% and 44%, respectively). Of the 230 leader–follower dyads, 95% of supervisors/managers and 94% of subordinates lived in the US while the rest lived in Canada or the UK. Of the 230 supervisors/managers, 50% were in the 31–40 age range, 71% held graduate degrees, 57% led fewer than 20 subordinates, and 64% had a working relationship of less than 15 years with their direct subordinates. Of the 230 subordinates, 52% were in the 31–40 age range, 58% held graduate degrees, 49% worked in business services, 31% worked in manufacturing, 17% in public administration, and 3% in the mining, oil, and gas industries.

Sample 2

The same two surveys used in Sample 1 were administered to individuals in supervisory or managerial roles and their subordinates as recruited using an online panel conducted by Qualtrics Inc.. They recruited 131 leader–follower dyads, subject to the restriction that: (a) the relationship between the two parties had to be in place for at least three months; and (b) that the average workweek was 30 hours or more. The dyads were comprised of 51 male and 80 female supervisors/managers (39% and 61%, respectively) and 61 male and 70 female subordinates (47% and 53%, respectively). Of the 131 supervisors/managers, 39% ranged in age from 31 to 40; while 26% were in each of the 41–50 and 50+ age ranges; 41% held graduate degrees, and 49% held undergraduate degrees. Years of leadership experience varied, i.e., less than 1 year, 9%; 1–5 years, 44%; 5–10 years, 21%; 10–20 years, 15%; and 20+ years, 11%. Of the 131 subordinates, 35% were age 31–40, 21% were 41–50, and 24% were 50+. Of the subordinates, 40% held graduate degrees while 50% held undergraduate degrees. Of the 131 dyads, 27% worked in business services, 24% worked in manufacturing and/or manual labor, 23% in public administration, 10% in retail and/or food services, 9% in healthcare, 5% in information technology, and 2% in education.

Measures

Each of the scales below used a five-point Likert-type scale (1 = Never; 5 = Always) response format.

Virtuous Leadership

Followers evaluated leader virtuousness using the 18-item VLQ (Wang and Hackett 2016; α = 0.96). An overall score was formed by averaging the scores associated with the 18 items. e.g., “My supervisor acts with sustained initiative, even in the face of incurring personal risk”; “My supervisor exercises sound reasoning in deciding on the optimal courses of action.”

Ethical Leadership

Followers completed the 10-item ELS (Brown et al. 2005; α = 0.90 to 0.94); e.g., “Conducts his/her work life in an ethical manner”; “Discusses success not just by results but also by the way that they are obtained.”.

Idealized Influence and Inspirational Motivation

Followers completed four idealized influence and four inspirational motivation items from the MLQ short-form (Bass and Avolio 2000; α = 0.92); e.g., “My supervisor talks about his/her most important values and beliefs,” and “My supervisor specifies the importance of having a strong sense of purpose.” Together, these dimensions reflect socialized charismatic leadership (e.g., Galvin et al. 2010; Howell 1988).

Leader Effectiveness

Followers evaluated leader effectiveness using four MLQ items (Bass and Avolio 2000; α = 0.87 to 0.93); e.g., “Is effective in meeting others’ job-related needs”; “Is effective in representing their group to higher authority.”.

Follower In-Role Performance

Leaders assessed in-role performance of their subordinates by completing four items adapted from the Lynch et al.’s (1999, α = 0.91) scale; e.g., “Fulfills responsibilities specified in job description”; “Performs tasks that are expected of him/her.”.

Follower Extra-Role Performance

Leaders assessed extra-role performance of their subordinates using six items adapted from the Lynch et al.’s (1999; α = 0.88) scale, e.g., “Helps co-workers who have been absent”; “Helps co-workers who have heavy workloads”.

Follower Ethicality

Leaders rated the ethicality of their direct subordinates using six items from Singer's (2000; α = 0.84–0.91) scale, e.g., “is honest”; “Can be trusted”.

Happiness

Both leaders and followers completed the five-item self-assessment adapted from the Personal State Questionnaire (Brebner et al. 1995; α = 0.93); e.g., “I always laugh these days”; “Things always work out the way I want them to”.

Life Satisfaction

Both leaders and followers completed the five-item self-assessment adapted from the Satisfaction with Life Scale (Emmons and Diener 1985; α = 0.86 to 0.90); e.g., “In most ways my life is close to my ideal”; “The conditions of my life are excellent”.

Analysis

As a first step in evaluating the data, composite reliabilities (Bearden and Netemeyer 1999) and average variance extracted (AVE) (Chin 1998) were examined for all constructs targeted in our samples.

For hypothesis testing, as in Brown et al. (2005, p. 129), structural equation modeling (SEM) was applied to both compare the magnitude of the relationships of the three leadership measures to the various outcomes of interest, and to assess the incremental prediction associated with each of the leadership models in relation to the outcomes. First, using AMOS, a model relating the three leadership approaches to the targeted outcomes was estimated using bootstrapping with 1000 iterations and 95% bias corrected confidence intervals, so that the standardized parameter estimates associated with each leadership-outcome could be compared. Then, the incremental validity of each leadership approach was tested by constraining the path of a given leadership approach to equal zero, so that the level of degradation associated with the loss of that approach, in relation to each of the outcomes of interest, could be directly assessed.

Results

Descriptive statistics for all study variables for both Samples 1 and 2 are shown in Table 1.

Table 1 Means, standard deviations, and correlations among study variables
Full size table

Construct Reliability and Validity

Table 2 shows the items associated with each of the targeted constructs in Samples 1 and 2. It is notable that in both samples, all of the items loaded significantly on their respective factors as intended; moreover, the reliability of each of the constructs was supported by composite reliabilities among the independent variables that were all above 0.90; the dependent variables were all 0.82 and higher (Bearden and Netemeyer 1999). Moreover, across samples, AVE either exceeded or was very close to (i.e., Sample 2, ethical leadership = 0.49; virtuous leadership = 0.47) the recommended threshold of 0.50 (Chin 1998; Hair et al. 2017).

Table 2 Measurement model: Item loadings, composite reliabilities, average variance extracted
Full size table

Discriminant validity was examined by comparing the square-root of AVE for each variable (see Table 2 for the AVE values) against the average of the intercorrelations (see Table 1) involving the target variable. In line with the criterion recommended by Fornell and Larcker (1981) and Hair et al. (2017) among others, evidence of discriminant validity was obtained for each variable; the square-root of AVE well exceeded the mean of the intercorrelations in both samples.

Three Leadership Models: Comparison of Standardized Parameter Estimates

In relation to hypothesis testing, Tables 3 and 4 (for Samples 1 and 2, respectively) compare standardized parameter estimates and associated bias corrected 95% confidence intervals (based on 1000 iterations) for each of the three leadership approaches in relation to each of the targeted outcomes.

Table 3 Results of the structural model: Sample 1
Full size table
Table 4 Results of the structural model: Sample 2
Full size table

H1, that follower-rated virtuous leadership, ethical leadership, and the inspirational motivation and idealized influence components of TL each associate positively with follower-rated leader effectiveness was not supported. The only parameter estimate that was significant across samples was associated with ethical leadership (Sample 1: 0.814; p < 0.002; Sample 2: 0.795; p < 0.001), though, for Sample 2 only, virtuous leadership was also significant (0.210; p < 0.027). The TL components were not related to leader effectiveness in either sample (Sample 1: 0.039; ns; Sample 2: − 0.073; ns).

H2a, that follower-rated virtuous leadership, ethical leadership, and the inspirational motivation and idealized influence components of TL will associate significantly and positively with follower in-role and extra-role performance as judged by leaders was not supported. In the context of all three leadership perspectives, ethical leadership was the only significant predictor of follower in-role performance across samples (Sample 1: 0.508; p < 0.002; Sample 2: 0.377; p < 0.014). Virtuous leadership (Sample 1: 0.068; ns; Sample 2: 0.081; ns) and the TL components (Sample 1: − 0.024; ns; Sample 2: 0.036; ns) were unrelated to follower in-role performance across both samples. H2a was also unsupported for follower extra-role performance, as ethical leadership again, was the only significant predictor across samples (Sample 1, 0.369; p < 0.045; Sample 2, 0.419; p < 0.011), though virtuous leadership was a significant predictor in Sample 1 (0.257; p < 0.043) as were the TL components in Sample 2 (0.172; p < 0.047).

H2b, that the TL components would be the single strongest predictor of both followers’ in-role and extra-role performance, was not supported. As reflected in the findings presented above, the TL components were unrelated to in-role performance in either sample. Also, although the TL elements related to extra-role performance in Sample 2, the strength of the association was weaker in magnitude relative to the influence of ethical leadership (i.e., 0.172 vs. 0.419).

H3a, that all three forms of follower-rated leadership would associate positively with leader-rated follower ethicality was not supported. Specifically, only ethical leadership related positively to follower ethicality across both samples (Sample 1: 0.469; p < 0.002; Sample 2: 0.394; p < 0.016); virtuous leadership (Sample 1: 0.118; ns; Sample 2: 0.089; ns) and the TL components did not (Sample 1: 0.057; ns; Sample 2: 0.084; ns). H3b was supported; of the three leadership models, the ELS had the strongest positive association with leader-rated follower ethicality.

H4a, that follower-rated virtuous leadership will relate more strongly to leader self-reported happiness and life satisfaction than the other two leadership approaches received partial support in Sample 1, where it was the only significant predictor of leader happiness (0.38; p < 0.005), but this finding did not hold in Sample 2 where the TL components were the only significant predictor (0.273; p < 0.019). With regard to leader life satisfaction, there were trends supporting the role of both virtuous leadership (0.245; p < 0.063) and the TL components (0.223; p < 0.065) in Sample 1, with the TL components only achieving significance in Sample 2 (0.253; p < 0.022).

H4b, that follower-rated virtuous leadership would be more strongly related to both follower self-reported happiness and life satisfaction relative to the other two leadership models received partial support in Sample 1. Specifically, virtuous leadership related to both follower happiness (0.347; p < 0.005) and life satisfaction (0.332; p < 0.003), whereas the TL components did not (0.128; ns, and 0.162; ns, respectively). Nonetheless, contrary to H4b, ethical leadership was similarly related to these two outcomes (0.335; p < 0.002 for follower happiness and, 0.207; p < 0.049 for life satisfaction). Also contrary to H4b, in Sample 2, only the TL components associated with follower happiness (0.319; p < 0.003) and life satisfaction (0.303; p < 0.020); ethical leadership (− 0.005; ns, and 0.220; ns, respectively) and virtuous leadership (0.220; ns, and 0.061; ns, respectively) did not.

Three Leadership Models: Relative/Incremental Validity

In each of the hypotheses tested above, our interest was in comparing the magnitude of the relationships observed in a context in which all three leadership approaches are simultaneously applied. As a follow-up, as in Brown et al. (2005, p. 129), we also used SEM to address a related issue, the incremental predictive power of a given leadership model in relation to each outcome. Relative to our initial analysis, we now address the extent to which a given leadership model is a unique predictor of the targeted outcome, relative to the remaining two approaches.

The first model examined for each outcome was one in which the path from ethical leadership to that outcome was fixed to zero, in comparison to the original corresponding model in which all the paths were freely estimated (see Tables 5 and 6 for Samples 1 and 2, respectively). Across both samples, dropping the path from ethical leadership resulted in a significant increase in chi-square, reflecting degradation of the model fit for leader effectiveness, follower in-role performance, follower extra-role performance, and follower ethicality. In Sample 1 only, dropping ethical leadership also degraded model fit for follower happiness and follower life satisfaction.

Table 5 Incremental validity analysis: Sample 1
Full size table
Table 6 Incremental validity analysis: Sample 2
Full size table

Next, a model for each of the outcomes was examined in which the path from virtuous leadership to the outcome of interest was fixed to zero, in comparison to the original model in which all the paths were freely estimated. As shown in Table 5, for Sample 1, dropping virtuous leadership resulted in a significant degradation of fit for the models involving follower extra-role performance, follower happiness, follower life satisfaction, and leader happiness. These findings did not hold in Sample 2, wherein dropping virtuous leadership did not degrade the fit for any of the eight outcomes (see Table 6).

Finally, we also examined a model for each of the outcomes in which the path from the TL components was fixed at zero, in comparison to the original model where none of the paths was constrained. In Sample 1, dropping the TL components did not degrade model fit for any of the outcomes (see Table 5). In comparison, dropping the TL components in Sample 2 resulted in a significant degradation of the models associated with the four affective outcomes, i.e., happiness and life satisfaction among both followers and leaders (see Table 6).

Discussion

The clearest, most robust finding to emerge from our research was that, among the three leadership approaches examined, ethical leadership, as rated by subordinates, was the only significant, unique predictor of: (a) leader effectiveness (as rated by subordinates); (b) follower in-role performance (as rated by supervisors); and (c) followers’ ethicality (as rated by leaders). This was true across two independently collected samples. Overall, our findings support the idea that ethical leadership, virtuous leadership, and socialized charismatic leadership differ in their prediction of a range of personal- and organizationally-valued outcomes. For example, in Sample 2 only, follower-rated virtuous leadership (VLQ, Wang and Hackett 2016) also (i.e., along with ethical leadership) positively associated with leader effectiveness, but this finding did not hold up in the incremental validity analysis, wherein (in both samples) the only significant drop in variance accounted for occurred when ethical leadership was removed from the overall model. Virtuous leadership (VLQ, Wang and Hackett 2016), and the TL-based (Bass and Avolio 2000) follower judgements of inspirational motivation and idealized influence, positively related to extra-role performance, but only in Sample 1 and Sample 2, respectively; it is notable that the Sample 2 TL finding did not hold up in the incremental validity analysis, wherein the only significant drop in variance accounted for occurred when ethical leadership was removed from the overall model. As a group these findings are consistent with the Bedi et al. (2016) suggestion that the impact of ethical leadership is broader in scope than the formal definition (cf., Brown et al. 2005) would imply.

The only substantial variation across samples involved the leader/follower happiness/life satisfaction findings. For example, consistent with H4a, the VLQ was the single best (and only) predictor of leader’s self-reported happiness in Sample 1, but this did not hold in Sample 2 wherein the TL components were the only significant predictors of leader’s self-reported happiness and life satisfaction. Analogously, Sample 1 was also partially supportive of H4b in that virtuous leadership was significantly related to both followers’ self-reported happiness and life satisfaction (as was ethical leadership), but these findings did not hold in Sample 2 where the TL components were the only significant predictors of these follower outcomes.

In summary, ethical leadership was the strongest and often only predictor of the behavioral outcomes (i.e., leader effectiveness, follower in-role performance, follower extra-role performance, and follower ethicality) whereas the more affective laden outcomes (leader happiness, follower life satisfaction) were tied more to virtuous leadership in Sample 1, and to TL only in Sample 2.

Leadership Effectiveness

In the context of ethical leadership (Brown et al. 2005), virtuous leadership (Wang and Hackett 2016), and the idealized influence and inspirational motivation components of TL, (Bass and Avolio 2000) only ethical leadership was significantly and uniquely related to leadership effectiveness. This is surprising given compelling theory suggesting that all three models should have unique ties to effectiveness. As explained earlier, while ethical leadership was expected to build legitimacy to help foster follower compliance and support, it was also anticipated that virtuous leaders would be seen as striving to attain overall excellence, and that TL would provide an inspirational, engaging vision. Still, ethical leadership dominated the prediction of perceived leader effectiveness.

In expecting all three models to be significantly related to leader effectiveness, we assumed that followers would be motivated and able to use the full range of associated scale content to make meaningful behavioral distinctions among leaders. In retrospect, the extent to which this was likely may have varied across the questionnaires. This is because, as argued by Hansbrough et al. (2015), among others, for behaviors to be the basis for differentiating among leaders, followers must have both encoded them into memory and successfully retrieved them for the rating task. Importantly, the application of dual-processing models of person perception (e.g., Fiske and Taylor 2013) to study leadership suggests that many factors work against this. For example, implicit leadership theory (ILT; Shondrick et al. 2010) suggests that followers’ leadership judgments are often anchored in relatively automated, quickly formed categorical impressions of general traits, including sensitivity, intelligence, dedication, and dynamism (Epitropaki and Martin 2004). Indeed, the extent to which leaders are perceived to match followers’ ideal standing on these traits relates positively to leader-member exchange quality (Epitropaki and Martin 2005). For our purposes, the crucial point is that controlled purposeful processing—beyond quickly formed impressions—is likely to occur only if the follower is especially motivated (Epitropaki and Martin 2005), and if the required cognitive resources are available (Lord 1985; Lord and Maher 1990). Indeed, when raters are poorly motivated and/or if cognitive resources are lacking, they are vulnerable to filling in memory gaps with leader actions aligning with their overall impression, but which did not occur (Hansbrough et al. 2015).

Given the realities of the leader effectiveness rating process described above, it is necessary to consider the potential impact of differences in the cognitive demands likely required to carefully respond to each of the leadership scales we used. For example, in accounting for the dominance of the ELS (Brown et al. 2005) in relation to follower perceptions of leader effectiveness, it is notable that its content is more focused and proximal in comparison to both the VLQ (Wang and Hackett 2016) and the idealized influence and inspirational motivation components of TL (Bass and Avolio 2000). For example, the ELS does not venture beyond the topic of ethics per se, whereas the VLQ and the MLQ have wide-ranging content that requires additional thought. Relatedly, much of the ELS is relatively concrete, concerning matters such as listening skills and administering discipline. In comparison, both the VLQ and MLQ involve issues relating to leader behavior over longer periods of time, in which there is greater subjective inference, such as inspiration (the MLQ) and good judgment (the VLQ). In all, from a cognitive perspective, the relatively more focused, proximal content of the ELS may have been more available to followers, and hence more influential, as they contemplated leader effectiveness shortly after completing the three leadership questionnaires.

Even though the potential relative availability of ELS content as discussed above provides a viable explanation for the dominance of ethical leadership in relation to judgments of leader effectiveness, it does not easily account for the lack of significant associations involving the other two leadership models. As we have explained, there is compelling theory linking all three approaches to leadership effectiveness. Moreover, our findings do not align well with ILT either, because dynamism (i.e., energetic, strong, dynamic) is one of only four positive traits implicated in the formation of followers’ overall impression of the leader (Epitropaki and Martin 2004) and yet, the TL components, including inspirational motivation, were not associated with subordinates’ perceptions of effectiveness.

Follower In-Role and Extra-Role Performance

As explained above, outcomes involving subordinates’ judgments may have been unduly influenced by the availability of ELS content. Importantly, this explanation does not apply to leader-sourced followers’ in-role performance wherein ethical leadership (Brown et al. 2005) was again the only significant predictor. As for extra-role performance, the ELS was the only significant, unique, predictor across samples; it was joined by the VLQ in sample 1 only. Thus, in these cases, when either of these scales were dropped from the model involving all three leadership perspectives, a significant degradation in model fit was observed.

The biggest surprise concerning the subordinate performance outcomes was that the socialized charismatic components of TL (Bass and Avolio 2000) did not hold up as a predictor in the context of the other two forms of leadership. This discrepancy relative to meta-analytic findings indicating that TL and subordinate performance are typically related (e.g., Hoch et al. 2018; Wang et al. 2011) may be because we did not assess the dimensions of TL that focus on individual followers, such as intellectual stimulation and individualized consideration (Bass 1985; Bass and Avolio 2000; Bass and Riggio 2006). Nonetheless, since these other dimensions are highly correlated with charisma (0.82 and 0.81, respectively; Avolio et al. 1999) it is unlikely that their omission alone accounts for our results. Our findings are also surprising relative to the meta-analytically derived estimate that ethical leadership only yields a 1% increase in the variance accounted for in subordinates’ in-role job performance, relative to TL (Hoch et al. 2018). In comparing our findings with Hoch et al.’s (2018), we again note that although using meta-analysis to generate and compare effect sizes is advantageous with respect to accounting for sampling error, the accuracy of the estimates are nonetheless impacted by significant unexplained across-study variance. In contrast, our results are grounded in two samples in which the all three leadership assessments were collected from the same groups of participants.

In examining our findings involving leader-provided ratings of subordinate performance, it is also important to consider that formal performance appraisals are a component of complex performance management systems (Schleicher et al. 2018) that have long been viewed as suffering from weaknesses analogous to those associated with leadership effectiveness ratings. That is, performance appraisals are also likely subject to the cognitive and motivational limitations of the rater (e.g., Feldman 1981) such that a variety of factors can affect the amount of deliberative processing that takes place (e.g., Schmidt, in press; Schleicher et al. 2018); as such, they may not be especially accurate reflections of subordinate behavior. As with judgments of leader effectiveness, automated rather than conscious controlled processing (Fiske and Taylor 2013) is likely to be involved. Once the subordinate has been categorized in terms of their performance, various biases may prevent the recognition and recall of behavior that deviates from the initial impression (Feldman 1981). Relatedly, the use of memory in relation to performance related judgments is subject to a variety of contextual factors (Woehr and Feldman 1993) as is the degree of rater motivation to provide an accurate appraisal (Harris 1994; Park 2014; Schleicher et al. 2018). In all, it is not surprising that performance appraisals are often more reflective of idiosyncratic rater differences than variations in subordinate performance (Scullen et al. 2000). Dedicated tools, including using performance diary information and structured recall techniques may be required to promote the conscious purposeful processing needed to yield performance evaluations reflective of behavioral differences among employees (DeNisi and Peters 1996).

The above considerations notwithstanding, it is noteworthy that not everyone has abandoned the possibility of making meaningful behavioral distinctions in performance appraisal contexts. For example, Russell et al. (2017) recently developed rating scales that situate ethical behavior per se within the overall nomological network of job performance. In line with their orientation, while acknowledging that both follower-provided ratings of leadership, and leader-provided judgments of subordinate performance, are likely subject to the cognitive and motivational limitations of the parties involved, our findings are consistent with the view that there is at least some additional controlled purposeful processing going on. First, attributing our findings to an overall ILT-based leadership impression (Epitropaki and Martin 2005) does not align well with the fact that in Sample 1, both the relatively focused proximal content of the ELS (Brown et al. 2005) and the much broader more inferential VLQ (Wang and Hackett 2016) accounted for unique variance in leader-rated extra-role performance. Second, analogous to our earlier point involving leader effectiveness, the lack of a unique association between the TL components and either in-role or extra-role performance is inconsistent with the view that leadership perceptions reflect a general impression solely, because dynamism, important to inspirational motivation, is one of a small set of traits crucial to the formation of followers’ impressions (Epitropaki and Martin 2004).

A relatively recent contribution to the leadership literature, implicit follower theory (IFT), also has potential implications for interpreting associations between leader behavior and follower performance. Specifically, IFT suggests that just as followers form relatively automatic quick categorical judgments of leaders that are not easily changed, leaders may do the same with respect to subordinates (Foti et al. 2017). From this perceptive, relational schema (i.e., cognitive structures reflecting expectations of regular patterns of relatedness between the parties) are crucial to understanding the ties between leadership and follower performance. There is also growing appreciation for the notion that leadership judgments can only be fully understood by, for example, accounting for the dynamic organizational context in which such judgments are made (Lord and Dinh 2014). The potential role of the social environment has long been appreciated in the performance appraisal literature as well (Levy and Williams 2004; Schleicher et al. 2018). As such, synergies may be gained in our understanding of leadership and performance by simultaneously accounting for the ideas and findings of both literatures.

Follower Ethicality

Although we expected all three follower-rated leadership approaches to associate positively with follower ethicality, only ELS-based ethical leadership (Brown et al. 2005) was a significant predictor. This was true in both samples and was not a total surprise; as explained earlier, of the three models, the ELS most clearly focused on ethicality. Interestingly, the ELS is the only model to explicitly include the possibility of punitive action against subordinates who violate ethical standards. Perhaps the possibility of sanctions is especially impactful relative to, for example, the intrinsic benefits of being a virtuous person. Finally, given that the TL components (Bass and Avolio 2000) do not especially emphasize ethicality, their lack of association with follower ethicality in the context of ethical and virtuous leadership is less surprising. Importantly, this pattern of findings helps reduce concerns about the level of construct redundancy in leadership research (e.g., Banks et al. 2018; Den Hartog 2015), especially with regard to ethical leadership and TL (Bedi et al. 2016) and adds to the conceptual arguments used to legitimize the newer models that emphasize leader ethics.

Happiness and Life Satisfaction Among Followers

Concerning happiness and life satisfaction among followers, our hypothesis that, of the three leadership approaches, the follower-rated VLQ would be most strongly related to these outcomes was partially supported, but in Sample 1 only. Specifically, as hypothesized, virtuous leadership associated positively with both follower outcomes, whereas the TL components did not. Contrary to expectations in Sample 1, ethical leadership also had relationships with these variables that were similar in magnitude to that of the VLQ. Moreover, in Sample 1, both virtuous leadership and ethical leadership were unique predictors of happiness and life satisfaction among followers; dropping either the ELS or the VLQ resulted in a significant degradation of fit in the models, whereas dropping the TL components did not. These findings are consistent with the notion that leader-modeled virtue has the potential to enhance followers’ happiness and life satisfaction as leader and follower begin to behave similarly, and that ethical leaders are supportive of employees in a manner that fosters trust and two-way communication that contributes to an overall positive work experience (Yang 2014) to ultimately impact happiness and life satisfaction (Avey et al. 2012; Judge and Watanabe 1993; Rode 2004). Yet, in Sample 2, the TL components were the only significant (and unique) predictors of followers’ happiness and life satisfaction. These findings are consistent, for example, with the view that TL contributes to followers’ psychological well-being through subordinates’ self-efficacy (Nielsen et al. 2009), trust in the leader (Kelloway et al. 2012), and its association with meaningful work (Arnold et al. 2007; Nielsen et al. 2008). It is also possible that in the process of modeling the inspirational motivation and idealized influence components of TL, leader displays of positive affect (e.g., Damen et al. 2008) impacted followers job satisfaction (Inceoglu et al. 2018) directly, to ultimately enhance their happiness and life satisfaction as well.

While it might be tempting to speculate about the role of simple demographic differences to account for the variations in our cross-sample follower affect findings, only in rare cases have such differences been shown to significantly impact the nature of leader–follower outcomes (e.g., Hackett et al. 2018). Rather, to fully account for the variation, it is likely necessary to better understand the nature of the competing mediational mechanisms apt to be operative to differing degrees within any given setting. Specifically, in their review of the literature, Inceoglu et al. (2018) found evidence of five different theory-driven sets of mechanisms that could underlie the ties between leadership and various indices of employee well-being. These include, as reflected in the TL literature, social-cognitive processes in which leaders, for example, help build followers’ self-efficacy (Nielsen et al. 2009); motivation through job design, wherein leaders ensure the work is meaningful (Arnold et al. 2007; Nielsen et al. 2008); leaders’ displays of positive emotion associated with the communication of an inspiring vision (Damen, et al. 2008); and relational processes that for example, in relation to TL, build subordinates’ trust in the leader (Kelloway et al. 2012). Thus, the differences in our findings could be accounted for by variations in the extent to which any of these processes were relatively more operative in Sample 2 versus Sample 1. The assessment of these and other possibilities, requires studies dedicated to subordinates’ well-being in which competing mediational mechanisms are directly measured in conjunction with alternative leadership approaches. The inclusion of multiple leadership models in the same study is especially important as many of the mediational processes cited above in relation to TL have been implicated in the operation of other leadership approaches as well (e.g., trust as a mediator of ethical leadership and various outcomes; Ng and Feldman 2015). Research of this kind is lacking, though this is likely to change as interest in employee well-being increases in the leadership literature (Inceoglu et al. 2018), and the potential financial implications associated improvements in these outcomes are more widely appreciated (Tziner et al. 2014).

A comprehensive interpretation of our findings concerning follower happiness and life satisfaction must also account for the fact that these outcomes were collected in the same rating session as the followers’ judgements of the three leadership approaches. As discussed earlier regarding the leader effectiveness outcome, the proximal and focused content of the ELS may have been most available to followers as they rated their own happiness and life satisfaction. This explanation, however, does not account for the similarly strong, unique VLQ-based association with these variables in Study 1, nor does it account for the lack of similar associations in Sample 2, wherein the TL components were dominant.

Happiness and Life Satisfaction Among Leaders

One of the few areas in which follower-rated ethical leadership (Brown et al. 2005) was not a dominant predictor involved leader self-rated happiness and life satisfaction. In Sample 1, in the context of the three leadership models, only the follower-rated VLQ (Wang and Hackett 2016) associated positively with leader happiness. Dropping virtuous leadership from the overall model of leader happiness resulted in a significant degradation of model fit, whereas dropping either of the other two approaches did not. Nonetheless, as noted earlier, this finding did not hold in Sample 2, where only the TL components were positively linked with leader happiness and life satisfaction.

The Sample 1 finding that subordinate-provided ratings of virtuous leadership were the single best predictor of leaders’ self-rated happiness supports our hypothesis that the practice of overall leadership excellence, reflected by the VLQ (Wang and Hackett 2016), should contribute to the greatest happiness (Flynn 2008), reflective of a good life. Moreover, as was the case with follower extra-role performance, these findings imply that as raters, both followers and leaders engaged in at least some level of purposeful processing, because it is difficult to explain how quick impression-based judgements from each party would result in a link between follower-perceived leader virtue and leaders’ self-rated happiness. The same can be said in relation to Sample 2 wherein followers’ ratings of the TL components were uniquely associated with both leaders’ self-rated happiness and life satisfaction. These findings are consistent with the view that leaders experience happiness and life satisfaction tied to their work role(s) of formulating and supporting a compelling, meaningful vision, which strengthens the ties between their self-identity and that of the organization (Shamir et al. 1993). Nonetheless, personal dispositions may also have played a role in these relationships. For example, leaders viewed as transformational score higher on emotional intelligence, conceptualized as an innate trait that enables and fosters well-being (Harms and Credé 2010). Finally, the positive links between the follower-rated TL components and leaders’ happiness and life satisfaction are also consistent with the long-established idea that expressions of positive emotion are important contributors to perceptions of leader charisma (e.g., Damen et al. 2008).

In addressing the differences across samples concerning leaders’ happiness and life satisfaction outcomes, the point made earlier with regard to followers’ affective outcomes also holds for leaders; that is, research is needed in which competing mediational mechanisms are directly measured in conjunction with alternative leadership approaches in the same study. The need here is even greater than in the case of subordinates because, as reflected in the Inceoglu et al. (2018) review, leadership research is almost exclusively focused on the well-being of followers, not leaders.

Applied Implications

In the context of the three leadership models we examined, the ELS (Brown et al. 2005) was the only significant, unique predictor of perceived leader effectiveness, followers’ in-role performance and follower ethicality. These findings add to other positive evidence concerning the incremental validity of this construct (Hoch et al. 2018; Ng and Feldman 2015). As such, our findings suggest that if businesses were to select and develop leaders based on a single approach, it should be the Brown et al. (2005) conceptualization of ethical leadership. Nonetheless, since virtuous leadership (along with ethical leadership) accounted for unique variance in followers’ extra-role performance, it cannot be disregarded in organizations looking, for example, to maximize citizenship. Moreover, if employee well-being is the focus (Inceoglu et al. 2018; Tziner et al. 2014), leader virtue (Sample 1) and the TL components (Sample 2) cannot be ignored.

Our finding that more than one leadership model may be required to optimize the full range of the outcomes we examined raises a practical issue in that followers perceive most leaders as practicing only a single leadership approach (O’Shea et al. 2009). For example, although the correlation between TL and contingent reward is substantial (e.g., 0.77; Avolio et al. 1999), O’Shea et al. (2009) found that only 38% of followers regarded their leaders to be frequent users of both. Most leaders seem either unwilling or unable to use more than one approach. As such, it should not simply be assumed that leaders typically have the complete range of behavioral options needed to foster all the outcomes examined in this study.

Limitations and Future Research

Our research has limitations that set the stage for future studies, beyond the possibilities suggested earlier. First, since all our measures were ratings-based, the influence of common method bias (Conway and Lance 2010) complicates the interpretation of the findings. Still, all the scales we used have good evidence of construct validity, which is among the strongest means available to counter method variance concerns (Conway and Lance 2010). Moreover, in most cases, our hypotheses involved ratings collected from different sources i.e., followers and their leaders. The only exception involved the analyses of the follower happiness and life satisfaction outcomes; here, despite the common method variance issue, we used followers because they were the most appropriate source given the nature of the research question involved (cf. Conway and Lance 2010). One way to further mitigate common method variance concerns, concerning the performance-based outcomes at least, is to use indices that do not involve leader judgment, such as absenteeism and other non-rating quality and quantity measures (Bycio 1992).

Our study design in both samples was cross-sectional, which limits our ability to make causal inferences. It may be, for example, that followers who are happy, life satisfied, and ethical, tend to see their leaders in a positive light (Hansbrough et al. 2015; Wang and Hackett 2016). Our lack of a longitudinal design also precluded evaluation of a range of mechanisms underlying each of the leadership models. Such designs are also needed to address more complex hypotheses involving leader ethics, including the notion that extraordinary performance is grounded in the progressive entanglement of leader character and competency (Sturm et al. 2017).

Given that our findings concerning leader/follower affect did not extend across samples, further research, as noted earlier, is required to ascertain the conditions under which different leadership approaches relate to different leader/follower affective outcomes. As explained earlier, follow-up studies should incorporate competing mediational mechanisms that may be of differential importance to the various leadership approaches. Moreover, concerning generalizability, it is notable that both of our samples were dominated by North American respondents having leadership preferences that differ from other societal clusters (House et al. 2004). Any given sample is also likely to be systematically affected by a variety of rater biases and random error (e.g., Feldman 1981; Hansbrough et al. 2015). Finally, opportunities to engage in virtuous behavior is significantly impacted by several aspects of organizational structure (Vriens et al. 2018), which could further limit generalizability of ethics-related empirical findings across settings.

Efforts to replicate our findings could also be expanded to include a wider range of leadership comparisons. For example, the Hoch et al.’s (2018) meta-analysis examined the degree to which ethical (Brown et al. 2005), authentic (Avolio and Luthans 2006) and servant leadership (Greenleaf 1977; Eva, et al. 2019) could be differentiated from TL. As referred to earlier, in the context of these comparisons, findings concerning the relative impact of TL (Bass and Avolio 2000) were much more favorable. Comparisons involving group and organizational-level performance indices are also needed since there is evidence suggesting that TL is more strongly linked to them, relative to the individual-level outcomes we collected (Wang et al. 2011). Relatedly, we did not consider innovation and adaptive performance as outcomes, which are areas of direct relevance to TL (e.g., Charbonnier-Voirin et al. 2010).

Finally, additional field studies that examine the relationship between ILTs (Epitropaki, and Martin 2005), IFTs (Foti et al. 2017), and other widely used leadership measures, are needed. It would not be surprising, for example, if a general impression of the leader, grounded in sensitivity, intelligence, dedication, and dynamism (Epitropaki and Martin 2005) or subordinates’ affect (Martinko et al. 2018), variously impacts followers’ perceptions of ethical, virtuous, and TL-based leadership, irrespective of the construct-relevant behaviors the leader exhibits (cf. Hansbrough et al. 2015). Nonetheless, our findings are consistent with the contention that leadership perceptions involve more than a quickly formed general impression. As explained earlier, evaluations of leadership based solely on a general impression do not align well with the instances in which our models did not predict, as anticipated, various outcomes. Still, the relative role of automated versus controlled purposeful processing in forming leadership-related judgments remains a compelling research topic.

To conclude, as others have noted (e.g., DeRue et al. 2011; Den Hartog 2015), there are few empirical investigations comparing a range of competing leadership approaches in the same study. By providing a within-study comparison of ethical, virtuous and socialized charismatic leadership we supplement meta-analytically derived findings (Bedi et al. 2016; Hoch et al. 2018; Ng and Feldman 2015). As evidenced by our results and study design limitations, several research opportunities remain that could contribute to addressing construct redundancy concerns in the leadership literature (Banks et al. 2018; Bedi et al. 2016; Den Hartog 2015; DeRue et al. 2011; Hoch et al. 2018).