Introduction
President Dwight D. Eisenhower (Reference Eisenhower, Peters and Woolley1954) described leadership as ‘the art of getting someone else to do something that you want done because he wants to do it, not because your position of power can compel him to do it’ (p. 1). Over 60 years later, scientists and practitioners are still working to understand the fundamental factors driving this process. To date, researchers have offered several models to frame the elements of power (e.g., Peabody, Reference Peabody1961; Kipnis, Schmidt, & Wilkinson, Reference Kipnis, Schmidt and Wilkinson1980; Shukla, Reference Shukla1982; Yukl & Falbe, Reference Yukl and Falbe1991; Brauer & Bourhis, Reference Brauer and Bourhis2006; Reiley, Kelley, & Soboroff, Reference Reiley, Kelley, Soboroff, Watola and MacIntyre2018). However, French and Raven's (Reference French, Raven and Cartwright1959) original taxonomy of social power remains the most prominent classification used in empirical studies (Cobb, Reference Cobb1980; Rahim, Antonioni, Krumov, & Illieva, Reference Rahim, Antonioni, Krumov and Illieva2000; Wood, Reference Wood2014).
French and Raven (Reference French, Raven and Cartwright1959) categorized five bases of social power (legitimate, reward, coercive, expert, and referent power) which agents (e.g., leaders) use to influence targets (e.g., followers). Legitimate power is position-based, formal authority which stems from the targets’ belief an agent has a right to influence them and they are obligated to accept this influence. Reward power is also position-based, but drawn from an agent's perceived ability to control resources and compensate the target for compliance actions. Coercive power is derived, similarly, from the target's expectation of punishment as a consequence of failing to obey or achieve the agent's goals. Expert power is rooted in the target's perception of the agent's personal, task-relevant knowledge or abilities. Referent power originates from the respect and attraction the target has toward the agent.
Podsakoff & Schriesheim's (Reference Podsakoff and Schriesheim1985) review and critique of the power literature based on French and Raven's (Reference French, Raven and Cartwright1959) taxonomy noted most studies focused only on correlating the direct relationships found between ratings on these five power bases and various dependent variables, such as measures of productivity, compliance, absenteeism, and retention. These studies offered limited theoretical insight into how the application of power led to organizational outcomes and sometimes yielded inconsistent findings, in part, because these direct approaches failed to recognize other key characteristics or dispositions among agents and targets within the context of the situation. More researchers have since joined Podsakoff and Schriesheim's (Reference Podsakoff and Schriesheim1985) appeal for deeper empirical investigations of mediating variables, which may explain how or why the use of these powers affect behavioral outcomes (e.g., Carson, Carson, & Roe, Reference Carson, Carson and Roe1993; Rahim, Antonioni, & Psenicka, Reference Rahim, Antonioni and Psenicka2001; Sturm & Antonakis, Reference Sturm and Antonakis2015).
Unfortunately, power's nefarious reputation has tainted the mere mention of the term in western, democratic societies and caused many good people to avoid power entirely (Gardner, Reference Gardner1993). This lingering negative association seems to also drive many leadership researchers away from investigating the topic and, as a result, fundamental power mechanisms are often overlooked in the leadership process (Pfeffer, Reference Pfeffer1992; Collinson & Tourish, Reference Collinson and Tourish2015). To overcome this research trend, the current study examined not only the direct relationships between a leader's power use and followers' performance, but also accounted for power's intermediate effects on the followers' attitudes, which provide a more nuanced perspective of these mechanisms and outcomes.
While power use does not necessarily equate to leadership, contemporary leadership approaches inherently employ some form of power (Reiley, Kelley, & Soboroff, Reference Reiley, Kelley, Soboroff, Watola and MacIntyre2018). For example, contingency theories, relationship-based theories, charismatic leadership styles, and transactional–transformational approaches call for various degrees of formal support, contingent rewards, corrective punishment, charismatic appeal, expert guidance, or intellectual engagement to address the needs of followers in a given situation. Nevertheless, prevailing power research tends to focus on the power holder (Sturm & Antonakis, Reference Sturm and Antonakis2015), with little regard for the broader leadership process.
The power to influence others resides within a complex confluence of leadership factors related to the leader, the followers, and the situation (Avolio, Reference Avolio2007; Bennis, Reference Bennis2007). However, as Collinson and Tourish (Reference Collinson and Tourish2015) warned, mainstream leadership studies and teachings:
tend to privilege and romanticize individual leaders whilst also underestimating the dynamics of power, the influence of context and the significance of follower dissent and resistance. They tend to assume that the interests of leaders and followers automatically coalesce, that leadership is an uncontested form of top-down influence, follower consent is its relatively unproblematic outcome and resistance is abnormal or irrational. (p. 580)
This highlights the need to assess the fundamental dynamics of power use, and its associated outcomes, within the broader leadership dynamic.
Indeed, President Eisenhower's leadership advice goes beyond top-down power-wielding and compliance, pointing us to examine the mediating factors linking leader power use to the performance of others. The current study investigated these complex relationships, because important mediating variables help define the nature of the leader–follower dynamic and provide a more complete picture of the social patterns shaping behavioral outcomes. Specifically, the intermediary effects of leader power use on follower attitudes may help explain the overall relationship between leader power use and follower performance. For example, followers may feel more satisfied with their job and consequently excel in their positions when their leaders provide knowledgeable support and expertise. On the other hand, followers' job satisfaction may wane under the condition of working for a leader who relies on his or her status and position to sway subordinates, ultimately degrading the followers' performance. When organizational leaders gain the respect and admiration of their followers, these followers may feel more deeply committed to the success of their organization and even perform extra tasks and functions not formally assigned to them. Conversely, experiencing persistent coercive pressure from their leaders may make followers question their loyalty to the organization and their commitment to working under such conditions, which may diminish their willingness to engage in these ‘extra-role’ behaviors. In these cases, the direct effects of a leader's power use on followers' performance is better understood by evaluating how this power use relates to the followers' attitudes, which may ultimately promote or deter performance outcomes.
The current study addressed noted gaps in the power and leadership literatures by examining the indirect mechanisms by which leader power use may lead to performance. Specifically, this study investigated the mediating effects of followers' job satisfaction and organizational commitment on the relationships found between their leaders' perceived use of power and the followers' performance.
Theoretical Development and Hypotheses
Bases of power
Bass (Reference Bass1960), and later Etzioni (Reference Etzioni1975), grouped French and Raven's (Reference French, Raven and Cartwright1959) original five power bases and classified legitimate, reward, and coercive power as positional powers drawn from a person's formal role in an organization. They classified referent and expert power as personal powers derived from an individual's inherent abilities or attributes. Examinations of these powers have suggested the ability to influence others is rooted in not only an individual's position or personal characteristics, but also in the target's perception and belief that the influencing agent has the right or capacity to influence him or her (Hinkin & Schriesheim, Reference Hinkin and Schriesheim1989; Barbuto, Reference Barbuto2000; Kelley, Dobbs, Lucas, & Lovaglia, Reference Kelley, Dobbs, Lucas and Lovaglia2017; Reiley, Kelley, & Soboroff, Reference Reiley, Kelley, Soboroff, Watola and MacIntyre2018).
Personal power (i.e., expert and referent power) has been shown to be related positively to several desirable workplace outcomes, such as productivity (Fiorelli, Reference Fiorelli1988), satisfaction with supervision (Bachman, Reference Bachman1968; Busch, Reference Busch1980), reduced absences and turnover (Student, Reference Student1968), compliance (Bachman, Bowers, & Marcus, Reference Bachman, Bowers, Marcus and Tannenbaum1968; Rahim, Reference Rahim1989), extra-role performance beyond defined job tasks (Reiley & Jacobs, Reference Reiley and Jacobs2016), and trust (Raven, Reference Raven2017). However, legitimate and coercive power are generally related negatively to these outcomes. In spite of these more common findings, legitimate and coercive power have sometimes been found to be unrelated to these and other target outcomes, and reward power has yielded both positive and negative effects on the same constructs across studies (Podsakoff & Schriesheim, Reference Podsakoff and Schriesheim1985; Carson, Carson, & Roe, Reference Carson, Carson and Roe1993).
In addition to the noted theoretical limitations associated with the power literature's historical focus on direct relationships, Podsakoff and Schriesheim (Reference Podsakoff and Schriesheim1985) argued the lack of consistent relationships between agent power and target outcomes may also be due to widespread measurement limitations. Many power studies utilized ipsative scales, which tend to force negative empirical relationships among the power bases and may cause lower ranked powers to be related to outcome variables in the opposite direction of higher ranked powers, and/or single-item measures, which may be less reliable and often had questionable content validity (Podsakoff & Schriesheim, Reference Podsakoff and Schriesheim1985; Schriesheim, Hinkin, & Podsakoff, Reference Schriesheim, Hinkin and Podsakoff1991; Rahim et al., Reference Rahim, Antonioni, Krumov and Illieva2000). Drawing from the sparse number of studies employing Likert-type items and more refined multi-item measures, expert, referent, and reward power seem to be related positively to performance, while legitimate and coercive power appear to be negative or weak predictors of job performance (Rahim, Antonioni, & Psenicka, Reference Rahim, Antonioni and Psenicka2001; Rahim, Reference Rahim, Tjosvold and Wisse2009). Clearly, more contemporary research into these relationships is needed.
While the field has been slow to respond to Podsakoff et al.'s appeal for more robust and complex investigations of French and Raven's (Reference French, Raven and Cartwright1959) power bases, a growing amount of research has begun to explore the intermediary links that help explain and contribute to the effects of power use. Rahim and colleagues' (Rahim & Psenicka, Reference Rahim, Psenicka, Rahim, Golembiewski and Lundberg1996; Rahim, Antonioni, & Psenicka, Reference Rahim, Antonioni and Psenicka2001) analyses of these power bases provided a clearer understanding of how these power sources affect one another. Their research suggests legitimate power influences perceptions of coercive power negatively, and expert power influences referent power positively (Rahim & Psenicka, Reference Rahim, Psenicka, Rahim, Golembiewski and Lundberg1996). Rahim, Antonioni, and Psenicka's (Reference Rahim, Antonioni and Psenicka2001) investigation of 1,116 manager–subordinate dyads found both legitimate and reward power influenced perceptions of expert power positively, and this expert power, in turn was positively related to referent power. They then showed referent power was related positively to the use of a more integrative and less avoidant problem solving style, which in turn positively influenced employees' job performance. These findings supported Carson, Carson, & Roe's (Reference Carson, Carson and Roe1993) meta-analytical conclusion that position powers also influenced subordinate outcome variables through their indirect effects (IE) on personal powers.
Other power studies have identified associated mediating variables that may enable or obstruct power–outcome relationships. For example, Jain, Giga, and Cooper's (Reference Jain, Giga and Cooper2011) analysis of mid-level executives found reward, expert, and referent power were related positively to organizational citizenship behaviors (OCBs) (Organ, Reference Organ1988a, Reference Organ1988b), and these power bases contribute to both personal and organizational effectiveness through these supportive citizenship behaviors. Liao (Reference Liao2008) demonstrated trust provided a link between managers' use of power and R&D employees' engagement in knowledge-sharing behavior. Additionally, employees' feelings of procedural justice have been shown to mediate the relationships between their leaders' power use and the employees' affective work reactions (Mossholder, Bennett, Kemery, & Wesolowski, Reference Mossholder, Bennett, Kemery and Wesolowski1998). However, more research is needed to empirically identify and understand adjuvant attitudinal factors bridging the crucial power–performance relationship.
Job satisfaction and commitment mediating the power–performance relationship
Efforts to understand the criterion of performance have resulted in the notion individuals contribute to organizational effectiveness both directly and in ways beyond their formal, ‘in-role’ job requirements (Borman & Motowidlo, Reference Borman, Motowidlo, Schmitt and Borman1993). Organ (Reference Organ1988a, Reference Organ1988b) classified this ‘extra-role’ performance dimension as Organizational Citizenship Behavior (OCB), and defined it as discretionary individual behavior beyond the prescribed role or tasks of the job not explicitly or directly recognized by the formal system of compensation, which collectively promotes effective organizational functions (p. 4). While both in-role and extra-role activities are important elements of performance, OCBs differ from in-role, task activities in that they support the social and psychological environment in which task performance occurs (Borman & Motowidlo, Reference Borman, Motowidlo, Schmitt and Borman1993). Investigations into the impact of leader power use on performance should examine both similar and differential effects on in-role and extra-role behaviors, as these behaviors define and support organizational effectiveness in distinct and essential ways.
Since performance is ultimately a central concern for organizations, the behavioral and psychological importance of employee engagement and work attitudes, as drivers of performance, cannot be overstated among scientists and practitioners today. Consequently, the practical need to understand potential relationships between leader power use and follower attitudes – as well as the association between these attitudes and performance – has spurred investigations into these discrete elements. Job satisfaction is the degree to which individuals have a positive affective orientation toward their job experiences (Locke, Reference Locke and Dunnette1976; Price, Reference Price1997). Mowday, Steers, and Porter (Reference Mowday, Steers and Porter1979) framed organizational commitment as ‘the relative strength of an individual's identification with and involvement in a particular organization’ (p. 226). Much has been written about the positive effect of each of these variables on performance with a variety of results reported. When investigating the role of power use on performance these attitudinal factors are important to consider, since they shape followers' behavioral intentions and predict performance (Ajzen & Fishbein, Reference Ajzen and Fishbein1980; Fishbein & Ajzen, Reference Fishbein and Ajzen2010; Raymer, Kelley, & Laney, Reference Raymer, Kelley, Laney, Watola and MacIntyre2017).
Prior research (e.g., Rahim & Afza, Reference Rahim and Afza1993; Elangovan & Xie, Reference Elangovan and Xie2000; Pierro, Raven, Amato, & Bélanger, Reference Pierro, Raven, Amato and Bélanger2013; Junaimah, See, & Bashawir, Reference Junaimah, See and Bashawir2015) has shown supervisors' expert, referent, and reward power are related positively to subordinates' satisfaction and commitment, while legitimate and coercive power are related negatively to these work attitudes. Examinations of power have also suggested person-based powers (i.e., expert and referent power) are related more positively to organizational commitment than position-based powers (i.e., coercive, reward, and legitimate power) through the building of stronger interpersonal bonds between leaders and followers (Etzioni, Reference Etzioni1975; Franklin, Reference Franklin1975). At the heart of this relationship is the likelihood that person-based powers help to build trust between leader–follower dyads versus an impersonal, utilitarian approach.
Beyond these links between leader power use and follower attitudes, both job satisfaction and organizational commitment have been associated, to various degrees, with in-role performance (e.g., Williams & Anderson, Reference Williams and Anderson1991; Judge, Bono, Thoresen, & Patton, Reference Judge, Bono, Thoresen and Patton2001; Wright & Bonett, Reference Wright and Bonett2002; Fisher, Reference Fisher2003) and extra-role OCBs (e.g., Bolon, Reference Bolon1997; Ackfeldt & Coote, Reference Ackfeldt and Coote2005). Bateman and Organ (Reference Bateman and Organ1983) drew from social exchange theory (Adams, Reference Adams1965) to rationalize job satisfaction influenced performance – and particularly citizenship behaviors – since ‘people seek to reciprocate those who benefit them. To the extent a person's satisfaction results from the efforts of organizational officials and such efforts are interpreted as volitional and non-manipulative in intent, the person will seek to reciprocate those efforts’ (p. 588). Furthermore, Riketta's (Reference Riketta2002) meta-analysis indicated organizational commitment was also related to performance – and marginally stronger for extra-role performance – across a range of organizations. Meyer and Allen (Reference Meyer and Allen1991) proposed organizational commitment has three dimensions: affective (the desire to stay with the organization), normative (the obligation to the organization), and continuance (the perceived cost of leaving the organization). Among these dimensions, affective organizational commitment most closely captures an individual's emotional attachment to, identification with, and involvement in the organization, which may provide insight into the psychological state and attitudes followers hold toward the organization relative to the leader power use they experience. Affective organizational commitment focuses on an individual's motivational desire – rather than need – to stay with the organization and has distinct positive correlations with both job performance and OCBs (Meyer, Stanley, Herscovitch, & Topolnytsky, Reference Meyer, Stanley, Herscovitch and Topolnytsky2002; Gellatly, Meyer, & Luchak, Reference Gellatly, Meyer and Luchak2006; Budihardjo, Reference Budihardjo2017).
Extending these collective relationships, Yukl (Reference Yukl2005) theorized a leader's use of power should have its most direct effect on subordinates' attitudes and behavioral compliance, and power should influence subordinate performance indirectly through these linkages. Thus, the ultimate impact of leader power use on follower performance may be attenuated or enhanced through its intermediate effect on followers' job satisfaction and organizational commitment.
Hypotheses
Building on this review of the power, job satisfaction, organizational commitment, and performance literatures, the current study hypothesized followers' job satisfaction and organizational commitment mediate the relationships found between their leaders' perceived power use and the followers' performance. Figure 1 illustrates these proposed relationships.
Hypothesis 1
Followers' job satisfaction mediates the negative effects of leaders' legitimate (H1a) and coercive (H1b) power use on the followers' in-role performance and OCB.
Hypothesis 2
Followers' job satisfaction mediates the positive effects of leaders' expert (H2a), referent (H2b), and reward (H2c) power use on the followers' in-role performance and OCB.
Hypothesis 3
Followers' organizational commitment mediates the negative effects of leaders' legitimate (H3a) and coercive (H3b) power use on the followers' in-role performance and OCB.
Hypothesis 4
Followers' organizational commitment mediates the positive effects of leaders' expert (H4a), referent (H4b), and reward (H4c) power use on the followers' in-role performance and OCB.
Figure 1. Job satisfaction and organizational commitment mediating the relationships between leader power use and follower performance.
Method
Sample
To test these hypotheses, the current study utilized a sample of 365 cadets at the U.S. Air Force Academy (referred to hereafter as the Academy). The Academy is a 4-year college and military training institution; upon graduation, cadets commission as officers in the U.S. Armed Forces. In addition to their undergraduate, academic course requirements, cadets also serve in formal job positions which provide a structure for military training and administrative duties in a traditional top-down, military hierarchy. Although the Academy is an academic institution, these functional job duties and formal structures differentiate this sample from that of a traditional-college-based convenience sample of undergraduate students. Sampling a population of military cadets was seen as an appropriate focus for assessing the power, attitudes, and performance relationships in the current study, since leadership and leader development are central to this population's success.
Among the 365 cadets in this sample, 72% were male and 28% were female. The sample's ethnicity was 71% Caucasian, 9.5% Hispanic, 6.6% Asian/Pacific Islander, 6% African-American/Black, 0.9% Native American, and 6% other. The mean age for the sample was 20.15 years. Finally, 55% of these cadets were freshmen, 1% were sophomores, 43% were juniors, and 1% were seniors. Since the subject pool was comprised of cadets in freshmen- and junior-level courses, this distribution was to be expected.
Procedure
A research subject pool coordinator, who was not involved directly with the study, e-mailed participation requests to 998 cadets enrolled in the two core psychology courses required for all academic majors at the Academy. In exchange for voluntarily completing an online survey, participants received extra credit in their respective courses as compensation for choosing to participate in the study. From this convenience sample of military cadets in the subject pool, we received 405 survey responses and excluded 32 participants due to grossly incomplete or missing data. In addition, the majority of participants required at least 30 min completing the survey; due to the length of the assessment and the median completion time, we excluded eight responses completed in under 8 min to avoid potentially erroneous or careless data (see Meade & Craig, Reference Meade and Craig2012). These adjustments garnered the final sample of 365 participants and represented a final response rate of 36.6%. This final sample size of 365 participants offered sufficient power to detect the focal effects of the current study with a degree of confidence at a 0.05 level of significance. All participants provided informed consent prior to being included in the study; details that might disclose the identity of these individuals were not retained.
Measures
Power use
Rahim's (Reference Rahim1988) 29-item Leader Power Inventory was adapted to assess followers' perceptions of their leaders' power use in the Academy environment. Prior to the study, all adapted items were reviewed by subject matter experts and pre-tested with Academy graduates not associated with the study for clarity and understanding to maintain nomological expectations (see Hinkin & Schriesheim, Reference Hinkin and Schriesheim1989). Using this modified scale, cadets rated their immediate cadet leader's power use on a Likert-type scale, ranging from ‘1’ (Very Rarely) to ‘5’ (Very Often). Expert power use (6 items) α = .80; sample items: ‘When a tough job comes up my leader uses technical “know how” to get it done’ and ‘Subordinates prefer to do what my leader suggests because my leader demonstrates high professional expertise.’ Referent power use (6 items) α = .82; sample items: ‘My leader makes subordinates want to identify with him (her)’ and ‘My leader makes subordinates want to develop a good interpersonal relationship with him (her).’ Reward power use (6 items) α = .89; sample items: ‘If a subordinate puts forth extra effort, my leader takes it into consideration to determine the subordinate's rewards’ and ‘My leader does not reward subordinates even if subordinates do their job well’ (reverse scored). Coercive power use (5 items) α = .86; sample items: ‘My leader keeps a subordinate from advancing if the subordinate's work is unsatisfactory’ and ‘My leader punishes a subordinate if the subordinate is late for duty.’ Legitimate power use (6 items) α = .70; sample items: ‘My leader relies on his (her) rank to influence subordinates to carry out his (her) instructions’ and ‘My leader uses his (her) position to gain subordinates support for my leader's policies.’
Job satisfaction
Evaluations of job satisfaction were based on cadets' attitudes toward their current jobs in their unit (squadron) and measured using items from Hackman and Oldham's (Reference Hackman and Oldham1975) Job Diagnostic Survey. Participants reported their agreement with three statements related to their current job on a 7-point, Likert-type scale ranging from ‘1’ (Strongly Disagree) to ‘7’ (Strongly Agree); sample items: ‘Generally speaking, I am very satisfied with my job’ and ‘I am generally satisfied with the kind of work I do in this job’ (α = .93).
Organizational commitment
The organizational commitment construct was assessed with adapted items from Meyer and Allen (Reference Meyer and Allen1997) to gage participants' affective attitudes indicating their motivational desire to stay with their squadrons. Participants rated their agreement with six statements on a 7-point, Likert-type scale ‘1’ (Strongly Disagree) to ‘7’ (Strongly Agree); sample items: ‘My squadron has a great deal of personal meaning for me’ and ‘I really feel as if my squadron's problems are my own’ (α = .91).
Job performance
In-role job performance was measured using each cadet's grade point average (GPA) and military performance average (MPA). A cadet's GPA is equivalent to those in traditional colleges and derived from academic course grades evaluated by the cadet's academic faculty (scaled from 0.00 to 4.00 [maximum]). A cadet's MPA is similar to, but determined independently from, academic GPA and is composed of objective and subjective evaluations of military job performance from multiple raters (scaled from 0.00 to 4.00 [maximum]). For example, subjective components are comprised of qualitative and behaviorally anchored ratings from the cadet's active-duty commanding officer, the cadet's immediate cadet supervisor, a senior cadet rater in the cadet's chain of command, and academic instructors' evaluations of the cadet's professional military conduct in the classroom. Objective rating components consist of elements such as the cadet's military knowledge test scores, room inspections, and personal appearance inspections. GPA and MPA correlate .39.
Organizational citizenship behaviors (OCB)
Followers' OCBs were measured using Lee and Allen's (Reference Lee and Allen2002) 16 items, which were also adapted slightly to include the Academy's nomenclature. Cadets indicated the frequency of behaviors they performed personally over the past semester in the accomplishment of their job in their unit on a Likert-type scale, ranging from ‘1’ (Never) to ‘7’ (Always); sample items: ‘Offer ideas to improve the functioning of the squadron’ and ‘Help others who have been absent’ (α = .90). Given the high alpha among these items, a single OCB factor was used, versus two discrete factors for individual- and organization-focused OCBs, to avoid unnecessarily complicating the analyses or creating redundant effects.
Analyses
The current study utilized IBM Statistical Package for the Social Sciences (SPSS) Version 23 and tested the elements outlined by Baron and Kenny (Reference Baron and Kenny1986) to assess the mediated relationships proposed in the hypotheses: (1) a significant relationship between the independent variable and mediator; (2) significance between the independent and dependent variable; (3) significance between the mediator and dependent variable; and (4) controlling for the influence of the mediator, determine whether the original relationship between the independent and dependent variables is reduced to non-significance or becomes smaller, which provides evidence for full or partial mediation. In addition, the current study employed Preacher and Hayes's (Reference Preacher and Hayes2008) bootstrapping approach for multiple mediator models to examine the significance of the IE through the two potential mediators (i.e., job satisfaction and organizational commitment). Compared to other commonly used mediation tests, such as the Sobel (Reference Sobel and Leinhart1982) test, the bootstrapping approach has more power and does not rely on the assumption total and IE are normally distributed (see Preacher & Hayes, Reference Preacher and Hayes2004, Reference Preacher and Hayes2008). These bootstrapped analyses utilized 5,000 random resamples with replacement from the full sample to construct bias-corrected confidence estimates. The specific IE of the independent variable on the dependent variable via the mediators is statistically significant if the IE does not include zero at the 95% confidence interval (CI; Preacher & Hayes, Reference Preacher and Hayes2008).
Results
The descriptive statistics, reliability estimates, and zero-order correlations for all variables included in the current study are given in Table 1. Leaders' perceived use of expert power was correlated highly with their perceived use of referent (r = .70, p < .01) and reward (r = .66, p < .01) power. Reward power use was also correlated highly with followers' perceptions of their leaders' referent power use (r = .62, p < .01). These relationships suggest any concomitant relationships linked to these power bases are potentially replications of the same effect.
Table 1. Descriptive statistics, reliability estimates, and correlations for all variables
Note. N = 365. GPA, grade point average; MPA, military performance average; OCB, organizational citizenship behavior. Internal consistency statistics (Cronbach's alphas) appear in bold along the diagonal.
*p < .05; **p < .01; ***p < .001.
Results indicated leaders' perceived use of legitimate (r = −.12, p < .05) and coercive (r = −.11, p < .05) power were correlated negatively with followers' MPA. However, expert (r = −.01, p > .05), referent (r = −.03, p > .05), and reward (r = −.05, p > .05) power use were not significantly related to MPA. None of the five power bases (expert [r = −.02, p > .05], referent [r = −.05, p > .05], reward [r = .00, p > .05], coercive [r = −.07, p > .05], or legitimate [r = −.01, p > .05] power) were significantly related to followers' GPA. Leaders' perceived use of expert (r = .17, p < .01), referent (r = .12, p < .05), and reward (r = .15, p < .01) power were related positively to followers' OCB. Legitimate (r = .06, p > .05) and coercive (r = .06, p > .05) power use, however, were not related to followers' OCB at a statistically significant level.
Table 2, Model 1 shows leaders' perceived use of expert (β = .20, p < .001), referent (β = .16, p < .01), and reward (β = .22, p < .001) power were related positively and significantly to followers' job satisfaction. Since job satisfaction was not related significantly to legitimate (β = −.02, p > .05) or coercive (β = .05, p > .05) power, there is no statistically significant evidence to support a mediated path from these powers to the performance variables through job satisfaction. Therefore, Hypotheses 1a and 1b, proposing job satisfaction would mediate the relationships between these power bases and followers' in-role performance and OCB, were not supported. Model 2 shows job satisfaction was significantly related to GPA (β = .11, p < .05) and OCB (β = .39, p < .001), but not MPA (β = .08, p > .05). Model 3 illustrates expert (β = .17, p < .05), referent (β = .12, p < .05), and reward (β = .15, p < .01) power correlated positively with followers' OCB, but not GPA (β = .09, p > .05; β = −.13, p > .05; and β = .05, p > .05 respectively) or MPA (β = −.05, p > .05; β = −.03, p > .05; and β = −.05, p > .05 respectively). Finally, support for the mediation hypotheses exists in situations where initially significant relationships found between the independent variables and dependent variables are lessened or eliminated once the mediating variable is added to the regression equation. As Model 4 shows, after job satisfaction is added to the regression model, the relationships between expert (β = .10, p > .05), referent (β = −.01, p > .05), and reward (β = .02, p > .05) power and followers' OCB were all reduced to non-significance. In addition, Preacher and Hayes's (Reference Preacher and Hayes2008) bootstrapping analyses for multiple mediators indicated the specific IE of expert (IE = .05; SE = .02; 95% CI = [.02 to .10]), referent (IE = .04; SE = .02; 95% CI = [.01 to .08]), and reward (IE = .05; SE = .02; 95% CI = [.02 to .10]) power on OCB, through job satisfaction, were all statistically significant at the 95% CI. Thus, Hypotheses 2a, 2b, and 2c, proposing job satisfaction mediates the positive relationships between these three powers and OCB, were supported; however, there were no statistically significant direct or indirect relationships between leaders' perceived use of these powers and followers' in-role performance (i.e., GPA or MPA).
Table 2. Regression analyses for the mediated effects of job satisfaction
Note. N = 365. Values are standardized coefficients. GPA, grade point average; MPA, military performance average; OCB, organizational citizenship behavior.
*p < .05; ** p < .01; *** p < .001.
Table 3, Model 1 illustrates leaders' perceived use of expert (β = .11, p < .05), referent (β = .13, p < .05), and reward (β = .15, p < .01) power, were related positively to followers' organizational commitment. Since organizational commitment was not related significantly to legitimate (β = .01, p > .05) or coercive (β = .00, p > .05) power, Hypotheses 3a and 3b, asserting organizational commitment would mediate the relationships between these two powers and followers' in-role performance and OCB were not supported. Model 2 indicates organizational commitment was related significantly to followers' MPA (β = .17, p < .01) and OCB (β = .55, p < .001), but not GPA (β = .07, p > .05). Model 3 shows leaders' perceived use of expert (β = .17, p < .05), referent (β = .12, p < .05), and reward (β = .15, p < .01) power were related significantly to followers' OCB, but not GPA or MPA (as discussed previously). Model 4 shows, once organizational commitment is controlled for, the significant, positive relationship initially found between expert power use and followers' OCB was reduced slightly (β = .14, p < .05), but remained statistically significant (bootstrapped direct effect = .13; SE = .05; 95% CI = [.02 to .24]). This indicated the relationship between leaders' expert power use and followers' OCB was partially mediated by organizational commitment (bootstrapped IE = .06; SE = .03; 95% CI = [.01 to .13]), supporting Hypothesis 4a. In addition, controlling for organizational commitment reduced the previously significant relationships between both referent (β = −.04, p > .05) and reward (β = .00, p > .05) power use and followers' OCB to non-significance. Preacher and Hayes's (Reference Preacher and Hayes2008) bootstrapping analyses for multiple mediators further indicated the specific IE of referent (IE = .06; SE = .03; 95% CI = [.01 to .12]) and reward (IE = .08; SE = .03; 95% CI = [.03 to .14]) power on OCB through organizational commitment were both statistically significant at the 95% CI. Therefore, Hypotheses 4b and 4c, predicting organizational commitment mediates the positive relationships between these powers and OCB, were supported. However, there were no statistically significant direct or indirect relationships between leaders' perceived use of these three powers and followers' GPA or MPA.
Table 3. Regression analyses for the mediated effects of organizational commitment
Note. N = 365. Values are standardized coefficients. GPA, grade point average; MPA, military performance average; OCB, organizational citizenship behavior.
* p < .05; **p < .01; ***p < .001 .
Finally, contrasting the IE of expert (IE = −.01; SE = .03; 95% CI = [−.08 to .05]), referent (IE = −.02; SE = .03; 95% CI = [−.08 to .03]), and reward (IE = −.03; SE = .03; 95% CI = [−.10 to .03]) power on followers' OCB, through both job satisfaction and organizational commitment, showed no statistically significant difference in magnitude between the two mediating effects (Preacher & Hayes, Reference Preacher and Hayes2008). That is, since zero is contained in each of these intervals, distinct IE through job satisfaction and organizational commitment cannot be distinguished at the 95% confidence level. Therefore, while both job satisfaction and organizational commitment mediate the specific IE of expert, referent, and reward power on followers' OCB, neither mediator contributes to these effects more than the other.
Discussion
The current study's analyses and reported results met its stated purpose by empirically identifying the relationships among leader power use, follower attitudes, and performance. First, the current study addressed gaps identified in existing power research by empirically identifying practical intermediate variables affecting the power–performance relationship. Although previous studies have focused on person-based powers' positive, direct relationships and position-based powers' often negative, direct relationships with a range of outcomes (Podsakoff & Schriesheim, Reference Podsakoff and Schriesheim1985; Carson, Carson, & Roe, Reference Carson, Carson and Roe1993), one form of power use is not necessarily universally better or worse than other forms. The relationship between power use and target outcomes (e.g., performance) is often contingent upon the extent the power use affects other key intermediate variables (in this case, the targets' attitudes).
As predicted, job satisfaction and organizational commitment mediated the relationships between followers' perceptions of both referent and expert power use, and the followers' OCB. Consequently, the positive relationship between personal power use and follower OCB is better explained by these attitudinal variables. Followers may feel more satisfied with their jobs and committed to an organization when leaders engage power dynamics based on respect and admiration; these positive attitudes are then associated with higher OCBs. Similarly, followers may feel more satisfied with their jobs and committed to an organization where leaders exhibit expertise; in which case, these attitudes are also positively related to the followers' OCBs.
Likewise, followers' job satisfaction and organizational commitment fully mediated the relationship between reward power use and OCBs. This implies that, although the use of positional power does not necessarily promote followers' extra-role performance directly, followers who believe their efforts will be rewarded appropriately may feel more satisfied with their job; this satisfaction is then associated with greater levels of OCB. Equally, when leaders demonstrate a proclivity toward rewarding behavior in a thoughtful and meaningful way, followers may feel committed to the organization and its success to ensure they receive these rewards; this commitment is then positively related to the followers' OCB. Although reward power is based on position, Jain, Giga, and Cooper (Reference Jain, Giga and Cooper2011) argued it is a positive, pro-social power and found it to be positively related to OCBs. The current study provides evidence to suggest this positive relationship is better understood by accounting for the IE of reward power on followers' job satisfaction and commitment.
Leaders' perceived use of legitimate and coercive power was both negatively related to followers' in-role performance in terms of MPA; however, these relationships were not mediated by job satisfaction or organizational commitment. This could suggest, while the use of legitimate or coercive power may be negatively related to in-role performance outcomes, this power use does not necessarily diminish followers' organizational commitment or personal job satisfaction. Still, these impersonal, utilitarian approaches to power (Etzioni, Reference Etzioni1975; Franklin, Reference Franklin1975) are unlikely to promote the trusting and supportive relationships found in person-based power dynamics or the use of fair contingent rewards, which are more central to promoting performance beyond the scope of an individual's formal job role (Podsakoff, MacKenzie, Moorman, & Fetter, Reference Podsakoff, MacKenzie, Moorman and Fetter1990; MacKenzie, Podsakoff, & Rich, Reference MacKenzie, Podsakoff and Rich2001). Since individuals' attitudes shape their distinctive behavioral intentions (Ajzen & Fishbein, Reference Ajzen and Fishbein1980), these findings help to explain some of the inconsistent performance outcomes associated with power use noted in previous reviews of the literature.
The current study examined some of the fundamental power mechanisms often overlooked – or taken for granted – in contemporary leadership approaches (Pfeffer, Reference Pfeffer1992; Collinson & Tourish, Reference Collinson and Tourish2015). Examining the direct and indirect relationships associated with French and Raven's (Reference French, Raven and Cartwright1959) power bases provides deeper insight into modern leadership methods. Contemporary leadership approaches employ varying degrees of power; their success depends on how this power use is perceived and accepted by followers in a given situation. Understanding the cumulative relationships leader power use has with proximal follower attitudes and distal performance behaviors provides a deeper understanding of the broader leadership dynamic.
Results in this study showed followers' job satisfaction and organizational commitment played a more prominent mediating role among the relationships found between leaders' perceived power use and followers' OCBs, and these job attitudes had less of an impact on followers' in-role performance. These findings support Organ's (Reference Organ1977, Reference Organ1988b) assertion that the relationships between job attitudes and extra-role performance should be stronger than those found between attitudes and in-role performance, since followers have more discretionary control over their OCBs than their mandated in-role performance requirements. Furthermore, these findings also support Bateman and Organ's (Reference Bateman and Organ1983), and later Organ and Ryan's (Reference Organ and Ryan1995), contention that work attitudes have a greater impact on OCBs than in-role performance, since OCBs are less constrained by an individual's knowledge, skills, and abilities. Relatedly, there were no significant direct relationships between any forms of leader power use and followers' GPA. This may be due to academic GPA being more dependent on isolated individual characteristics, such as cognitive ability, which are more fixed and not as susceptible to the influence of power use or leadership styles. Followers' MPA and OCB may be affected more by their leaders' influence and predicated on social interactions within the organization; therefore, MPA and OCB may not be as constrained, with respect to basic ability, as GPA. Managers who seek to use their power to drive performance may have limited or no success with complex tasks and behaviors strongly dependent on cognitive abilities (Hunter & Hunter, Reference Hunter and Hunter1984).
The current study answered researchers' rarely headed call (Podsakoff & Schriesheim, Reference Podsakoff and Schriesheim1985; Schriesheim, Hinkin, & Podsakoff, Reference Schriesheim, Hinkin and Podsakoff1991; Rahim et al., Reference Rahim, Antonioni, Krumov and Illieva2000) for better scale measurement of French and Raven's (Reference French, Raven and Cartwright1959) social power bases by utilizing a validated, multi-item, Likert-type measure to assess followers' perceptions of leader power: The Rahim Leader Power Inventory (RLPI; Rahim, Reference Rahim1988). The RLPI was selected over other widely used ipsative ranking scales (e.g., Bachman, Smith, & Slesinger, Reference Bachman, Smith and Slesinger1966; Student, Reference Student1968) or single-item measures (e.g., Busch, Reference Busch1980; Cobb, Reference Cobb1980). As discussed previously, these scales may have severe measurement limitations, which have muddied the waters of power research and contributed to inconsistent outcomes and conclusions. For example, ipsative ranking scales asking subordinates to stratify the importance of power, based on which form of power led to their compliance, do not allow independent comparisons of these power bases (Podsakoff & Schriesheim, Reference Podsakoff and Schriesheim1985; Johnson, Wood, & Blinkhorn, Reference Johnson, Wood and Blinkhorn1988). Other researchers (e.g., Nunnally, Reference Nunnally1978; Schriesheim, Hinkin, & Podsakoff, Reference Schriesheim, Hinkin and Podsakoff1991) have also cautioned single-item instruments are generally less reliable than multi-item instruments since these single-item instruments do not allow for internal consistency reliability testing using dependable validation methods, such as factor analysis, and may weaken or obscure relationships among variables. Therefore, the RLPI employed in the current study was deemed a more accurate measure of perceived power use and allowed for more appropriate examinations of power's relationship to performance outcomes.
Finally, this study contributes a unique perspective by exploring a view of leadership and power use in an interesting context: a military organization. The possession of certain formal powers in military organizations is clearly defined (e.g., positional power is based on military rank), but there may still be significant variance among each leader's reliance on these power types. This exemplifies and extends Hinkin & Schriesheim's (Reference Hinkin and Schriesheim1989), as well as Barbuto's (Reference Barbuto2000), theorization that the power to influence others relies not only on a leader's formal position and personal abilities, but also on followers' belief they have the authority or capability to influence them. This understanding offers insight into nuanced power dynamics within the leadership process and particular value to organizations with similar organizational dynamics, cultural characteristics, or leader–follower paradigms.
Practical implications
When leaders exercise power, regardless of the type of power it may be, they are entering into a complicated process that goes well beyond a simple action–reaction (i.e., power exerted–performance impacted) exchange. Effective leaders are those who understand the dynamic relationship between their behavior (action), the potential response by followers to the leader's behavior (intentions), the specific perspectives of their followers (attitudes), and the outcome they wish to achieve (reaction). This study helps us better understand the importance of intervening variables between action and reaction and can help leaders more broadly inventory the terrain they traverse in ‘…getting someone else to do something that you want done.’ Followers clearly evaluate the type of power their leaders employ; the extent this power use promotes or deters critical attitudes – including job satisfaction and organizational commitment – and the roles these attitudes play in significantly shaping followers' behavioral intentions and performance. Successful leaders are those who can fully appreciate and navigate the complexities involved in the exercise of power through this more complicated leadership labyrinth.
Whether a leader is relying on personal leadership strengths, position, or influencing a subordinate through rewards or potential penalties, it behooves the leader to simultaneously consider both the desired outcome (e.g., performance) and the impact of the approach on the follower (e.g., the follower's attitude). Thinking through this process and applying it to an academic setting, invoking deadlines with a graduate student in terms of the positive outcome of graduating versus losing funding as a result of falling behind may not address the key attitudes driving or surrounding those events. Certainly, deadlines must be made clear, but tying the discussion of deadlines with the potential satisfaction of accomplishing a long sought-after goal can only enhance the likelihood that use of position power, along with appeals for increased satisfaction, will have an even more positive outcome. Within an institution like the Academy, simply giving orders may be efficient in terms of time, but not necessarily effective in terms of overall performance because of the lack of focus on the underlying dynamics between the leader and the follower. Here, additional attention to the relationship between accomplishment and achieving personal satisfaction from reaching a goal should enhance performance above and beyond simply giving an order. In an even broader application, managers who address employee satisfaction and commitment may expose limitations or potentially strengthen more favorable conditions surrounding their use of formal authority or penalties to drive performance. Managers who are attentive to these underlying attitudes may better reinforce and capitalize on subordinates' willingness to engage in helping behaviors beyond their job descriptions when these managers leverage incentives, expert knowledge, or personal appeal to influence them.
Overall, organizational leaders who rely on certain personal or positional powers to generate performance may only be effective to the extent these approaches increase or decrease their followers' job satisfaction or organizational commitment. Based on the findings in this study it is clear that simultaneously using some form of power can help to build performance to the degree the leader exercising that power pays close attention to how the subordinate perceives the link between performance and intervening attitudes such as job satisfaction and organizational commitment.
Limitations
When designing and conducting any research, there are methodological decisions and tradeoffs, which introduce limitations that must be considered. First, the design of the current study is cross-sectional and the relationships examined are correlational; causal inferences among the variables of interest must be interpreted conservatively.
Second, although this study utilized multi-rater-based GPAs and MPAs as measures of in-role performance, followers' personal attitudes, perceptions of power use, and OCBs were self-rated at a single point of time. Therefore, these data may be susceptible to common method bias, wherein the observed variance is a result of the collection methodology rather than the measurement items (Podsakoff, MacKenzie, Lee, & Podsakoff, Reference Podsakoff, MacKenzie, Lee and Podsakoff2003). Harman's single-factor test was used to detect the pervasiveness of common method variance in the current study. First, we conducted an exploratory factor analysis of all items included in the current study's 10 scales (i.e., expert, referent, reward, coercive, and legitimate power; job satisfaction; organizational commitment; OCB; MPA; and GPA). Results indicated more than one factor emerged and the general factor explained only 17.36% of the variance. Second, a confirmatory factor analysis of these scales indicated the goodness-of-fit indices of the 10-factor model had a better fit to the data (χ2 = 802.73, df = 656, N = 365, p < .001, CFI = .96, TLI = .96, RMSEA = .03) than did the one-factor model (χ2 = 4323.11, df = 1034, N = 365, p < .001, CFI = .76, TLI = .75, RMSEA = .09). These analyses provided evidence to suggest common method variance is not a serious concern in the current study and did not likely affect the results.
Finally, the uniqueness of the current study's sample, while interesting, may present some concerns over the generalizability of these findings. This field sample was based on a population of military cadets in a western culture with a mean age of 20.15 years (i.e., Millennials). While this sample represents a functioning, real-world organization operating under a top-down organizational hierarchy, the generalization of the relationships found among the current study's variables to other age groups, cultures, countries, or organizations with less clearly defined lines of authority, or other unique characteristics, may potentially be limited. Still, the overall implications and relationships found in the current study are believed to be informative and relevant. Counter to the common stereotype implying military leadership is simply superiors issuing orders to subordinates who automatically comply with and support them, the current study showed military leaders' use of position-based power was negatively related to followers' job performance, while other, more nuanced, power use was positively related to subordinate outcomes. This contributes to the growing assertion by many leadership researchers that military leadership employs competencies and approaches that are valuable and applicable to a broad range of organizations and contexts (Wong, Bliese, & McGurk, Reference Wong, Bliese and McGurk2003).
Future research
Future research should build on the current study's findings by examining other types of organizations (e.g., civilian organizations with flatter organizational structures, lower power distances [Hofstede, Reference Hofstede1980], or broader employee age ranges) to identify and further investigate potential boundary conditions associated with power use, attitudes, perceptions, and behavioral outcomes. These studies should incorporate observational field procedures examining leaders in their natural work environments directly and recording their behaviors clearly, through either empirically derived categories or a narrative description of their observed behaviors. Researchers could couple these observations with the leaders' self-reported assessments of their own power use to detect differences in followers' perceptions versus the leaders' intended power use. Kumar, Stern, and Anderson (Reference Kumar, Stern and Anderson1993) argued the use of multiple key informants increases reliability and validity, and this methodology would enrich analyses through the triangulation of different methods (i.e., a combination of qualitative and quantitative data).
Next, the current study highlights the importance of followers in the leadership dynamic by demonstrating the intermediary effects of followers' attitudes in the power–performance relationship. Judge and colleagues noted a range of distinctive, personal characteristics (e.g., personality, self-esteem, and general self-efficacy) are key determinants of an individual's job attitude (Judge & Bono, Reference Judge and Bono2001; Judge, Bono, Ilies, & Gerhardt, Reference Judge, Bono, Ilies and Gerhardt2002); these determinants could inform an even broader model of the leadership dynamic. Beyond these characteristics, social psychological research suggests other follower-centric considerations (e.g., the follower's efficacy toward the behavior, perceived behavioral control, anticipated behavioral consequences, personal habits, and past experiences) also shape his or her behavioral intentions and predict behavior (Ajzen & Fishbein, Reference Ajzen and Fishbein1980; Ajzen, Reference Ajzen2002; Fishbein & Ajzen, Reference Fishbein and Ajzen2010).
Furthermore, Lord (Reference Lord, Staw and Cummings1985) hypothesized a follower's willingness to accept a leader's influence may be governed by the extent the follower's personal leadership beliefs align with his or her perception of the leader. The current study showed leaders' use of expert, referent, and reward power have significant, positive relationships with both followers' job satisfaction and organizational commitment, which were also reflected in positive performance outcomes. This could imply these types of power use were welcomed and consistent with the followers' Implicit Leadership Theories (Lord, Foti, & De Vader, Reference Lord, Foti and De Vader1984; Offermann, Kennedy, & Wirtz, Reference Offermann, Kennedy and Wirtz1994). However, followers with different expectations or beliefs about leader behavior might, for example, feel less satisfied or form other attitudes about the leader or the leadership dynamic, ultimately yielding different performance outcomes.
Therefore, followers' reactions to their leaders' power use may also be shaped by the followers' own characteristics, beliefs, or perceptions. For example, Reiley and Jacobs (Reference Reiley and Jacobs2016) found followers who perceived their leaders to be ethical exhibited more OCBs when these leaders utilized reward, referent, and expert power; conversely, when followers perceived their leaders to be less ethical, they engaged in fewer contextual performance behaviors when these leaders attempted to influence them using rewards.
Finally, Gustafson and Mumford (Reference Gustafson and Mumford1995) found individuals' personalities were better predictors of job attitudes and performance when situational factors were taken into account. This also aligns with a broader, leader–follower–situational perspective of leadership as a complex and interactive process. While followers' perceptions and attitudes play a role in both task performance and their willingness to engage in supportive contextual behaviors, the situational context of the job at hand may affect these elements dramatically. For example, Faiz (Reference Faiz2013) found managers' use of power and its effects on employees' job satisfaction differed between the contexts of the private sector versus the public sector. Fishbein and Ajzen (Reference Fishbein and Ajzen2010) suggested injunctive norms (i.e., what people in the organization expect an individual to do in a situation), descriptive norms (i.e., perceptions of what other people actually do), time, situational capacity, and other environmental factors play a role in predicting behavior. Therefore, future research should investigate how these situational factors interplay with followers' individual differences and leader power use to drive performance and other critical organizational outcomes.
Conclusion
In closing, like President Eisenhower's leadership philosophy, the power to influence performance lies, in part, in the effects leaders have on followers' attitudes toward the job. This study provided empirical evidence to suggest job satisfaction and organizational commitment explain many of the relationships found between leader power use and follower performance.
Author ORCIDs
Peter J. Reiley, 0000-0002-9144-9406.
Acknowledgements
The authors would like to thank Weichun Zhu for his contributions to earlier drafts of this manuscript. The views expressed in this article are those of the authors and do not necessarily reflect those of the U.S. Department of Defense or the United States Air Force.
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
