LEADERSHIP STYLES AS PREDICTORS OF WORK ATTITUDES: A MODERATED–MEDIATION LINK

Drawing on the recent locus and mechanism models of leadership (Eberly, Johnson, Hernandez & Avolio, 2013), the purpose of this paper is to begin to explore the role of leadership styles from two different loci standpoints – (1) dyadic-focused, transformational leadership; and (2) leader-focused, transactional leadership—as important antecedents to individual and organizational outcomes in the organizational context. Among employees in several organizations in Israel (N=265), we investigated the relationships between (1) several individual and organizational work outcomes (i.e., job engagement, work enjoyment); and (2) the two leadership styles (the predictors, namely, transformational and transactional leadership). In addition, we explored the roles of (3) a possible mediational mechanism through which we posited the leadership styles operate (i.e., work drive); and (4) a possible moderator (i.e., organization types). These associations were presented as a model that was both tested via multi-group moderation structural equation modeling (SEM) and through moderated–mediation analyses via competing models of demographical differences. The findings illustrated that both transformational and transactional leadership styles have a direct, positive influence on outcomes. However, with regard to the intermediary moderator and mediator variables, the results demonstrate varied and interesting relationships in current study, the indication being that each of the two leadership styles, when interfacing with unique combinations of moderator and mediator, produce outcomes specific to the leadership style. Important concepts, recommendations, and implications are discussed.


Introduction
We begin our discussion by noting that after years of research in the field of leadership, Hackman and Wageman (2007, p. 43) assert that "there are no generally accepted definitions of what leadership is, no dominant paradigms for studying it, and little agreement about the best strategies for developing and exercising it." Nevertheless, in what appears to be a major departure from the classical views concerning leadership that stress the leader-and-the-led, recent research into the concept of leadership has focused on leadership as an "ongoing process of social influence" (Yukl, 2010) whereby leader and followers mutually influence each other. A recent paper by Eberly, Johnson, Hernandez, and Avolio (2013) significantly refined this notion of mutuality by regarding leadership as: "the exertion of social influence between and among multiple loci of leadership (leader, follower, leader-follower dyad, collective, and context) working toward a common goal, via the leadership mechanisms of traits, behaviors, affect, and cognition, through a series of event cycles that may or may not include the same mechanisms and/or loci" (Eberly et al., 2013, p. 439).
Of note is that one of the most common contexts where such a paradigm of leadership is extant is the work environment, "where we tend to invest most of our waking hours" (Landy & Conte, 2016), and seem to be doing [that] at an increasing rate in recent years (Lee, McCann & Messenger, 2007), such that it becomes clear that, "work captures an essential share of our lives" (Arvey, Harpaz, & Liao, 2004). Since most of us pursue these activities in the context of interacting with our managers-the organizational leaders-it would appear to be of paramount importance to investigate the intricate relationships between managers and their subordinates.
Thus, in the current study, we approached this route of inquiry by exploring the effects that management leadership styles have on their employees' work attitudes and experiences. Following recommendations in the literature (e.g., Staw & Cohen-Charash, 2005), we paid particular attention to the organizational context in which the leader-follower interactionthe dyadic exchange-takes place, an aspect of leadership in the organizational context only scarcely scrutinized in recent years. Specifically, while the present study aims to shed additional light on the outcomes of leadership styles, it concurrently represents an attempt to examine insufficiently studied mediation and moderation effects that are likely to impact the link between leadership styles and outcomes (Avolio, 2007) (see below).
Recently, Hernandez, Eberly, Avolio, and Johnson (2011), and later Eberly et al. (2013), created a "locus-mechanism" leadership model consisting of two "composing" elements. These are the locus, the origin point of the leadership (e.g., the leader, the follower, the context), while the mechanism is how the leadership is communicated and transmitted (e.g., direct leadership behaviors or indirect influences on cognitions and emotions of followers).

Leadership Styles-Transformational and Transactional
Leadership styles, of course, vary among leaders. The literature has most recently favored distinguishing between transformational and transactional leadership styles that are distinct but not necessarily mutually exclusive. Thus, within an individual's leadership style repertoire one style can exist exclusively or that specific approach (towards leading others) can also coexist with another (Fein, Tziner, & Vasiliu, 2010;Kark, Van Dijk, & Vashdi, 2018;Xenikou, 2017). Hence, we believe that pigeon-holing a leader strictly and solely into either style is artificial and does not represent the managerial reality well.
Nevertheless, although we have argued that leadership styles are not necessarily exclusive, looking at the dichotomy between transformational and transactional leadership styles enables us to perceive the possible range of leadership mechanisms likely to be employed in the workplace.
Thus, formally, a transformational (or charismatic) leader inspires subordinates (or followers) and entire collectives by influencing and managing their behaviors via their belief systems (cognitions) and emotions (affect) through the expression of a collective vision and positive emotions that induce inspiration (Bass, 2007;Jung & Avolio, 1999;Kark et al., 2018;Yaffe & Kark, 2011). Following the two-dimensional framework indicated above (Hernandez et al., 2011), the locus in transformational leadership is dyad focused, such that there is an extant reciprocal dyadic process, rather than a leadership paradigm that is unidirectional. While the commands filter down the hierarchy, the leader is still open to debate and may be influenced by the followers as well. This transformational style tends to influence the "followers" through an entire spectrum of mechanisms: affect, cognitions, behaviors and traits.
As opposed to the transformational style of leadership, the locus of transactional (or monitoring) leadership is the person who is the (traditional) leader: the source of the leadership initiative originates from the leader and the leader alone (commands go down the hierarchy and are unidirectional or even unilateral). The leader will thus more likely invoke a mechanism that is authoritative and direct in order to influence followers' behaviors. This style has been conceptualized in terms of an exchange process, in which rewards are offered for compliance and punishment for noncompliance (Bass, 2007;Jung & Avolio, 1999;Yaffe & Kark, 2011). The transactional leader sets standards and norms and highlights obligations, while directing subordinates to perform tasks in the "correct and expected way," which encourages conformity and compliance (Bass, 1985;Gorman et al., 2012;Kark, Katz-Navon, & Delegach, 2015;Kark et al., 2018;Neubert, Kacmar, Carlson, Chonko, & Roberts, 2008;Tseng & Kang, 2009).
In the present study, we investigated the links between these two leadership styles and two f work outcomes (i.e., job engagement, work enjoyment) while, concurrently, investigating the role of a possible mediator, namely, "work drive," and "organization type" as moderator. The research model is depicted in Figure 1.

Work Drive (WD) and Work Enjoyment (WE)
Regarding the proposed mediators in the leadership style-outcome paradigm, we first picked up on the concept of work drive (WD) as the mediator and work enjoyment (WE) as the outcome, based on Shkoler, Rabenu, and Tziner's (2017) study on workaholism, in which they differentiated between these two distinct and independent factors (after Spence and Robbins's [1992] workaholism triad). WD is considered as a dispositional trait that constitutes the inner pressures that compel the employee to act within the work setting. WE is the actual pleasure derived from the work itself (Spence & Robbins, 1992). Shkoler et al. (2017), focusing on the distinctions between these two dimensions, advocated for an order of precedence to be established between them. Thus, as WD is a dispositional trait, so "the uncontrollable internal drive to work precedes the experiences that might derive from the work itself (e.g., work enjoyment)" (Shkoler et al., 2017, p. 194). Notably, the researchers' two-study research yielded a positive association between these two dimensions (even though the authors did not assume a one-tailed directional hypothesis in this regard). For the purposes of the current paper, we hypothesized that: H1: Work drive positively associates with work enjoyment.

Leadership Styles and the Relationship Between Work Drive and Work Enjoyment
While the concepts of WD and WE are interesting in and of themselves, we wished to add to our understanding of these constructs by examining the role of the leader in affecting (1) the drive to work; and (2) the enjoyment derived from the work itself via leadership styles. For this purpose, we drew upon the Trait Activation Theory (Tett & Burnett, 2003), in which "trait activation is the process by which individuals express their traits when presented with trait relevant situational cues" (p. 502). These cues can activate personality traits that may be related to job tasks and organizational expectations (e.g., job performance) or not (Tett & Burnett, 2003).
In this sense, we suggest that leadership style may act as a trigger (i.e., a situational cue) for the activation of a job-related trait such as work drive, manifested in the job context. Both leadership styles-transformational leadership and transactional leadership, respectivelyare conducive to making the employee more involved with the work one way or another, whether by the "carrot and stick" approach or through the employment of more inspirational/motivational tactics. The worker (follower) perceives that a reaction (work) is required in response to the leader's cue: The leader thus activates the worker's drive to work. Hence, we hypothesized that: H2.1: Transformational leadership positively associates with work drive.
Furthermore, as stated above, the two leadership styles are distinguished by their loci and mechanisms and hence, presumably, are also differentiated in terms of their respective effects on the employees. In any event, we would expect subordinates exposed to either leadership style to experience enjoyment from work: under a transformational leader, they would enjoy personalized and inspiring attention with relatively intangible rewards (e.g., empowerment, mentoring), while, under the tutelage of a transactional leader they would enjoy tangible rewards, such as bonuses and similar material incentives. Hence, we hypothesized that: H3.1: Transformational leadership positively associates with work enjoyment.

Job Engagement and Leadership Styles
As stated above, it is fairly safe to assume that JE may be an attitudinal response to both situational cues (i.e., leadership) and dispositional traits (i.e., work drive). The rationale behind the effect of leadership is nested within the unique management type each style promotes. Thus, job engagement may stem from both motivational and mentoring support (i.e., transformational leadership) or from monitoring compliance (i.e., transactional leadership). We argue that both styles of leadership are likely to nurture employee engagement.
In addition, we propose that work drive, as a dispositional trait that triggers high investment and efforts in the job, may also enhance another outcome, namely, the levels of workers' job engagement. Thus, the higher the manifestation of the work drive, the more work driven the employee becomes, resulting in increased overall engagement. Hence, we hypothesized that: H4.1: Transformational leadership positively associates with job engagement.
H5: Work drive positively associates with job engagement.

Leadership Styles, Work Drive, Work Enjoyment and Job Engagement
While the two leadership styles clearly may have a direct and independent effect on engagement and enjoyment in the job, we have also indicated that each leadership style activates the employee's drive to work and it is the work drive that may contribute towards (enhanced) enjoyment and higher engagement. As such, work drive acts as a mediational mechanism through which the leader can manage effects on workers' engagement and enjoyment. Hence, we further hypothesized that:

Organization Types
With respect to possible moderators on the predictor (leadership style)-outcome relationship (see Figure 1), we turned our attention to organization type. Of note, not much research has been conducted on the possible differences between sectors in the working market (e.g., private, public, government). Most of the extant research has been concerned with investigating specific sectors, independently of each other (e.g., Kunze & Miller, 2017;Yeo, Ananthram, Teo, & Pearson, 2015) and scarcely have the different sector types been compared (but see Johnson, Leenders, & McCue, 2017;Top, Akdere, & Tarcan, 2015).
Nevertheless, the available research has revealed that organizations in the public sector demonstrate greater bureaucracy, more formalization of rules, regulations, and hierarchical authority structures than their private sector counterparts (e.g., Boyne, 2002). Furthermore, public sector organizations are less innovative and less tolerant of risks (Aarons, Sommerfeld, & Walrath-Greene, 2009;Bozeman & Kingsley, 1998). Moreover, managers in government workplaces are less entrepreneurial than their counterparts in the private sector (Moon, 1999). Based on these highlighted differences between the sectors, we can assume that managers will need either to employ different leadership styles in order to effectively lead subordinates or that the same managerial skills will have differential effects on the followers (employees) in different types of organizations. Hence, we hypothesized that: H8: Organization types moderate the associations depicted in the model (H1-H7.2), such that they will vary according to organization type.

Measures
Leadership style was gauged using the 36-item Multifactor Leadership Questionnaire (MLQ; Avolio & Bass, 1991), on a Likert-type scale ranging from 1 ("strongly disagree") to 6 ("strongly agree"). Transactional leadership was gauged by 12 items; for example, "Your leader assists you based on effort." In the present study, reliability was adequate (alpha = 0.70, M = 3.25, SD = 0.65). Transformational leadership was measured by 24 items; for example, "Your leader teaches and coaches." In the present study, there was a high reliability (alpha = 0.96, M = 4.06, SD = 1.13).
Work drive and work enjoyment was gauged using the Workaholism Battery (Work-Bat; McMillan, Brady, O'Driscoll, & Marsh, 2002; see also Shkoler et al., 2017), consisting of 21 Likert-type items ranging from 1 ("does not describe my work at all") to 6 ("describes my work exactly"). Seven items measure the enjoyment factor of workaholism (e.g., "I like my work more than most people do"), and seven measure the drive factor of workaholism (e.g., "I feel guilty when I take time off work"). (The other seven items measured the involvement factor, but they were not included in the present study; for further reading, see Shkloler et al., 2017.) Cronbach's alpha of the enjoyment and drive factors, respectively, were 0.88 and 0.73 (Huang, Hu, & Wu, 2010). In the present study, there was a good reliability for work enjoyment (alpha = 0.89, M = 3.51, SD = 1.12) and for work drive (alpha = 0.84, M = 3.54, SD = 1.13).
Organization type was gauged by a single demographical item: "In which organization type are you currently working?" (1 = Private, 2 = Public, 3 = Government).

Procedure
The survey (pencil-paper) was given to working people in the various organizations to complete voluntarily. After we collected all the data, analysis was performed using SPSS (v. 22) and AMOS (v. 22) software packages.

Common-method bias (CMB).
Harman's single-factor test (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003) was used to assess the extent to which intercorrelations among the variables might be an artifact of common-method variance (CMV). The first general factor that emerged from the analysis accounted only for 28.22% of the explained variance. While this result does not rule out completely the possibility of same-source bias (i.e., CMV), according to Podsakoff et al. (2003) less than 50% (R 2 < 0.50) of the explained variance accounted for by the first emerging factor indicates that CMB is an unlikely explanation of our investigation's findings.

Results
In order to test the model of current study, we mainly employed structural equation modeling (SEM) with multi-group moderation analyses. The path diagram for model (across different organization types) is presented in Figure 2, with the coefficients and their significance levels (and fit indices).  Note: *p < 0.05, **p < 0.01, ***p < 0.001; TA = transactional leadership style; TF = transformational leadership style; WD = work drive; WE = work enjoyment. JE = job engagement.

The bivariate correlation matrix is presented in
Interestingly, Table 1 indicates that our hypotheses H1-H5 were both supported and rejected, sporadically.
As can be seen in Figure 2, the model's fit is in the absolute sense (see Byrne, 2010). However, in terms of mediation effects, not all the mediation conditions were met in each model; significant effects of: (1) predictor → criterion; (2) predictor → mediator; (3) mediator → criterion; and (4) the direct effect (path c) should be less ("weaker") than the total effect (path c) (for further reading, see Baron & Kenny, 1986;Frazier, Tix, & Barron, 2004;Hayes, 2013).
Note: *p < 0.05, **p < 0.01, ***p < 0.001. chi square(df) = 3.97 (3)  Therefore, when testing for the significance of the mediation effect via bootstrapping (see Preacher & Hayes, 2008), we chose only the paths that actually met all of the aforementioned mediation conditions and whose indirect effects were statistically significant. The findings are presented in Table 2.

Distinctions between organization types
Government sector. As can be seen in Table 2, the government sector did not have any mediation effects within.
Private sector. Regarding the private sector, both transactional and transformational leadership styles affected job engagement through the work drive-as a full mediator for transactional leadership and as a partial mediator for transformational leadership. However, no mediation effect occurred for work enjoyment.
Public sector. In addition, regarding the public sector, both transactional and transformational leadership styles affected work enjoyment through the drive to work (as a full mediator for transactional leadership, and as a partial mediator for transformational leadership); however, no mediation effect occurred for job engagement. A summary of the results in regard to the hypotheses is displayed in Table 3. Note: WD = work drive; WE = work enjoyment; JE = job engagement. A positive mathematical sign (+) indicates a hypothesis about positive correlations; (n.s.) = nonsignificant correlation. (neg) = although statistically significant, the (negative) relationship is contrary to the hypothesis.

Discussion and Conclusions
The present research can be considered as exploratory and aimed at identifying possible roles of leadership styles as important predictors in the organizational context. To this end, we drew upon the recent locus-mechanism model of leadership (Eberly et al., 2013;Hernandez et al., 2011) to investigate two different loci of leadership: (1) dyadic focused (transformational); and (2) leader focused (transactional). Implementing two independent yet related studies within the framework of an overall model of the association of leadership style and work outcomes (see Figure 1), we investigated across three types of work settings-government, public and private-two outcomes of these styles (i.e., job engagement, work enjoyment), a possible mediational mechanism through which we posited the leadership styles operate (i.e., work drive), and a possible moderator (i.e., organization types).
Most of our research hypotheses were supported in the private sector, but less so in the public sector and only three hypotheses were supported in the government sector: • In all sectors, both transformational and transactional leaderships led to increased WD and WD led to enhanced WE.
• In all sectors, transformational leadership led to increased WE.
• Under transformational leadership in the private and public sectors (but not in the governmental sector), WD led to increased JE.
• By contrast, transactional leadership did not lead to WE, at all.
• In the private sector, both leadership styles led to increased JE, and for the public sector transformational leadership (but not transactional leadership) also led to JE.
• In addition, in the public sector, two effects (transactional → WE, and transactional → JE) were found to be statistically significant but negatively signed, contrary to our hypotheses.
• Mediation effects (of leadership styles-through-WD) were found on JE, only in the private sector.
• Mediation effects (of leadership styles-through-WD) were found on WE, only in the public sector.

Theoretical implications
Our findings stress the importance of examining different loci and mechanisms of leadership styles (see Eberly et al., 2013;Hernandez et al., 2011) for they may end in different outcomes, as observed in the results of this study (see results).
As mentioned above, the investigated mediator-work drive-affected outcomes as predicted, albeit each in a different way and if not totally under all circumstances, then at least partially. The generic point, however, is that independent of the specific outcome, the principle holds that managers and supervisors in an organizational setting can influence their employees through activating their drive to work (see Tett & Burnett, 2003).
We also demonstrated the importance of testing competing models through multiple-group moderation, and not through specific moderation/interaction effects. This enables us to test models in different "settings," while remaining as parsimonious as possible. In the context of the present research, such a method also allows us to identify cognitive, affective, and behavioral patterns in a broader fashion than more traditional methods.

Practical implications
While in some instances both leadership styles studied indicated the same positive outcomes, independent of which end of the dichotomy of styles each represented, from a managerial standpoint, our findings also indicate that each of the two styles, when interfacing with combinations of various moderators and mediators produced outcomes specific to those combinations and leadership style. Consequently, managers, supervisors and other "loci" would do well to exercise flexibility and to adjust their (preferred) styles accordingly with respect to specific outcomes that they favor. Moreover, based on the various and "mixed" results of these studies, we could assert in general terms, and state more categorically, that under the same rooftop a leader needs to exert a leadership style (apply the appropriate mechanism) most suited to the organizational context (type of work setting: governmental, public, private) as revealed in current study.
Although the literature has opted to dichotomize leadership into two somewhat opposing ends of a leadership scale-and our subjects responded to the leadership questionnaire accordingly, isolating, as it were, their supervisors' tendencies to adapt one style of leadership over the other-we nevertheless argue that there is no "one best leadership style," because the exercising of the style is actually dependent on: (1) the situation (organizational context; type of work setting); (2) the followers' individual differences (i.e., work drive); and (3) the context in which the skills are utilized. Additionally, if we take these conclusions seriously, then we bring to mind that despite the dichotomy of styles presented in the model, we can understand that leadership styles are not necessarily mutually exclusive. From this perspective, effective leadership is like a river: it flows as the currents takes it but with clear boundaries. By contrast, leadership that is less effective is like a mountain-it struggles against the wind of change, unmoving and uncompromising.

Limitations
The use of self-reporting measures may prove a limitation. While several of our variables are cross-sectional (e.g., organization types), we measured leadership only from the point of view of leader (or giver), but not from the perspective of the follower (or receiver). And since we are discussing dyadic relationships, as far as transformational relationships are concerned, we may yet have missed some further intriguing aspects of the mechanisms of leadership in the workplace; for instance, regarding the effect that "followers" (the locus) have on their leaders and consequently on the (official) leaders' leadership styles and decision-making processes.
We investigated a delimited number of outcomes but we did not tap into several other work outcomes that are surely affected by the leadership styles of managers and supervisors, such as organizational citizenship behaviors (OCBs), counterproductive workplace behaviors (CWBs), and de facto turnover intentions.

Future research
Clearly, there is room for much more investigation and analysis of the specific results across types of leadership, type of organization, age levels and their relationship with specific outcomes. By way of illustration, just a few examples from our study in this paper indicate how we might want to tease out the explanations for the following (tentative) results:

Differences between sectors:
 In the government sector, no mediator effects were indicated at all; while in the private sector, work drive was found to mediate for JE but not for WE. Yet, conversely, in the private sector, work drive was found to mediate for JE but not for WE; yet in the public sector, WD found for WE but not for JE!  Under transformational leadership in the private and public sectors (but not in the government sector), WD led to increased JE.
 Only in the private sector were mediation effects found for leadership styles-through-WD on JE, and only in the public sector were mediation effects found for leadership stylesthrough-WD on WE.
How do we explain the differences between the sectors?

Differences between leadership styles:
In all sectors, transformational leadership led to increased WE. By contrast, transactional leadership did not lead to WE at all. The result is strange because conventional wisdom would say that a dyadic, interpersonal empowering leadership style would induce greater work enjoyment than an authoritarian style transformational leadership style.
So, we reiterate that it would appear that more replications and trial experimentation with these and other possible variables will reveal much more needed information. This information will enable the theoretical underpinnings of the various leadership mechanisms and their effectiveness among employees (varying in socio-demographic features; in type of organization; in workplace circumstances) to be better understood and explained. Specifically, we recommend that future investigations examine the full model displayed in Figure 1, from which the current study's model was derived.