Academics and practitioners have shown increased interest in a caring style of leadership (Peterson, Galvin, & Lange, Reference Peterson, Galvin and Lange2012; Hunter et al., Reference Hunter, Mitchell, Neubert, Perry, Witt, Penney and Weinberger2013; Udani & Lorenzo-Molo, Reference Udani and Lorenzo-Molo2013), a growing need for more people-centered management (LidenReference Liden,Wayne, Liao, & Reference Liden, Wayne, Liao and MeuserMeuser, 2014b; van Dierendonck, Reference van Dierendonck2011), and a concern for the success of all stakeholders (Walumbwa, Hartnell, & Oke, Reference Walumbwa, Hartnell and Oke2010; Sun, Reference Sun2013), as these factors contribute to organizations’ success in today’s marketplace. Successful leadership requires that followers be considered before leaders. However, current leadership theories mainly adopt a leader-first approach (Greenleaf, Reference Greenleaf2008, Reference Greenleaf1977). In these theories, the leader is the main player in the organization, and the followers simply ‘follow the leader.’ However, in today’s business environment, leaders need to serve their followers so that followers can develop into important players who serve their organization and who can achieve better results than a single leader (the main player) would be able to produce individually. Therefore, there is a strong need for a new type of leadership to address the academic and practitioner interest in follower-focused leadership in order to broadly resolve the stubborn problems and unsolved puzzles in our understanding of how leadership drives discretionary attitudes and behaviors (e.g., Berry, Parasuraman, & Zeithaml, Reference Berry, Parasuraman and Zeithaml1994; Ehrhart, Reference Ehrhart2004; Barbuto & Wheeler, Reference Barbuto and Wheeler2006; Walumbwa, Hartnell, & Oke, Reference Walumbwa, Hartnell and Oke2010; van Dierendonck, Reference van Dierendonck2011; LidenReference Liden,Panaccio, Meuser, Hu, & Reference Liden, Panaccio, Meuser, Hu and WayneWayne, 2014a; Reference Liden, Wayne, Liao and MeuserLiden et al., 2014b).
Servant leadership theory is uniquely positioned to address this need. Greenleaf (Reference Greenleaf1977: 66) describes servant leadership as follows:
The servant-leader is servant first—… It begins with the natural feeling that one wants to serve, to serve first. Then conscious choice brings one to aspire to lead. He is sharply different from the person who is leader first, perhaps because of the need to assuage an unusual power drive or to acquire material possessions. For such it will be a later choice to serve—after leadership is established. The leader-first and the servant-first are two extreme types. Between them there are shadings and blends that are part of the infinite variety of human nature.
The term servant leadership was coined by Greenleaf (Reference Greenleaf1977) to emphasize that leadership is enacted in a for-profit organization by meeting the highest priority needs of employees first. As easily recognized from the above anecdote, the servant-first nature of servant leadership distinguishes this concept from other theories related to leadership (Graham, Reference Graham1991; Ehrhart, Reference Ehrhart2004; Judge & Piccolo, Reference Judge and Piccolo2004; Ilies, Nahrgang, & Morgeson, Reference Ilies, Nahrgang and Morgeson2007; LidenReference Liden,Wayne, Zhao, & Henderson, Reference Liden, Wayne, Zhao and Henderson2008; Walumbwa, Hartnell, & Oke, Reference Walumbwa, Hartnell and Oke2010; Hu & LidenReference Liden,Reference Hu and Liden2011; Reference Liden, Wayne, Liao and MeuserLiden et al., 2014b). Servant leadership challenges those who want to be servant leaders because it requires them to set aside their personal desires, needs, and/or wishes to first meet the highest priority needs of their followers (or employees).
Previous studies on servant leadership have generally linked leadership to employees’ workplace attitudes and behaviors (e.g., Ehrhart, Reference Ehrhart2004; Janssen & van Yperen, Reference Janssen and van Yperen2004; Judge & Piccolo, Reference Judge and Piccolo2004; Barbuto & Wheeler, Reference Barbuto and Wheeler2006; Ilies, Nahrgang, & Morgeson, Reference Ilies, Nahrgang and Morgeson2007; Gonzalez & Garazo, Reference Gonzalez and Garazo2008; Walumbwa, Hartnell, & Oke, Reference Walumbwa, Hartnell and Oke2010; van Dierendonck, Reference van Dierendonck2011; Zhang, Kwan, Everett, & Jian, Reference Zhang, Kwan, Everett and Jian2012; Hunter et al., Reference Hunter, Mitchell, Neubert, Perry, Witt, Penney and Weinberger2013; Udani & Lorenzo-Molo, Reference Udani and Lorenzo-Molo2013; Reference Liden, Wayne, Liao and MeuserLiden et al., 2014b) as well as the workplace’s psychological climate (e.g., Schneider, Ehrhart, Mayer, Saltz, & Niles-Jolly, Reference Schneider, Ehrhart, Mayer, Saltz and Niles-Jolly2005). However, servant leadership studies on for-profit organizations consider servant leadership mainly from unit-, group-, and team-level constructs to investigate its direct effect on employee attitudes and behaviors and its effects through the mediating processes of climate and culture (e.g., Ehrhart, Reference Ehrhart2004; Liden, Wayne, Zhao, &Henderson, Reference Liden, Wayne, Zhao and Henderson2008; Walumbwa, Hartnell, & Oke, Reference Walumbwa, Hartnell and Oke2010; Hu & LidenReference Liden,Reference Hu and Liden2011; Reference Liden, Wayne, Liao and MeuserLiden et al., 2014b). We know that servant leadership focuses on meeting the needs of individual followers and that a servant leader’s relationship with her/his followers is one to one (Reference Liden, Panaccio, Meuser, Hu and WayneLiden et al., 2014a, in press-b). However, the effects of individual-level servant leadership on diverse employee attitudes and behaviors have not been comprehensively studied in a for-profit organizational context (see Parris & Peachey, Reference Parris and Peachey2013 for the most current and extended review of the servant leadership literature in organizational contexts), although there is a strong need for individual-level analysis on the effects of servant leadership (Manz & Sims, Reference Manz and Sims1987; Davis, Schoorman, & Donaldson, Reference Davis, Schoorman and Donaldson1997; Kozlowski & Klein, Reference Kozlowski and Klein2000; Barbuto & Wheeler, Reference Barbuto and Wheeler2006; Reference LidenLidenet al., Reference Liden, Wayne, Zhao and Henderson2008, 2014a; van Dierendonck, Reference van Dierendonck2011).
We develop a theoretical model to address these issues and to enhance our understanding of how leadership drives discretionary attitudes and behaviors. Figure 1 shows our model, which considers the individual level of analysis for all of the constructs. Accordingly, servant leadership is thought to influence psychological climate (e.g., van Dierendonck, Reference van Dierendonck2011), job satisfaction (e.g., Barbuto & Wheeler, Reference Barbuto and Wheeler2006; Gonzalez & Garazo, Reference Gonzalez and Garazo2008; Mayer, Bardes, & Piccolo, Reference Mayer, Bardes and Piccolo2008; van Dierendonck & Nuijten, Reference van Dierendonck and Nuijten2011), and organizational citizenship behavior (OCB) (e.g., Barbuto & Wheeler, Reference Barbuto and Wheeler2006; Liden, Reference Liden2013). We also link servant leadership to job satisfaction through psychological climate for reasons that we discuss in detail in the related section below. Given that the most important component of servant leadership is to serve first (Greenleaf, Reference Greenleaf1977), we articulate that servant leadership is enacted in a for-profit organization by a servant leader showing managerial competence and consistency in helping followers solve the various problems that they confront daily, by satisfying followers’ workplace needs so that followers commit themselves fully to their jobs in the organization, and by creating a climate where cooperation and coordination is valued among followers when they perform their jobs. We address three research questions: (1) How does servant leadership affect job satisfaction, psychological climate, and OCB? (2) Does psychological climate mediate a possible relationship between servant leadership and job satisfaction? (3) Which variable(s) among servant leadership, job satisfaction, and psychological climate best predict(s) OCB?
Research model
THEORY AND HYPOTHESIS DEVELOPMENT
Servant leadership and OCB
In addressing our research questions, we draw on social exchange theory (SET) (Blau, Reference Blau1964). SET greatly emphasizes reciprocity as the best-known exchange rule in social interactions (Gouldner, Reference Gouldner1960; Blau, Reference Blau1964; Cropanzano & Mitchell, Reference Cropanzano and Mitchell2005). Reciprocity is defined as ‘the pattern of exchange through which the mutual dependence of people, brought about by the division of labor, is realized. Reciprocity, therefore, is a mutually gratifying pattern of exchanging goods and services’ (Gouldner, Reference Gouldner1960: 169, 170). Social exchange is defined as the ‘voluntary actions of individuals that are motivated by the returns they are expected to bring and typically do in fact bring from others’ (Blau, Reference Blau1964: 91). We argue that servant leadership leads to OCB through the SET tenet that ‘effective management is impossible within the confines of formal authority alone …. An alternative strategy a manager can use is to provide services to subordinates that obligate them to comply with his directives’ (Blau, Reference Blau1964: 206). Because servant leaders do not use coercive power in their interactions with followers, a social exchange relationship in the form of services provided to followers is used by servant leaders to obligate their followers to get things done. After the followers receive services from their servant leaders, they feel obligated to reciprocate those services by practicing servant leadership behaviors that benefit their followers. Therefore, the followers’ actions are contingent upon the servant leader’s behavior. Of course, there is no explicit bargaining in the social exchange. The social exchange relationship logic applies through reciprocation: because my leader served me to learn my job, I will serve my subordinate to learn her/his job, too. The objective of the servant leader is to serve the highest priority needs of the followers, and this tenet of servant leadership depends on the following statement from Greenleaf (Reference Greenleaf1977: 7):
The difference manifests itself in the care taken by the servant-first to make sure that other people’s highest priority needs are being served. The best test, and difficult to administer, is: do those served grow as persons; do they, while being served, become healthier, wiser, freer, more autonomous, and more likely themselves to become servants? And, what is the effect on the least privileged in society; will he benefit, or, at least, will he not be further deprived?
Barbuto and Wheeler (Reference Barbuto and Wheeler2006) describe servant leadership as consisting of dimensions such as altruistic calling, emotional leadership, wisdom, persuasive mapping, and organizational stewardship. Leaders’ level of desire to serve the highest priority interests of their followers instead of their own represents their altruistic calling. The level of enthusiasm that leaders bring to committing themselves and performing competently to provide their followers with relief from any hardship refers to their emotional leadership. Wisdom is described as leaders’ level of scholarship and their consciousness of the events that are happening or that are about to happen in their environments. The level of mastery that leaders display in convincing their followers to attain organizational aims denotes their persuasive mapping. Organizational stewardship refers to the degree to which a leader considers the entire organization to be a family that exists to develop the society in which it resides. A leader manifests servant leadership by exerting these different subdimensions to serve her/his followers’ top priority needs (Greenleaf, Reference Greenleaf2008, Reference Greenleaf1977; van Dierendonck, Reference van Dierendonck2011; Zhang et al., Reference Zhang, Kwan, Everett and Jian2012).
Because SET is appropriate for explaining the exchange between a leader and her/his followers, many previous studies have used SET to explain the possible effects of leadership (e.g., leader-member exchange) on employee attitudes (e.g., job satisfaction) and behaviors (e.g., OCB) (Settoon, Bennett, & LidenReference Liden,Reference Settoon, Bennett and Liden1996; Wayne, Shore, & LidenReference Liden,Reference Wayne, Shore and Liden1997; Masterson, Lewis, Goldman, & Taylor, Reference Masterson, Lewis, Goldman and Taylor2000; Erdogan & LidenReference Liden,Reference Erdogan and Liden2002). Given the definition of reciprocity and the intention of servant leadership, when a servant leader’s subordinates continue to receive desired services from their servant leaders (e.g., through altruistic calling, emotional leadership, wisdom, persuasive mapping, and organizational citizenship), they discharge their obligation because their leader served them earlier (Gouldner, Reference Gouldner1960). For example, servant leaders’ practice of emotional leadership leads their followers to develop individual behaviors that are proactive and that seek to avert undesirable occurrences in the workplace. Previous studies provide support for the discharge of reciprocity argument (e.g., Ehrhart, Reference Ehrhart2004; Walumbwa, Hartnell, & Oke, Reference Walumbwa, Hartnell and Oke2010; Liden, Reference Liden2013; Reference Liden, Wayne, Liao and MeuserLiden et al., 2014b). For example, Reference LidenLiden(2013) emphasizes that ‘helping to explain how servant leadership influences outcomes, such as helping citizenship behaviors, creativity, and community service behaviors is the finding that servant leadership cultivates within followers a desire to fulfill one’s inner potential by seeking opportunities that help to develop skills and abilities.’ Organ (Reference Organ1997: 95) defines OCB as ‘performance that supports the social and psychological environment in which task performance takes place.’ In the present study, we adopt the concepts of Vey and Campbell (Reference Vey and Campbell2004) and Williams and Shiaw (Reference Williams and Shiaw1999), as identified much earlier by Organ (Reference Organ1988).
Barbuto and Wheeler (Reference Barbuto and Wheeler2006) and van Dierendonck (Reference van Dierendonck2011) theoretically argue that individual-level servant leadership precedes OCB. Although no study has empirically investigated the possible relationship between servant leadership and OCB at the individual level, theoretical work and work at the unit, group, and team level (e.g., Ehrhart, Reference Ehrhart2004; Walumbwa, Hartnell, & Oke, Reference Walumbwa, Hartnell and Oke2010; Hu & LidenReference Liden,Reference Hu and Liden2011) support the link between servant leadership and OCB. We agree with van Dierendonck (Reference van Dierendonck2011) that individual needs are the primary subject of servant leadership, and we therefore accept servant leadership as a micro- and individual-level construct. Given that no study has empirically tested the effect of servant leadership on OCB, we propose the following hypothesis:
Hypothesis 1: Servant leadership is significantly and positively related to OCB.
Psychological climate and OCB
Psychological climate refers to a follower’s perception of the psychological impact of the work environment on her/his well-being (Jones & James, Reference Jones and James1979; James & James, Reference James and James1989). Scholars distinguish between psychological climate and organizational climate in the literature (e.g., Hellriegel & Slocum, Reference Hellriegel and Slocum1974; James & Jones, Reference James and Jones1974; Schulte, Ostroff, & Kinicki, Reference Schulte, Ostroff and Kinicki2006; James et al., Reference James, Choi, Ko, McNeil, Minton, Wright and Kim2008): whereas each individual’s perception of the work environment represents psychological climate at the individual level of analysis, organizational climate is accepted as a unit-, group-, team-, or organizational-level construct. We adopt psychological climate as an individual climate perception in this study, and we do not aggregate the construct at a unit, group, or team level (e.g., Parker et al., Reference Parker, Baltes, Young, Huff, Altmann, LaCost and Roberts2003). Using an aggregation of climate perception is unnecessary because the unique aspect of servant leadership is that it ensures that other people’s highest priority needs are being served, and each follower’s highest priority needs are different. Therefore, each follower evaluates her/his individual climate perception by determining whether her/his highest priority needs are met vis-à-vis their counterparts in the work environment. Therefore, each follower’s attitude depends on her/his own individual perception of whether her/his highest priority needs are being met, not shared perceptions of whether others’ highest priority needs are being met. Each individual’s perception of the work environment is shaped through managerial behaviors such as following through on commitments, considering employees when making decisions, acting on new ideas provided by employees, creating a trusting environment, showing employees that they are valued, and creating good working relationships between supervisors and subordinates (Rogg, Schmidt, Shull, & Schmitt, Reference Rogg, Schmidt, Shull and Schmitt2001). We contend that after followers perceive the serving behaviors of their servant leaders in their work environment, they feel obligated to reciprocate those serving behaviors by demonstrating similar serving behaviors to their followers. One servant leader may help a new subordinate learn how to do a specific part of her/his job, while another may help a subordinate to find a way to finish a job with a tight deadline. The social interaction between a servant leader and her/his followers thereby creates a psychological climate as perceived by followers that is very beneficial to their professional development. In return, because of the psychological climate created through the social interaction between a servant leader and the followers, we contend that followers reciprocate the services that advance their professional development by serving their subordinates to help them develop professionally.
Based on the previous discussion, we specifically contend that psychological climate has a positive effect on OCB. When the highest priority needs of employees are met through servant leadership, individual employees perceive the work environment as a place where managers are competent and consistent, employees follow through on commitments, and everyone cooperates and coordinates to achieve the organizational objectives. When these perceptions are cognitively processed by an employee, they ultimately persuade the employee of the importance of showing appropriate OCB to support the work environment. In return, followers reciprocate their servant leaders’ helping or serving behaviors by helping their new subordinates perform their jobs better. We contend that reciprocation is a mechanism for followers to respond to the services they receive from their servant leaders and therefore that psychological climate affects OCB through the effect of reciprocation. Based on recent studies supporting a positive and significant effect of psychological climate on OCB (e.g., Carr, Schmidt, Ford, & DeShon, Reference Carr, Schmidt, Ford and DeShon2003; Parker et al., Reference Parker, Baltes, Young, Huff, Altmann, LaCost and Roberts2003; Schneider et al., Reference Schneider, Ehrhart, Mayer, Saltz and Niles-Jolly2005; D’Amato & Zijlstra, Reference D’Amato and Zijlstra2008; van Dierendonck, Reference van Dierendonck2011), we propose Hypothesis 2 below. In addition, because we consider that servant leadership, job satisfaction, and psychological climate are individual determinants of OCB, we further argue that the three determinants taken together are positive and significant determinants of OCB (Hypothesis 3). In other words, each determinant contributes to explain OCB given the availability of the other two determinants. Therefore, we propose the following:
Hypothesis 2: Psychological climate is significantly and positively related to OCB.
Hypothesis 3: Servant leadership, job satisfaction, and psychological climate taken together are significant and positive determinants of OCB.
Servant leadership and job satisfaction
Servant leadership begins when leaders accept the role of serving their followers over their own interests, and it assigns two roles—servant and leader—to one person, regardless of that person’s status or calling (Greenleaf, Reference Greenleaf1977). As discussed previously, SET informs several previous studies on leadership with respect to the significant antecedents of job satisfaction in individual relationships (e.g., Janssen & van Yperen, Reference Janssen and van Yperen2004; Judge & Piccolo, Reference Judge and Piccolo2004; Walumbwa, Avolio, Gardner, Wernsing, & Peterson, Reference Walumbwa, Avolio, Gardner, Wernsing and Peterson2008). However, servant leadership is treated as an exception because it differs from transformational leadership, charismatic leadership, leader-member exchange, and authentic leadership (e.g., Graham, Reference Graham1991; Ehrhart, Reference Ehrhart2004; Barbuto & Wheeler, Reference Barbuto and Wheeler2006; Reference LidenLidenet al., Reference Liden, Wayne, Zhao and Henderson2008; Walumbwa, Hartnell, & Oke, Reference Walumbwa, Hartnell and Oke2010).
Previous empirical studies do little to inform us regarding the direct effect of servant leadership on job satisfaction in a for-profit organizational context. We consider a for-profit organizational context in this study because Yukl (Reference Yukl2006) emphasizes that leadership cannot be understood when it is investigated in isolation from the context in which it occurs. For example, Barbuto and Wheeler (Reference Barbuto and Wheeler2006) develop a servant leadership scale and report that the self-reporting and rater-reporting servant leadership subscales are significantly and positively correlated with employee satisfaction, but not necessarily with job satisfaction. Similarly, in an empirical study, Mayer, Bardes and Piccolo (Reference Mayer, Bardes and Piccolo2008) show that the effect of servant leadership on job satisfaction is not direct but rather indirect through a chain of direct and mediating effects. Accordingly, servant leadership affects the satisfaction of overall needs directly and indirectly through a mediating variable of overall organizational justice. Thereafter, the satisfaction of overall needs affects job satisfaction directly. van Dierendonck (Reference van Dierendonck2011) and van Dierendonck and Nuijten (Reference van Dierendonck and Nuijten2011) consider servant leadership characteristics (sub-dimensions of the servant leadership scale, not necessarily the servant leadership scale itself) such as empowering and developing people, humility, authenticity, interpersonal acceptance, providing direction, and stewardship as antecedents of job satisfaction because these characteristics create high-quality leader-follower relationships. Earlier, Chu (Reference Chu2008) and Mayer, Bardes and Piccolo (Reference Mayer, Bardes and Piccolo2008) report a significant, positive correlation between servant leadership and job satisfaction, although their research intention is not to test a hypothesis investigating a direct effect of servant leadership on job satisfaction. However, in a structural relationship model, Gonzalez and Garazo (Reference Gonzalez and Garazo2008) show that the effect of servant leadership on job satisfaction is not significant and suggest a need to further investigate the true effect of servant leadership on job satisfaction. Therefore, the inadequacy of the previous results reported in the extant literature on servant leadership theory in relation to for-profit organizations motivated us to further investigate the veracity of the direct, positive, and significant effect of servant leadership on job satisfaction.
Job satisfaction refers to the emotional state that emanates from an individual’s assessment of his or her job (Locke, Reference Locke1976) or to a psychological state concurrently signified by affective and cognitive indicators (Hulin & Judge, Reference Hulin and Judge2003). Job satisfaction is further defined as the extent to which an employee reports satisfaction with features of her/his job (Warr, Cook, & Wall, Reference Warr, Cook and Wall1979). The intrinsic features of job satisfaction include providing employees the autonomy to choose their own way of working, recognizing superior work, giving employees responsibility, and offering employees opportunities to use their abilities (e.g., Warr, Cook, & Wall, Reference Warr, Cook and Wall1979; Williams & Anderson, Reference Williams and Anderson1991). Some extrinsic features of job satisfaction include an employee’s immediate supervisor, her/his fellow workers, the physical work conditions, and the rate of pay (e.g., Warr, Cook, & Wall, Reference Warr, Cook and Wall1979; Williams & Anderson, Reference Williams and Anderson1991; Clark, Oswald, & Warr, Reference Clark, Oswald and Warr1996; Judge, Piccolo, Podsakoff, Shaw, & Rich, Reference Judge, Piccolo, Podsakoff, Shaw and Rich2010). For the purposes of this study, we adopt Warr, Cook and Wall (Reference Warr, Cook and Wall1979) definition of job satisfaction and specifically define it as an employee’s positive assessment of the intrinsic and extrinsic features of her/his job or of her/his experiences with those features. Because servant leadership depends on not only achieving organizational success but also serving subordinates, we contend that the more that employees are served by their leaders, the more that they will be satisfied with their jobs.
Given our previous discussion of the reciprocity of SET to explain the effect of leadership on certain employees’ workplace attitudes and behaviors, we further argue that followers reciprocate the services that they receive from their servant leaders by showing and voicing their satisfaction with the intrinsic and extrinsic features of their jobs. Specifically, we argue that when servant leaders serve the interests of their followers and positively affect their lives, followers tend to have, for example, the autonomy to choose their own ways of working, responsibility in their jobs, the right to use their abilities, encouraging physical work conditions, and satisfying rates of pay. In addition, when servant leaders show commitment and skill in assisting their followers in recovering from a personal trauma or hardship by listening or creating an environment in which followers are comfortable discussing their personal and professional problems, leaders create good working relationships with their followers. These good relationships may also help followers establish better relationships with their fellow coworkers. Finally, when servant leaders have knowledge and awareness about what is occurring and what is about to occur in their workplaces, they give their followers the responsibility of preventing potential problems and offer them the opportunity to use their abilities. We further articulate that when the followers of servant leaders hear statements about job satisfaction from their peers, subordinates, and supervisors, when they cognitively process how their servant leaders patiently meet their needs, and when they repeatedly experience the services of their servant leaders, followers reciprocate those experiences by voicing their satisfaction and showing that they enjoy a positive experience from the intrinsic and extrinsic features of their jobs. Therefore, we argue that servant leadership enhances followers’ job satisfaction because ‘the servant-leader concept emphasizes increased service to others; a holistic approach to work; promoting a sense of community; and the sharing of power in decision making’ (Robert K. Greenleaf Center, 1991). This argument is also supported by other leadership theories such as transformational leadership (e.g., Braun, Peus, Weisweiler, & Frey, Reference Braun, Peus, Weisweiler and Frey2013). Given the above discussion, we propose the following hypothesis:
Hypothesis 4: Servant leadership is significantly and positively related to job satisfaction.
Psychological climate as a mediating mechanism
In addition to directly affecting job satisfaction, servant leadership may influence job satisfaction by enhancing followers’ psychological climate. Because leadership is generally accepted to be a strong determinant of psychological climate (e.g., Schneider et al., Reference Schneider, Ehrhart, Mayer, Saltz and Niles-Jolly2005; Walumbwa et al., Reference Walumbwa, Avolio, Gardner, Wernsing and Peterson2008; van Dierendonck, Reference van Dierendonck2011; Liden, Reference Liden2013), we argue that servant leadership is a positive and significant antecedent of psychological climate because it creates reciprocal serving obligations through servant leadership practices that increase the collective interest of followers (Blau, Reference Blau1964).
We also argue that psychological climate has a positive influence on job satisfaction. Psychological climate is shaped by the perceived social environment in which meaning is created and particular information is brought to employees’ attention. Thus, an informational social influence grows, and it is reflected by the employees in the form of job satisfaction. Here, informational social influence becomes the mechanism through which an employee attributes meaning to managerial competence and consistency and employee commitment, cooperation, and coordination. Informational social influence is the mechanism through which job satisfaction is attained by an individual. Therefore, we propose that the social influence of information in the social environment helps to explain the relationship between psychological climate and job satisfaction. Accordingly, an individual’s level of job satisfaction is determined by the individual’s perceived overall satisfaction with their social environment, which her/his servant leader shapes through her/his serving behaviors. In addition, Zalesny and Ford (Reference Zalesny and Ford1990) provide empirical evidence for the effect of social information on job satisfaction, and previous research supports the effect of psychological climate on employee outcomes, including job satisfaction (e.g., Lyon & Ivanchevich, Reference Lyon and Ivanchevich1974; Brown & Leigh, Reference Brown and Leigh1996; Schulte, Ostroff, & Kinicki, Reference Schulte, Ostroff and Kinicki2006; Walumbwa et al., Reference Walumbwa, Avolio, Gardner, Wernsing and Peterson2008).
Psychological climate is also a critical mediating mechanism in the individual relationships between servant leadership and job satisfaction (van Dierendonck, Reference van Dierendonck2011). Psychological climate is defined as the experiential and perceptual components of an employee’s interaction with managers, departments, subordinates, supervisors, and coworkers regarding managerial competence and consistency and employee commitment, cooperation, and coordination. It follows that servant leadership plays a vital role in determining these individual climate perceptions and thus in creating the individual psychological climate. As SET notes, ‘managerial practices that advance the collective interest of subordinates create joint obligations’ (Blau, Reference Blau1964: 207). Lewin, Lippitt, and White (Reference Lewin, Lippitt and White1939) support this argument by demonstrating that job satisfaction results from different climates created by different leadership styles. We add to this proposition that climate creates a social environment that shapes individuals’ attitudes. The social information that an employee gains from her/his social environment affects her/his job satisfaction (Zalesny & Ford, Reference Zalesny and Ford1990; Walumbwa et al., Reference Walumbwa, Avolio, Gardner, Wernsing and Peterson2008), and social information about the needs of followers offers servant leaders an opportunity to provide service and support to subordinates in the form of technical advice on specific tasks or emotional support to help those in need of personal healing (Reference Liden, Wayne, Liao and MeuserLiden et al., 2014b). Followers show serving behaviors after they experience the benefits of their servant leader’s service to help them become healthier, wiser, freer, and more autonomous (van Knippenberg, van Knippenberg, De Cremer, & Hogg, Reference van Knippenberg, van Knippenberg, De Cremer and Hogg2004). Therefore, we maintain that the more that subordinates are psychologically fulfilled by the experience of serving others and the more that individuals benefit from the actions of their servant leaders, the more that the followers will be satisfied with the intrinsic and extrinsic features of their jobs. Reference Liden, Wayne, Liao and MeuserLiden et al., 2014b: 14) support this argument: ‘As a relational leadership style, the servant leader’s one-to-one attention to, and care for, followers magnifies the salience of followers serving fellow coworkers and customers.’ Recent theoretical and empirical studies also consider psychological climate to be a mediator in various relationships (e.g., Walumbwa, Hartnell, & Oke, Reference Walumbwa, Hartnell and Oke2010; van Dierendonck, Reference van Dierendonck2011; Peterson, Galvin, & Lange, Reference Peterson, Galvin and Lange2012). In parallel with these studies and in line with the above discussion, we propose that psychological climate enhances favorable follower attitudes and behaviors, both directly and indirectly, as implied by the norm of reciprocity described by SET. Therefore, we propose the following hypotheses:
Hypothesis 5: Servant leadership is significantly and positively related to psychological climate.
Hypothesis 6: Psychological climate mediates the relationship between servant leadership and job satisfaction.
Methods
Sample and procedures
The data collection method was survey questionnaire. The questionnaires were written in English and subjected to back-translation (Brislin, Reference Brislin1980). The participants received a survey packet containing a cover letter from the researchers that was endorsed by the company’s CEO requesting their participation. The cover letter provided assurances of confidentiality, explained the objectives of the study, emphasized the importance of responding to all of the surveys, and so forth.
In both of the studies we describe in detail below, the authors visited the CEOs of the companies, informed them about the objectives of the study, and requested their participation. Following the suggestions of the CEOs who agreed to participate in the study, we visited the heads of the human resources departments in each company to request their assistance in collecting the data. Each human resources department assigned a contact person to help us collect the data for the study. A stratified random sampling strategy (in which each employee in a stratum was given the same probability of being selected for the sample for the stratum) was used in both of the studies to ensure that all of the departments in each participating company were represented in the sample.
A full description of Study 1
Study 1 was a large study in which we surveyed 306 full-time employees of 12 companies. The companies were chosen from a variety of industries, such as banking, iron/steel, railroads, and dairy, located in Iskenderun, a heavily industrialized region in Turkey. We deliberately collected our data from many diverse industries because we wanted to prevent or minimize potential contextual restrictions related to any peculiar organizational forms in Study 1 and 2. Data are commonly collected from different companies in conducting research on servant leadership, climate, and OCB (e.g., Walumbwa, Hartnell, & Oke, Reference Walumbwa, Hartnell and Oke2010).
In Study 1, the contact person assigned a number to each participant and wrote this number on that participant’s questionnaire. Therefore, we were able to inform the contact person about nonrespondents so that they could send a second request to the participants. We did not have the lists or names of the participating employees. In addition, no one in the participating companies had access to the employee responses, which were submitted in sealed envelopes. In this way, the respondents were provided with anonymity, which increased our response rate.
The questionnaires were distributed to the employees on the list by the contact person, and the employees submitted their questionnaires in sealed envelopes. Ultimately, we collected 284 usable questionnaires (a response rate of 93%). In Study 1, we collected data from the employees in the beginning months of 2010 with one questionnaire that included all the study measures.
A full description of Study 2
In Study 2, 325 full-time employees of 15 companies were chosen from a variety of industries, such as textiles, banking, petroleum, and brick manufacturing, located in Iskenderun, Turkey. The companies in Study 1 and Study 2 were different. The response rate was improved by an assurance of anonymity. The questionnaires were distributed to the employees on the list by the contact person. The contact person also verbally identified the employee for whom each supervisor and peer would complete the questionnaire. Thus, no employee names appeared on the questionnaires that we collected from the supervisors and peers; only the previously assigned employee numbers were provided. Ultimately, we collected 286 usable questionnaires (a response rate of 88%) in Study 2 (the data in Study 2 also included the responses of 123 supervisors and 286 peers of the participating employees). We collected data from October 2011 to July 2012 and used two procedural remedies to control for common methods variance (CMV) (Podsakoff, MacKenzie, Lee, & Podsakoff, Reference Podsakoff, MacKenzie, Lee and Podsakoff2003; Podsakoff, MacKenzie, & Podsakoff, Reference Podsakoff, MacKenzie and Podsakoff2012). The data that were collected in Study 2 and their sources are discussed next.
At Time 1, we asked the participating employees to assess the servant leadership and psychological climate in their organizations and to provide their demographics (Questionnaire 1). The employees were asked to complete the servant leadership scale at work and the psychological climate scale at home because this temporal and locational separation would help the participants to eliminate common retrieval cues from responding to both scales. We allowed the participants 1 week to complete each survey. A total of 325 employee survey questionnaires were distributed at Time 1, and 310 completed questionnaires were returned (a response rate of 95%). At Time 2, ~3 weeks later, based on the information from the completed Time 1 surveys, we distributed a second survey (Questionnaire 2) to the employees that rated their job satisfaction. In all, 302 of 310 (97 per cent) completed surveys were returned at Time 2.
At Time 3, ~3 weeks after Questionnaire 2 was collected, the employees who completed the surveys at Time 1 and Time 2 were given a third survey (Questionnaire 3) that assessed employee-reported OCB (eOCB). Because OCB was the dependent variable in some of our hypotheses, we also collected data on each participant employee’s OCB from two other sources (source separation): their supervisors and their peers. Accordingly, Questionnaire 3 was modified into Questionnaire 4 to measure supervisor-rated OCB (sOCB) and into Questionnaire 5 to measure peer-rated OCB (pOCB). Therefore, at Time 3, our dependent variable was collected from three different sources, namely eOCB, sOCB, and pOCB. A total of 302 employee, supervisor, and peer surveys were distributed to the three separate sets of respondents at Time 3. Matching the data collected at Time 1, Time 2, and Time 3 produced a total of 286 usable surveys.
The description/demographic characteristics of the participants in Study 1 and Study 2 are presented in Table 1.
The description/demographic characteristics of the participants in study 1 and study 2
Measures
We used Likert-type scales with original anchors. The anchors were unique for each scale to further control for CMV (Podsakoff et al., Reference Podsakoff, MacKenzie, Lee and Podsakoff2003; Podsakoff, MacKenzie, & Podsakoff, Reference Podsakoff, MacKenzie and Podsakoff2012). In line with our hypothesized model, we used four latent constructs: servant leadership, job satisfaction, psychological climate, and OCB. Table 2 shows the factor loadings from our measurement models (λ1a=eOCB was used in Study 1; λ1b=eOCB, λ2b=sOCB, and λ3b=pOCB were used in Study 2). The standardized factor loadings in Table 2 were all significant (all z>2.58) and large, ranging from 0.62 to 1.00. All of the scales achieved high levels of internal reliability (alpha) in both Study 1 and Study 2 (see Table 4).
Factor loadings in measurement model
The lambdas (λ) are from the completely standardized solution; listwise an=284 (Study 1), listwise bn=286 (Study 2).
1eOCB (employee-reported OCB) was used. 2sOCB (supervisor-reported OCB) was used. 3pOCB (peer-reported OCB) was used. The standardized factor loadings were significant (all z>2.58).
Servant leadership
We used a 23-item scale devised by Barbuto and Wheeler (Reference Barbuto and Wheeler2006) to measure servant leadership. An example item stated, ‘This person puts my best interest ahead of her/his own.’ A 4-point Likert-type scale with anchors ranging from 1=‘strongly disagree’ to 4=‘strongly agree’ was used to indicate the participants’ agreement with the 23 items. The content, discriminant, convergent, and nomological validity of the servant leadership scale were previously assured in different contexts (e.g., Barbuto & Wheeler, Reference Barbuto and Wheeler2006; Sun & Wang, Reference Sun and Wang2009; Zhang et al., Reference Zhang, Kwan, Everett and Jian2012).
Psychological climate
We used a 17-item scale developed by Rogg et al. (Reference Rogg, Schmidt, Shull and Schmitt2001) to measure psychological climate. A sample item stated, ‘Employees say they are proud to work here.’ A five-point scale with response options ranging from 1=‘strongly disagree’ to 5=‘strongly agree’ was used to indicate the participants’ agreement with the 17 items.
Job satisfaction
We used Warr, Cook, and Wall (Reference Warr, Cook and Wall1979) 15-item job satisfaction scale to measure job satisfaction. A sample item stated, ‘My present job gives me the freedom to choose my own method of working.’ A 7-point Likert-type scale ranging from 1=‘I am extremely dissatisfied’ to 7=‘I am extremely satisfied’ was used to indicate the participants’ agreement with the 15 items.
OCB
We used a 19-item scale derived primarily from Vey and Campbell (Reference Vey and Campbell2004) and Williams and Shiaw (Reference Williams and Shiaw1999) to measure OCB. A sample item stated, ‘I don’t spend time in personal conversations.’ A 6-point Likert-type scale with anchors ranging from 1=‘strongly disagree’ to 6=‘strongly agree’ was used to indicate the participants’ agreement with the 19 items.
Control variables
On the basis of prior research, gender, marital status, age, tenure with current supervisor, organizational tenure, and educational level of the participating employees were used as control variables (e.g., Locke, Reference Locke1976; Williams & Shiaw, Reference Williams and Shiaw1999; Janssen & van Yperen, Reference Janssen and van Yperen2004). We measured gender and marital status as dichotomous variables, with gender coded as 0 for female and 1 for male and marital status coded as 0 for unmarried and 1 for married. Employee age was measured in groups as follows: 0=‘less than 20,’ 1=‘20–29,’ 2=‘30–39,’ 3=‘40–49,’ 4=‘50–59,’ and 5=‘60–69 years old.’ We measured tenure with current supervisor and organizational tenure in groups as follows: 0=‘less than 5,’ 1=‘5–9,’ 2=‘10–14,’ 3=‘15–19,’ 4=‘20–24,’ 5=‘25–29,’ and 6=‘30 years and over.’ Education was also measured in groups, ranging from 1=‘primary school-level degree’ to 7=‘doctorate-level degree.’ All the control variables are also presented in Table 1.
Results
Study 1
We used a one-way analysis of variance in SPSS 20 to assess whether there were differences between the industries from which we collected our data (four service organizations [n=100] and eight manufacturing organizations [n=184]) in Study 1. The results revealed no statistical differences in servant leadership, job satisfaction, psychological climate, or OCB between the manufacturing and service organizations.
We first examined the adequacy of our measurement models by performing a confirmatory factor analysis in LISREL 8.80 (Jöreskog & Sörbom, Reference Jöreskog and Sörbom2006). As shown in Table 3, our hypothesized 4-factor model in Study 1 (Study 11) fit the data well (χ 2=222.22, df=84, χ 2⧸df=2.64, CFI=0.99, IFI=0.99, TLI=0.98, SRMR=0.052, RMSEA=0.059). A good fitting model is generally determined by a χ 2⧸df ratio of <3 or 5, CFI and IFI values >0.90, and SRMR and RMSEA values of about 0.05 (Kline, Reference Kline2011). Byrne (Reference Byrne2010) also proposes that CFI and TLI values above 0.95 indicate superior fit, whereas values between 0.90 and 0.95 indicate adequate fit. Overall, given Byrne (Reference Byrne2010) and Kline (Reference Kline2011) suggestions for model fit, the hypothesized model considered in Study 1 fit the corresponding data very well.
Comparison of alternative measurement models
1eOCB (employee-reported OCB) was used. 2sOCB (supervisor-reported OCB) was used. 3pOCB (peer-reported OCB) was used.
CFI=Comparative Fit Index; IFI=Incremental Fit Index; JS=Job Satisfaction; OCB=organizational citizenship behavior; PC=psychological climate; RMSEA=root mean square error of approximation; SL=servant leadership; SRMR=standardized root mean square residual; TLI=Tucker-Lewis Index (also known as NNFI [The Non-Normed Fit Index]).
a PC and JS combined.
b SL and JS combined.
c SL and PC combined.
d SL, PC, and JS combined.
e SL, PC, JS, and OCB combined.
We also investigated alternative measurement models (the 3-factor, 2-factor, and 1-factor models for Study 1 presented in Table 3) to ensure that our hypothesized 4-factor model best reflected the data structure. The results supported the distinctiveness of the 4-factor model. Next, we conducted a correlation analysis in SPSS 20 (see Table 4) and then a path analysis (see Figure 2 and Table 5) with our four latent variables in LISREL 8.80 (Jöreskog & Sörbom, Reference Jöreskog and Sörbom2006) to test our hypotheses. The details of each analysis and the process followed for each analysis are explained in the related sections below.
Results of study 1.
Means, standard deviations, reliabilities, and correlations among study variablesFootnote a, Footnote b
Two-tailed tests. Means and standard deviations are for the unstandardized variables.
Organizational Citizenship Behavior=OCB, eOCB=employee-reported OCB; pOCB=peer-reported OCB; sOCB=supervisor-reported OCB
a n=284 (Study 1 in lower left corner).
b n=286 (Study 2 in upper right corner).
*p<.05; **p<.01
Fit indices for structural models
Note: All alternatives were compared to hypothesized model.
1eOCB (employee-reported OCB) was used. 2sOCB (supervisor-reported OCB) was used. 3pOCB (peer-reported OCB) was used.
Alternative 1: no direct path between SL and OCB; Alternative 2: no direct path between JS and OCB; Alternative 3: no direct path between PC and OCB; Alternative 4: no direct path between SL and JS; Alternative 5: no direct path between SL and PC; Alternative 6: no direct path between PC and JS; Alternative 7: no direct path between SL and OCB, and PC and OCB; Alternative 8: no direct path between SL and OCB, and between JS and OCB; Alternative 9: no direct path between PC and OCB, and between JS and OCB; Alternative 10: no direct path between SL and JS, and between SL and PC; Alternatiave 11: no direct path between SL and JS, and between PC and JS; Alternative 12: no direct path between SL and PC, and between PC and JS.
CFI=Comparative Fit Index; IFI=Incremental Fit Index; JS=Job Satisfaction; OCB=organizational citizenship behavior; PC=psychological climate; RMSEA=root mean square error of approximation; SL=servant leadership; SRMR=standardized root mean square residual; TLI=Tucker-Lewis Index (also known as NNFI [The Non-Normed Fit Index]).
In Table 4, we report the means, standard deviations, reliability estimates, and correlations for all of the study variables (in bold in the lower left corner of Table 4). The hypotheses (see Figure 2) were tested with a structural equation model. Our 4-factor hypothesized model provided the best fitting model (see Study 1 in Table 5) (χ 2=222.22, df=84, χ 2⧸df=2.64, CFI=0.99, IFI=0.99, TLI=0.98, SRMR=0.052, RMSEA=0.059) among the other 12 alternative models that we considered in Study 1. With our 84 degrees of freedom, a power test of 0.80 provided a power estimate of 0.96244 for the test of exact fit (with a null value of RMSEA=0.00 and an alternative value of RMSEA=0.05), 0.9801 for the test of close fit (with a null value of RMSEA=0.05 and an alternative value of RMSEA=0.08), and 0.94856 for the test of not-close fit (with a null value of RMSEA=0.05 and an alternative value of RMSEA=0.01) based on α=0.05. The three power estimates are thus much larger than the target level of power, 0.80 (MacCallum, Browne, & Sugawara, Reference MacCallum, Browne and Sugawara1996; Kline, Reference Kline2011). In addition, the minimum sample size in Study 1 to achieve a level of power of 0.80 with 84 degrees of freedom was 180 for an exact fit, 148 for a close fit, and 197 a not-close fit, all of which are much lower than our sample size of 284 (see MacCallum, Browne, & Sugawara, Reference MacCallum, Browne and Sugawara1996). In addition, our LISREL 8.80 output provided a Critical N value of 203.5 for our hypothesized 4-factor model. Byrne (Reference Byrne1998) and Hoelter (Reference Hoelter1983) argue that a Critical N value >200 indicates that a model adequately represents the sample data. Based on the sample size of 284 and the Critical N value of 203.5 for Study 1, which is larger than the minimum cut-off value for Critical N, we can conclude that the fit of our 4-factor hypothesized model is adequate.
Furthermore, we ensured that our observed continuous variables were normally distributed with skewness and kurtosis values between ±0.65. We used a covariance matrix to conduct our latent variable structural path analysis in LISREL 8.80 in Study 1 and used maximum likelihood as the estimation method. The linearity between the observed variables was ensured. The results of latent variable structural path analysis are presented in Figure 2. Hypothesis 1 posited that servant leadership is positively and significantly related to OCB; however, we found a nonsignificant path between servant leadership and OCB (γ=0.01, n.s.=not significant). Therefore, Hypothesis 1 was not supported. Hypothesis 2 posited that psychological climate is positively and significantly related to OCB, but the data showed that psychological climate is a negative and significant antecedent of OCB (β=−0.19, p<.05). Thus, Hypothesis 2 was not supported. Hypothesis 3 posited that servant leadership, job satisfaction, and psychological climate collectively are significant and positive antecedents of OCB, yet the data showed that only job satisfaction is a positive and significant antecedent of OCB (β=0.66, p<.001). Nevertheless, given the results for Hypotheses 1 and 2, Hypothesis 3 was not supported. Hypothesis 4 posited that servant leadership is a positive and significant antecedent of job satisfaction, and the data revealed that servant leadership is a positive and significant antecedent of job satisfaction (γ=0.12, p<.05) (for the reduced form of the same relationship, γ=0.55, p<.001). Thus, Hypothesis 4 was supported. Hypothesis 5 posited that servant leadership is a positive and significant antecedent of psychological climate, and the data again revealed that servant leadership is a positive and significant antecedent of psychological climate (γ=0.60, p<.001). Thus, Hypothesis 5 was supported.
Hypothesis 6 posited that psychological climate mediates the relationship between servant leadership and job satisfaction. Our analysis revealed a significant path between servant leadership and psychological climate (γ=0.60, p<.001) and between psychological climate and job satisfaction (β=0.72, p<.001). Given these results, there is ‘some’ significant mediation in our data. Therefore, we computed the z score to test the relative sizes of the mediated (indirect) and direct paths (Iacobucci, Saldanha, & Deng, Reference Iacobucci, Saldanha and Deng2007). According to a Sobel test (Sobel, Reference Sobel1982), the indirect effect of servant leadership on job satisfaction was significant (indirect effect=0.432, z=105.57, p<.000), and the direct path between servant leadership and job satisfaction was also significant (γ=0.12, p<.05). Because both the indirect path, based on the z score (z=105.57), and the direct path from servant leadership to job satisfaction were significant, we can conclude that there is ‘partial’ mediation. We also constructed a continuous index and found that a substantially larger portion of the variance in job satisfaction due to servant leadership is explained via the indirect path (78%=[0.432/.55]×100) relative to the direct path (22%=[0.12/0.55]×100) (see Iacobucci, Saldanha, & Deng, Reference Iacobucci, Saldanha and Deng2007). MacKinnon, Coxe and Baraldi (Reference MacKinnon, Coxe and Baraldi2012: 6) suggest that ‘it will be important to obtain some measure of effect size to judge the importance or size of an effect.’ Following Preacher and Kelley (Reference Preacher and Kelley2011) and Kelly and Preacher (Reference Kelley and Preacher2012), we also computed the kappa (κ) value (for the maximum possible indirect effect in our mediation analysis and its standardized version) to assess the effect size of our partial mediation. The maximum possible indirect effect was 0.9879, yet the observed indirect effect was 0.432. Although the observed indirect effect was statistically significant, it was much smaller than the maximum possible effect. We nevertheless consider the standardized effect size to be a better indicator of effect size, and we obtained 0.432/0.9879=0.437 for the κ2 value, which indicates the proportion of the maximum possible indirect effect that we could have obtained. κ2=0 indicates no linear indirect effect, whereas κ2=1 indicates the maximum possible indirect effect. Further, κ is independent of sample size. Given these results and the interpretation of κ2 as the proportion of variance accounted for in one variable by another, we argue that the mediating effect in Study 1 is large (see Cohen, Reference Cohen1988; Preacher & Kelley, Reference Preacher and Kelley2011), supporting our previous finding that the partial mediating effect that we found in Study 1 is large.
Study 2
In Study 2, we followed the same one-way analysis of variance procedure that we followed in Study 1 to ensure that combining the data collected from six service companies (n=73) and nine manufacturing companies (n=213) was appropriate. The results showed that there were no statistical differences in manufacturing and service organizations across servant leadership, job satisfaction, psychological climate, eOCB, sOCB, or pOCB between the manufacturing and service organizations in Study 2.
The adequacy of the measurement models was assessed by confirmatory factor analysis in LISREL 8.80 (Jöreskog & Sörbom, Reference Jöreskog and Sörbom2006). Table 3 shows the hypothesized 4-factor models that were used in Study 2. The models fit the data very well (χ2=183.78, df=84, χ2⧸df=2.18, CFI=0.99, IFI=0.99, TLI=0.99, SRMR=0.049, RMSEA=0.045 when we considered eOCB (Study 21); χ2=206.90, df=84, χ2⧸df=2.46, CFI=0.99, IFI=0.99, TLI=0.99, SRMR=0.043, RMSEA=0.043 when we considered sOCB (Study 22); χ2=155.59, df=84, χ2⧸df=1.85, CFI=1.00, IFI=1.00, TLI=1.00, SRMR=0.024, RMSEA=0.013 when we considered pOCB [Study 23]).
Alternative measurement models were also investigated to ensure that our hypothesized 4-factor models best reflect the data structure. The results supported the distinctiveness of our 4-factor models (Study 21, Study 22, and Study 23 in Table 3). After the measurement model analysis, we conducted a correlation analysis in SPSS 20 and then a latent variable structural path analysis with the four latent variables in LISREL 8.80 (Jöreskog & Sörbom, Reference Jöreskog and Sörbom2006) to test the hypotheses of our study. Next, we present the details of each analysis and the processes that we followed in the related sections below.
The means, standard deviations, reliability estimates, and correlations for all of the study variables are reported in Table 4 (the upper right corner). We tested our hypotheses (see Figure 3) with a structural equation model. Our 4-factor hypothesized models provided the best fitting models (see Study 21, Study 22, and Study 23 in Table 5) among the other 12 alternative modes that we considered for Study 2. Based on an α of 0.05, we also obtained a power estimate of 0.96946 for the test of exact fit, 0.9801 for the test of close fit, and 0.96946 for the test of not-close fit with our 84 degrees of freedom. All three power estimates are thus much larger than the target level of power, 0.80 (MacCallum, Browne, & Sugawara, Reference MacCallum, Browne and Sugawara1996; Kline, Reference Kline2011). To achieve a minimum level of power of 0.80 with 84 degrees of freedom, the sample size needs to be 180 for an exact fit, 148 for a close fit, and 197 a not-close fit (MacCallum, Browne, & Sugawara, Reference MacCallum, Browne and Sugawara1996). Our sample size of 286 for Study 2 is much higher than each of these values. In addition, for our hypothesized 4-factor models, the LISREL output provided a Critical N value of 243.55 for eOCB (Study 21), 262.84 for sOCB (Study 22), and 377.07 for pOCB (Study 23). Previous studies have argued that a Critical N value larger than 200 indicates a model that adequately depicts the sample data (e.g., Hoelter, Reference Hoelter1983; Byrne, Reference Byrne1998). Based on the sample size of 286 for Study 2 and the Critical N values that were larger than the cut-off of 200, we can conclude that the fit of our hypothesized models in Study 2 is adequate.
Results of study 2.
Normality was sustained in the range reported previously for Study 1. We again used a covariance matrix to conduct our latent variable structural path analysis and used maximum likelihood as the estimation method. Further, we ensured the linearity between the observed variables, and our sample size was large. The results of our hypothesis tests are presented in Figure 3. In Hypothesis 1, we posited that a positive and significant relationship exists between servant leadership and OCB. The data revealed a nonsignificant path between servant leadership and eOCB (γ=0.03, n.s.), a positive and significant path between servant leadership and sOCB (γ=0.22, p<.05), and a positive and significant path between servant leadership and pOCB (γ=0.29, p<.01) (see Figure 3 for all the hypothesis tests for Study 2). Thus, Hypothesis 1 was supported when sOCB and pOCB was considered but not supported when eOCB was considered. In Hypothesis 2, we posited that a positive and significant relationship exists between psychological climate and OCB. The data revealed a positive and significant relationship between psychological climate and eOCB (β=0.21, p<.05) but no relationship between psychological climate and sOCB (β=0.00, n.s.) or between psychological climate and pOCB (β=−0.04, n.s.). Therefore, Hypothesis 2 was supported for eOCB but not for sOCB and pOCB. In Hypothesis 3, we posited that servant leadership, job satisfaction, and psychological climate collectively are significant and positive antecedents of OCB. However, the data showed that job satisfaction is not related to eOCB (β=0.05, n.s.), sOCB (β=0.03, n.s.), or pOCB (β=−0.01, n.s.). In addition, given the results for Hypotheses 1 and 2, Hypothesis 3 was not supported. In Hypothesis 4, we posited that a positive and significant relationship exists between servant leadership and job satisfaction, and the data revealed a positive and significant relationship between servant leadership and job satisfaction (γ=0.17, p<.05 for eOCB; γ=0.18, p<.05 for sOCB and pOCB) (for the reduced form of the same relationship, γ=0.52, p<.001 for eOCB, sOCB, and pOCB). Therefore, Hypothesis 4 was supported. In Hypothesis 5, we posited that there is a positive and significant relationship between servant leadership and psychological climate, and the data showed that there is a positive and significant relationship between servant leadership and psychological climate (γ=0.72, p<.001 for eOCB, sOCB, and pOCB). Therefore, Hypothesis 5 was supported.
In Hypothesis 6, we posited that the relationship between servant leadership and job satisfaction is mediated by psychological climate. We found a positive and significant path between servant leadership and psychological climate (γ=0.72, p<.001 for OCB) and between psychological climate and job satisfaction (β=0.48, p<.001 for eOCB; β=0.47, p<.001 for sOCB and pOCB) (Figure 3). Therefore, there is ‘some’ significant mediation in our data. We thus computed the z scores to test the relative sizes of the indirect and direct paths (Iacobucci, Saldanha, & Deng, Reference Iacobucci, Saldanha and Deng2007) by using a Sobel test (Sobel, Reference Sobel1982). We found that the indirect effect of servant leadership on job satisfaction was positive and significant (indirect effect=0.3456, z=68.639, p<.000 for eOCB; indirect effect=0.3384, z=67.2095, p<.000 for sOCB and pOCB) and that direct path between servant leadership and job satisfaction was also significant (γ=0.17, p<.05 for eOCB; γ=0.18, p<.05 for sOCB and pOCB). Since both the indirect path, based on the z scores (z=68.639 for eOCB and z=67.209 for sOCB and pOCB), and the direct path from servant leadership to job satisfaction were significant, we can conclude that the mediation is ‘partial.’ We also constructed a continuous index and found that a substantially larger portion of the variance in job satisfaction due to servant leadership is explained via the indirect path (66%=(0.3456/0.52)×100 for eOCB; 65%=(0.3384/0.52)×100 for sOCB and pOCB) relative to the direct path (34%=(0.1744/0.52)×100 for eOCB; 35%=(0.1816/0.52)×100 for sOCB and pOCB) (see Iacobucci, Saldanha, & Deng, Reference Iacobucci, Saldanha and Deng2007). We also calculated the effect size of the partial mediation (see MacKinnon, Coxe & Baraldi, Reference MacKinnon, Coxe and Baraldi2012), relying on Preacher and Kelley (Reference Preacher and Kelley2011) and Kelley and Preacher (Reference Kelley and Preacher2012) to calculate κ. We found that the maximum possible indirect effect was 1.2037 for eOCB, sOCB, and pOCB but obtained 0.3456 for the observed indirect effect for eOCB and 0.3384 for the observed indirect effects for sOCB and pOCB. Thus, although the observed indirect effects are statistically significant, they are much smaller than the maximum possible effects. Because the standardized effect size is considered to be a better indicator of effect size, we estimated κ-squared and obtained 0.3456/1.2037=0.2871 for eOCB and 0.3384/1.2037=0.2811 for sOCB and pOCB. Based on these results and the interpretation of κ2 as the proportion of variance accounted for in one variable by another, we conclude that the mediating effect in Study 2 is large (see Cohen, Reference Cohen1988; Preacher and Kelley, Reference Preacher and Kelley2011), supporting our earlier finding that the partial mediating effect in Study 2 is large.
Discussion
In this study, we investigated the direct and mediating effects of servant leadership on job satisfaction, psychological climate, and OCB at an individual level of analysis in a for-profit organizational context. We also examined servant leadership, job satisfaction, and psychological climate as antecedents of OCB. Based on two studies employing two different data collection procedures to account for CMV, we thus contribute to the related research and extend our knowledge on servant leadership, job satisfaction, psychological climate, and OCB in at least three ways.
First, our study shows that the servant-first nature of servant leadership directly, positively, and significantly influences job satisfaction, psychological climate, and OCB (see sOCB and pOCB in Figure 3), confirming the theoretical argument in the literature that servant leadership produces these expected benefits (e.g., Greenleaf, Reference Greenleaf1977; Barbuto & Wheeler, Reference Barbuto and Wheeler2006; Reference LidenLidenet al., Reference Liden, Wayne, Zhao and Henderson2008; Walumbwa, Hartnell, & Oke, Reference Walumbwa, Hartnell and Oke2010; van Dierendonck, Reference van Dierendonck2011; Reference Liden, Wayne, Liao and MeuserLiden et al., 2014b). Further, servant leadership appears to have a stronger effect on psychological climate and job satisfaction than on OCB. Second, we provide empirical evidence that servant leadership affects job satisfaction through the partial mediation mechanism of psychological climate at the individual level of analysis in a for-profit organizational context. Finally, this study confirms that servant leadership (sOCB and pOCB in Study 2 in Figure 3), job satisfaction (in Study 1 in Figure 2), and psychological climate (in Study 2 in Figure 3) positively and significantly affect OCB. Interestingly, psychological climate (in Study 1 in Figure 2) also negatively and significantly affects OCB. When servant leadership, job satisfaction, and psychological climate are considered together as the antecedents of OCB, only one of the constructs, not all three, emerges as a positive and significant antecedent of OCB. Thus, the other two constructs are less relevant predictors of OCB, depending on the responder for the response variable and the time at which the data for the predictor and response variables were collected.
Theoretical contributions
We focused on servant leadership to advance our understanding of employees’ attitudes, psychological climates, and behaviors at work. Moreover, given the increasing importance of servant leadership and the lack of both theoretical and empirical research at the individual level of analysis on this topic in for-profit organizations (except for a few studies, e.g., Barbuto & Wheeler, Reference Barbuto and Wheeler2006; Chu, Reference Chu2008; Gonzalez & Garazo, Reference Gonzalez and Garazo2008; van Dierendonck, Reference van Dierendonck2011), we theoretically developed and empirically tested a model to contribute to our understanding of servant leadership. Our research demonstrates that followers reciprocate their servant leaders’ behaviors by demonstrating similar serving behaviors, leading to desired employee behaviors that foster a psychological and social environment in which relevant organizational tasks can be accomplished. Followers respond to servant leadership by demonstrating OCB through reciprocation and altruism, as described by SET. Accordingly, servant leadership has a positive impact on employees’ lives, and it is thus well accepted by employees. In addition, this study shows that leadership consistently and positively contributes to job satisfaction and psychological climate. Therefore, our findings demonstrate that servant leadership positively contributes to OCB, job satisfaction, and psychological climate. Moreover, as its most interesting finding, this study reveals that leadership has a stronger influence on psychological climate and job satisfaction than on OCB. We attribute this result to the unique ‘servant first’ aspect of servant leadership, which affects the inner feelings of followers and thus encourages them to psychologically associate themselves with their servant leaders. Therefore, we argue that a unique aspect of servant leadership is its ability to establish an inner attachment within followers that creates a strong psychological climate.
Our focus on servant leadership also advances our understanding of the partial mediating mechanism of psychological climate through which leadership influences job satisfaction. Our study shows that followers reciprocate their servant leaders’ servant-first approach by being intrinsically and extrinsically satisfied in their jobs. Such satisfaction arises because servant leaders aim to make their followers healthier, wiser, freer, and more autonomous, meeting followers’ intrinsic and extrinsic job features. Second, our findings show that this direct effect is dampened by the partial mediation of psychological climate, which arises from servant leaders’ serving behaviors. Together, these two findings imply that leadership contributes both directly and indirectly, through the partial mediation of psychological climate, to followers’ job satisfaction. Thus, employees’ perception of their work environment, which is shaped by their leaders’ competence and consistency and employees’ commitment, cooperation, and coordination, absorbs some of the direct effect of leadership on job satisfaction. This absorption occurs because leadership positively contributes to the creation of a social environment in which followers understand that their leaders’ highest priority is to serve their followers’ most important individual needs. Followers also individually experience and attribute meaning to the service of their leader, and they obtain information from their peers in the social environment regarding their servant leader’s service, further favorably shaping their attitudes toward the leader. Therefore, the information and experience that followers gain in their social environment partially shapes their leader’s influence on their job satisfaction.
Our results regarding the effects of servant leadership, job satisfaction, and psychological climate on OCB further advance our understanding of OCB. The results show that all three constructs individually, positively, and significantly contribute to OCB, despite the two different data collection procedures that we applied in the two studies. Accordingly, our results strongly support the theoretical argument (e.g., Barbuto & Wheeler, Reference Barbuto and Wheeler2006; van Dierendonck, Reference van Dierendonck2011) that leadership leads to OCB. Additionally, our results support the argument that psychological climate is an antecedent of OCB (van Dierendonck, Reference van Dierendonck2011). Regarding the servant leadership-OCB relationship, we argue that servant leadership affects OCB through reciprocity, as described by SET, meaning that the logic applies that ‘because you serve me, I will serve you.’ Therefore, leadership works positively to affect OCB. This result was expected because people desire to serve those who help them eliminate hurdles or obtain benefits at work. With respect to the psychological climate-OCB relationship, we argue that when peers remark positively in their social interactions at work about their servant leaders’ competence and consistency in helping followers, these statements structure the followers’ attentional processes regarding the value of helping others. Thus, the followers interpret the value of helping others as genuinely important for the sake of everyone in the organization, and they act upon it accordingly, thus increasing OCB at work. In addition, because the followers cognitively process the serving behavior of their servant leaders as critical and important to the organization’s well-being, they show similar serving behaviors, thus increasing their OCB at work. Because the followers have repeatedly observed the serving behaviors of their servant leaders, they also accept these behaviors as legitimate and apply them to serve others at work, thus increasing OCB at work further.
This study contributes to the literature by showing that leadership, but not job satisfaction and psychological climate, leads to sOCB and pOCB when the three items are considered together as antecedents of sOCB and pOCB (see Study 2, Figure 3 for sOCB and pOCB). We also reveal that psychological climate, but not servant leadership and job satisfaction, is a positive and significant antecedent of eOCB. These findings also show how the source of data can affect the results and why considering CMV is important. In Study 1, only job satisfaction, not leadership and psychological climate, was a significant and positive predictor of OCB. In Study 2, when the same individual reported both OCB and the predictor variables, psychological climate emerged as the sole positive and significant antecedent of OCB, making the effects of leadership and job satisfaction nonsignificant. When OCB was reported by the individual employee’s supervisor and peers, leadership emerged as the sole positive and significant antecedent of OCB, making the effects of job satisfaction and psychological climate nonsignificant and, as a result, less relevant for predicting OCB. If we accept that supervisors and peers provided objective evaluations of employees’ OCB, then based on our empirical evidence, this paper extends existing knowledge by demonstrating that leadership fosters employees’ OCB. This result shows the power of the servant-first nature of leadership in fostering employees’ OCB vis-à-vis job satisfaction and psychological climate. Given that servant leadership also individually affects job satisfaction and psychological climate, we argue that examining employees’ workplace attitudes, psychological climates, and behaviors from a servant leadership perspective advances our understanding of and theory on employees’ workplace attitudes and behaviors in a for-profit organizational context.
How do we expand SET? The underlying principle of servant leadership is to meet subordinates’ highest priority needs to help them grow personally, indicating that the servant leader is a servant first and not a leader first. In the leader-first context, the social exchange proceeds as follows: a leader supplies rewarding services to a subordinate, who feels obligated to reciprocate, and in return, the subordinate furnishes benefits back to the leader (providing a path for service exchanges of leader→follower→leader). The same logic applies in the servant leadership context, but with an extended process. Accordingly, the servant leader supplies rewarding services to the subordinate, who feels obliged to reciprocate, and as a result, the subordinate emulates the servant leader and furnishes benefits to their subordinates in return. The subordinate does not furnish benefits to the servant leader at this stage because the servant leader expects that the followers will be better off serving their followers. Finally, the last subordinate who is served by the immediate servant leader feels obliged to serve the servant leader, who initiated the serving process (providing a path of extended service exchange of servant leader→follower (becomes a servant leader)→follower (becomes the subsequent servant leader)→…→follower at the lowest level of the organizational hierarchy (becomes the last servant leader)→original servant leader). Therefore, we expand the reciprocal exchange principle of SET to also include an extended-reciprocal exchange principle in the servant leadership context. This extended exchange occurs because the unique tenet of servant leadership is the servant-first approach to making the served healthier, wiser, freer, and more autonomous so that they can also become servant leaders. Thus, the servant leader does not expect her/his followers to feel obliged to immediately return the service but instead expects them to serve others so that they become servant leaders as well. Our extension is not contrary to but consistent with the definition of social exchange provided earlier in this work. We argue that reciprocation results from a chain of exchange processes that occur at a different level of the organizational hierarchy in which each servant leader serves to encourage her/his followers to become servant leaders. In the end, all of the servant leaders at all of the different levels of the organizational hierarchy accept and consequently serve the servant leaders so that they are empowered to lead the organization.
Managerial implications
Although it is implied, it merits repeating that an important practical implication of this study is that organizations should devote resources to the development of servant leadership. When servant leadership is established, an organization develops the desired outcomes in the form of OCB, job satisfaction, and psychological climate. Our results show that servant leadership is more helpful in developing an individual psychological climate and job satisfaction than in developing OCB when servant leadership is considered the unique antecedent to the three constructs. We also add that if the sole objective of management is to achieve a higher level of job satisfaction and an improved psychological climate, then servant leadership will help to achieve these objectives. If the sole objective is to achieve a higher level of OCB, servant leadership helps beyond job satisfaction and a positive psychological climate. In any case, managers should invest more in psychological climate because it partially mediates the relationship between servant leadership and job satisfaction. Investing in psychological climate requires that managers devote more time to following through on their commitments, including employees in decision making, creating good working relationships with their followers, and so forth. Overall, our discussion suggests that investing in servant leadership contributes to desirable employee attitudes, behaviors, and psychological climates at work.
Conclusion
The results of this study indicate that servant leadership has both direct and mediating effects on employees’ discretionary attitudes and behaviors at the individual level of analysis in a for-profit organizational context. Thus, this study furthers the understanding of the effect of leadership on employees’ workplace outcomes. Regarding the strengths of the present study, first, we benefit from two sets of data that were collected through two different procedures. Our data collection procedures allowed us to detect CMV, providing us with information about which variables were affected by and which variables were free of CMV as well as the degree and direction of these effects (see Figures 2 and 3). Given that no reliable statistical method for detecting and correcting CMV is available (e.g., Richardson, Simmering, & Sturman, Reference Richardson, Simmering and Sturman2009; Conway & Lance, Reference Conway and Lance2010), this study contributes to the literature by showing that CMV must be seriously considered during the design of a study.
Second, the collection of data from 12 different organizations in Study 1 and 15 different organizations in Study 2 from a variety of industries allows us to generalize our results across industries. Third, we use random sampling by collecting data from many organizations, which is very difficult to achieve. Fourth, we enhance the growing literature on servant leadership, job satisfaction, psychological climate, and OCB in non-Western settings. Fifth, another strength of this study is the very high response rate (93% in Study 1 and 88% in Study 2) for our survey questionnaires. Finally, our study verifies the consistency of the scales that were chiefly developed in Western settings and that now have been applied to a non-Western setting.
As with many studies, this study has some limitations that should be addressed in future research. We present three of the limitations here. First, we collected our data from a variety of industries in Turkey. The world includes many cultures that mostly differ for each country, and data on different cultures could provide a broader understanding and more significant test of our theoretical model. Thus, the results of cross-national, cross-cultural research testing out model may differ from the findings of this country- and culture-specific study. Second, we did not control for differences between the samples of our two studies (there are multiple organizations in each sample). Study 1 included 12 companies, and Study 2 included 15 companies. We previously stated that deliberately used data on multiple organizations to minimize contextual restrictions. The implication of this research design preference is that there may be attributes of the organizations that better explain our outcomes. Finally, we used only one mediator (psychological climate) in our model. For-profit organizations include many boundary conditions (e.g., organizational culture, group formations) that are specific to each organization, and different boundary conditions (if they are included in the form of multiple mediators and/or moderators in our theoretical model) could provide a more fully elucidated theoretical model.
This study is based on a theoretical model that was tested by using two sets of data. Our model includes leadership as an individual-level construct to elucidate how leadership drives employees’ discretionary attitudes and behaviors, as a more employee-centered type of leadership style is currently being demanded. While studies in the literature have considered servant leadership as a unit-, group-, and team-level construct to study its effects on certain employee attitudes and behaviors (e.g., Ehrhart, Reference Ehrhart2004; Walumbwa, Hartnell, & Oke, Reference Walumbwa, Hartnell and Oke2010; Reference Liden, Wayne, Liao and MeuserLiden et al., 2014b), this body of literature remains limited, and in particular, studies reporting an individual level of analysis in a for-profit organizational context are lacking in the literature. Future research should thus focus on how servant leadership affects a broad array of employee attitudes and behaviors, such as task performance, life satisfaction, commitment, turnover intention, and job engagement, at an individual level of analysis. A better understanding of the influence of leadership on a wide variety of employee attitudes and behaviors could inform solutions that better address the demands for more people-centered management, a caring style of leadership, and concern for the success of all organizational stakeholders. It would also be very useful to examine additional mediators and moderators of the relationships found in this study, as an examination of mediating mechanisms seems to be a necessary direction for this line of inquiry, given our results. Finally, it would be very interesting to test our model with respect to other leader-first approaches to elucidate whether leader-first or follower-first leadership better explains employee attitudes and behaviors given the current desire for more follower-centered management.
Acknowledgements
The authors thank editor Peter Galvin, associate editor Stephen Teo and three anonymous reviewers for their constructive comments on previous versions of this present article.
