Setting the Scene: Partners’ Leadership Behavior and Employees’ Perceptions of Work Performance in Professional Service Firms

The effect of leadership behavior on work performance is highly context sensitive. We address this notion by investigating leadership behavior in one important but understudied organizational context—namely, professional service firms (PSFs). We examine how partners’ leadership behavior in a PSF relates to employee self-leadership, creative climate, and work performance (N = 442). The results show that partners’ consideration leadership behavior is positively related to employees’ perceived work performance. Moreover, partners’ consideration and intellectual stimulation leadership behavior are especially important drivers of self-leadership and creative climate in a PSF, which in turn are positively related to employees’ work performance. Theoretical and practical implications are discussed.

The search for leadership behaviors that improve organizational performance has aroused interest for decades. A key takeaway from prior studies on leadership behaviors is that no one behavior is the best (Derue, Nahrgang, Wellman, & Humphrey, 2011;Piccolo et al., 2012). Rather, the effects of leadership behavior on employees' work performance is highly context sensitive (Bryman, Stephens, & Campo, 1996;Johns, 2006;Porter & McLaughlin, 2006;Vroom & Jago, 2007). Thus, to inform leaders on how to enhance employees' work performance, understanding context is imperative.
One important context is that of professional service firms (PSFs). Indeed, the professional service industry in the United States contained almost 900,000 firms in 2014, employing more than 8.5 million people (SelectUSA, 2017). PSFs are firms that rely on a workforce with substantial expertise to produce and apply customized client solutions in everyday service delivery (Malhotra, Smets, & Morris, 2016;Nordenflycht, 2010). Employees in PSFs are expected to draw on their expertise to solve complex client problems, with the scope of creativity and autonomy shaped by their partners (Donnelly, 2009;Florida, 2002;Nordenflycht, 2010). Partners reside at the highest level in the hierarchy and are typically responsible for the "brain" work (Malhotra et al., 2016). As such, they "set the scene" for employees to excel, and the way partners use their leadership behaviors may directly affect PSFs' creative capacity and work performance. Unfortunately, knowledge about partners' leadership behavior in this context is surprisingly limited.
Thus, the purpose of this article is twofold. First, we examine how partners' more traditional (initiating structure and consideration) and recent (intellectual stimulation) leadership behaviors relate to employees' perceived work performance in PSFs. Second, we suggest that two components play an especially important role in PSFs: partners' ability to (a) develop employees' self-leadership and (b) generate a creative climate. Given the highly autonomous context of PSFs, facilitating self-leadership in which employees use specific sets of behaviors and cognitive strategies to improve their work performance (Neck & Houghton, 2006) is likely to be important. Furthermore, as employees' creativity can depend heavily on the climate (Isaksen & Ekvall, 2010;Mumford & Gustafson, 1988), partners may facilitate a culture that makes asking questions and making mistakes acceptable.
We use social learning theory (Bandura, 1977(Bandura, , 1986 to explain why partners' leadership behaviors are important for self-leadership, creative climate, and work performance, and we contribute to research on partners' leadership 781813J LOXXX10.1177 1 NHH Norwegian School of Economics, Bergen, Norway 2 BI Norwegian School of Management, Oslo, Norway 3 Norwegian Defense University College, Oslo, Norway behaviors by testing our proposed research model in a large international PSF. In the following, we describe the three partner leadership behaviors of interest and how they may be related to work performance of employees in PSFs. We then present the concepts of self-leadership and cultural climate and discuss their roles as potential mediating mechanisms. Next, we describe the study conducted in a large international PSF and show how partner leadership behavior is important for employee self-leadership, cultural climate, and employees' work performance.

Partners' Leadership Behaviors and Work Performance
In the study of partners' leadership behavior in PSFs, we integrate initiating structure and consideration leadership behaviors with intellectual stimulation from transformational leadership behavior (Derue et al., 2011;Ekvall, 1996;Judge, Piccolo, & Ilies, 2004;Lambert, Teppe, Carr, Holt, & Barelka, 2012;Piccolo et al., 2012). The first two leadership behaviors, initiating structure and consideration, were originally identified by researchers at Ohio State University at the beginning of the 1950s. Initiating structure is a leadership behavior that involves clarifying task responsibilities, providing direction, and communicating expectations to subordinates; it also includes directing and structuring subordinates' tasks (Bass & Stogdill, 1990). This behavior was largely absent in leadership literature for approximately 30 years (Judge et al., 2004) but has since made a comeback and remains an important leadership behavior (Keller, 2006), due to its predictive capacity (Judge et al., 2004).
The leadership behavior consideration refers to concern for employees' well-being, expression of support, and warmth and approachability (Fleishman, 1973). Leaders high in consideration show concern and respect for followers, are friendly and approachable, are open to input from others, and treat all group members as equals (Bass & Stogdill, 1990). Judge et al. (2004) found that consideration correlates more strongly with followers' satisfaction and motivation, while initiating structure correlates slightly stronger with criteria related to performance. Overall, meta-studies find that initiating structure and consideration are important for work performance (Derue et al., 2011;Judge et al., 2004). However, previous research also reports that employees in PSFs tend to dislike control-based leadership (Starbuck, 1992).
Interest in initiating structure and consideration has slowly given way to transformational leadership (Bass, 1985;Burns, 1978). In transformational leadership, intellectual stimulation is a leadership behavior that inspires followers' efforts to be creative. Intellectual stimulation includes behaviors such as questioning assumptions, reframing problems, and approaching old situations in new ways (Bass, 1985). When using this leadership behavior, partners include employees in the process of finding new solutions and avoid criticizing their ideas. Previous research finds that intellectual stimulation is conceptually distinct from initiating structure and consideration (Piccolo et al., 2012), and research has called for studies to focus on the specific dimensions (e.g., intellectual stimulation) of transformational leadership behaviors (van Knippenberg & Sitkin, 2013). Derue et al. (2011) andPiccolo et al. (2012) compared transformational leadership with the Ohio State University leadership behaviors (initiating structure and consideration) and found that transformational leadership and consideration were important predictors of employee outcomes. In addition, evidence shows that intellectual stimulation is related not only to employee creativity but also to subordinates' perceptions and organizational measures of work performance (e.g., Lowe, Kroeck, & Sivasubramaniam, 1996).
Social learning theory (Bandura, 1977(Bandura, , 1986) may help explain how the leadership behavior of partners in PSFs affects employees' work performance. This theory proposes that employees learn how to respond to the environment by observing leaders and emulating their behavior in their own work. Bandura (1977Bandura ( , 1986 proposed that humans copy the behavior of others, without the need for rewards or punishment to encourage them to do so, through behavioral modeling. If the behavior copied is successful, they are more likely to act in that way. At work, employees cognitively think about their leaders' behavior and its consequences before they imitate it or not. Behavioral modeling includes four cognitive processes at work. First, the leader must attract employees' attention; employees must notice and be exposed to the leader's behavior. The second factor is retention; employees must remember the leader's behavior in their own situations to be able to perform. The third is reproduction; employees must have the ability to perform the behavior the leader demonstrated. The fourth behavior is motivation; employees consider the reward and their willingness to imitate the leader's behavior. Overall, partners' initiating structure behavior should help employees improve their work performance (Judge et al., 2004). Partners' consideration behavior is also important, providing models employees can learn from and emulate partners to meet the challenges of their work. Partners' intellectual stimulation behavior helps employees feel safe to find new ways to work, learn from failures, and share ideas with colleagues. Thus, this type of behavior is important for employees to learn from their mistakes and improve ways to deliver results, increasing work performance. Thus, we suggest that partners' role-modeling of the three types of leadership behaviors will be positively related to work performance.
Hypothesis 1: Partners' initiating structure will be positively related to employees' work performance in a PSF. Hypothesis 2: Partners' consideration will be positively related to employees' work performance in a PSF. Hypothesis 3: Partners' intellectual stimulation will be positively related to employees' work performance in a PSF.

Self-Leadership as Mediator
The concept of self-leadership was inspired by Kerr and Jermier's (1978) notion of "leadership substitution" and emerged during the 1980s as an expansion of self-management (Manz & Sims, 1980). The fundamental idea is that employees can take on the responsibility for many of the influence processes normally carried out by external leaders (Lovelace, Manz, & Alves, 2007). A key aspect of this process is that self-leadership prompts employees to take control, which potentially leads to increased effectiveness and performance. Self-leadership strategies are normally grouped into three distinct areas (Neck & Manz, 2013). First, behavior-focused strategies include self-observation, self-goal setting, self-cueing, self-reward, and self-correcting techniques that help employees perform necessary but unpleasant tasks. Second, natural reward strategies focus on how employees can achieve desirable outcomes by using natural rewards or the environment to foster intrinsic motivation and enjoyable aspects of a task or an activity. Third, constructive thought patterns refer to self-analysis, mental imagery, and self-talk strategies that aim to replace dysfunctional beliefs and adopt more constructive thought processes in advance.
Several researchers have found empirical evidence of a positive relationship between self-leadership and work performance (for reviews, see Neck & Houghton, 2006;Stewart, Courtright, & Manz, 2011). However, few studies have examined self-leadership in PSFs, despite evidence suggesting that self-leadership is favorable in such contexts. Roberts and Foti (1998), for example, found that employees who reported low self-leadership were more satisfied in highly structured work environments while employees reporting high self-leadership were more satisfied in more unstructured work environments.
Partners' leadership behaviors may evoke self-leadership in employees through social learning. In the model of "superleadership," Manz and Sims (2001) described how leadership helps others lead themselves. A primary aim of the superleadership approach is to role-model behaviors that develop and encourage employees to be self-leaders (Manz & Sims, 2001). Conceptually, Houghton and Yoho (2005) suggest that empowering leadership behaviors encourages employees' self-leadership. Empirically, in their study of four Christian mission organizations, Amundsen and Martinsen (2015) found a direct positive relationship between empowering leadership behavior and self-leadership and between self-leadership and work performance.
We suggest that partners' initiating structure is associated with higher employee self-leadership. Employees high in self-leadership use self-goal setting, self-observation, and self-reward for their performance. As such, partners' directing and coordinating of task goals may help encourage and foster employees' self-leadership. We also suggest that partners' consideration behavior, by showing concern and respect for employees, contributes to higher self-leadership (Stewart et al., 2011). Finally, we suggest that partners' intellectual stimulation behavior, such as encouraging the exploration of new ways of doing work and learning from failures, contributes to employee selfgoal setting, self-observation, self-rewarding, and higher intrinsic motivations (Furtner, Baldegger, & Rauthmann, 2013). Empirically, Furtner et al. (2013) found that leaders' self-leadership was associated with transformational and transactional leadership behaviors. Andressen, Konradt, and Neck (2012) found that self-leadership is a process factor that mediates the relationship between transformational leadership and employee motivation. However, research has not examined the association between partners' initiating structure, consideration, and intellectual stimulation leadership behavior and self-leadership and its effect on work performance.

Hypothesis 4:
Employees' self-leadership will positively mediate the relationship between partners' (a) initiating structure, (b) consideration, and (c) intellectual stimulation and work performance in a PSF. Ekvall (1991) defined organizational climate as the observed and recurring patterns of behavior, attitudes, and feelings that characterize life in an organization. A creative climate encourages people to develop new ideas and helps the organization expand to achieve greater effectiveness (Ekvall, Arvonen, & Waldenström-Lindblad, 1983). Part of leadership is creating an appropriate climate in which employees can share and build on one another's ideas and suggestions (Isaksen & Ekvall, 2010). Denison (1996) evaluated 10 years of organizational climate and culture research and defined organizational climate as something that is changeable, is subject to direct control by leaders, and includes aspects of the social environment that is consciously perceived by organizational members. Moreover, Ekvall (1996Ekvall ( , 1997 found that creative climate explains innovative differences in organizations (i.e., number of patents obtained and success in developing new products). Although researchers have learned much about the determinants of creative climate and their consequences, scant research has examined the roles of leadership behaviors (Sundgren, Dimenäs, Gustafsson, & Selart, 2005;Sundgren, Selart, Ingelgård, & Bengtson, 2005).

Creative Climate as Mediator
Conceptually, several authors suggest the importance of examining how leadership behaviors influence a creative climate (Mumford & Gustafson, 1988;Mumford, Scott, Gaddis, & Strange, 2002;Oke, Munshi, & Walumbwa, 2009). Employees can learn from and mirror leaders with the intent to gain improvements and develop new ideas (Amabile, Conti, Coon, Lazenby, & Herron, 1996). A considerate partner may encourage employee development and participation to stimulate creativity (Oldham & Cummings, 1996). Similarly, partners' intellectual stimulation in modeling curiosity and generating new ideas will also stimulate employee creativity (Oldham & Cummings, 1996). As such, leaders who show concern for employees and stimulate creative behavior may achieve a supportive environment (Sarros, Cooper, & Santora, 2008;Shalley & Gilson, 2004). On the negative side, however, partners' initiating structure can undermine creativity by being too controlling or directive (Oldham & Cummings, 1996).
Empirically, Xenikou and Simosi (2006) examined 293 employees in 32 business units from a financial organization in Greece and found that an achievement cultural orientation mediated the relationship between transformational leadership and work performance. In another study, Ogbonna and Harris (2000), using a multi-industry sample from the United Kingdom, examined the mediating role of an innovative culture and found that participative leadership behavior had a positive indirect effect on work performance through an innovative culture while instrumental leadership had a negative indirect effect. Notwithstanding these previous studies, few studies have examined the mediating role of creative climate in a PSF context. However, we argue that the links among partner leadership behaviors, creative climate, and employees' work performance are central to understanding PSFs and making them effective. We suggest that initiating structure will have a negative indirect effect while consideration and intellectual stimulation will have a positive indirect effect.
Hypothesis 5: A creative climate will negatively mediate the relationship between (a) partners' initiating structure and work performance and positively mediate the relationship between (b) partners' consideration and (c) intellectual stimulation and work performance in a PSF.

Sample and Context
We recruited participants from a multinational PSF in Norway. The firm is one of the largest suppliers of professional services in auditing, consultancy, and legal services worldwide. We distributed 1,053 questionnaires, 559 of which were completed (response rate 53%). We restricted the sample to people who worked with clients (thus excluding administrative employees) and partners, which left us with a final sample of 442 participants. Of these, 47.5% were women, and the average age was 33.15 years (SD = 8.53). On average, participants had worked for the company for 4.08 years (SD = 5.39). Participants had a mean of 5.43 years (SD = 1.40) of university/higher education.
The firm has four basic levels: analyst, consultant, manager, and partner. Each step below partner also has a senior level (i.e., senior manager). Employees are expected to progress within fixed time frames, though only a few will ultimately become partners. The hierarchy reflects overall competence and experience. Those at the lower levels are typically less experienced and paid less than those above them. Partners reside at the highest level in the hierarchy, and they possess the most experience in the company; they are not just in charge of the operations but also the decisions makers in the company.
A partner in a PSF is a person who owns a portion of the firm. This is the main difference between a partner and other managers such as senior managers, who are salaried employees. Partners do not have a monthly "salary" but rather a monthly "income" determined by the revenue they bring to the firm. Partners also have signing responsibilities, while salaried employees cannot sign on behalf of the firm. Employees are motivated to work hard and economize on monitoring costs, because the billable hours are an effective mechanism that regulates promotion to a partner. Those who fail to become partners are expected to move on to another employer.
Each partner receives a competence sector for which he or she has revenue responsibility. In addition, all employees have one immediate partner within their competence area. This means that partners rely on their leadership behaviors to influence people working under them to achieve their overall targets and to generate profits. In addition, partners share their knowledge with the people under them (i.e., manager, consultants) to enable them to take on complex tasks and earn higher fees. For partners, this often means trading off between time and capacity to solve complex client problems, win new businesses, and share knowledge with employees. Increased delegation to senior managers means less time spent on knowledge sharing with employees and more time prioritizing business development. As such, partners stimulate a more creative climate when they enhance the transfer and growth of knowledge to employees. In addition, managers and consultants often operate on customer sites outside the firm, and self-leadership strategies are important for their work performance. Therefore, in this study we examine partners' leadership as the main source of leadership behaviors, as they are in the most prominent position to enhance self-leadership and develop a creative climate to improve work performance.

Procedures
We collected data using a web-based survey instrument (Qualtrics). To translate the items in the questionnaire, we followed Brislin's (1986) procedure. We first translated the items from English to Norwegian and then back-translated them from Norwegian to English to ensure equivalence. The study was approved by the Norwegian Social Science Data Services and supported by the company's top management. The questionnaire was prepared in collaboration with the HR department and tested on a few partners and employees in the firm to ensure that the questions were relevant and interesting. The company provided a mailing list of all employees in the Norwegian part of the firm, and each respondent was invited to participate by the senior HR manager. Employees were assured that all collected data would be completely confidential and have no bearing on their careers.

Measures
The items for all the variables in the study are reported in the appendix.
Leadership Behaviors. We measured leadership behaviors by asking respondents to assess, on each leadership item, their closest partner on a 5-point Likert-type scale (1 = Strongly disagree, 5 = Strongly agree). We measured initiating structure and consideration using a six-item version developed by Lambert et al. (2012) on the basis of the original Leader Behavior Description Questionnaire (Stogdill, 1963). An example of the three items measuring consideration is "My partner is acting friendly and approachable," and an example of the three items measuring initiating structure is "My partner lets me know what is expected of me." We measured intellectual stimulation using a three-item version from the Multifactor Leadership Questionnaire (MLQ Form 5x; Bass & Avolio, 1995). The MLQ Form 5x is copyright protected but is available from www.mindgarden.com. The internal consistency for initiating structure, consideration, and intellectual stimulation was .76, .88, and .90, respectively.
Self-Leadership. We measured employees' self-leadership using a nine-item version obtained from the Abbreviated Self-Leadership Questionnaire (Houghton, Dawley, & DiLiello, 2012) on a 5-point Likert-type scale (1 = Strongly disagree, 5 = Strongly agree). A sample item is "I establish specific goals for my own performance." The internal consistency was .81.
Creative Climate. We measured creative climate using a sixitem short-version scale adapted from Ekvall's (1996) creative climate instrument on a 5-point Likert-type scale (1 = Strongly disagree, 5 = Strongly agree). A sample item is "Workers in the company can come up with new ideas and opinions without being criticized." The internal consistency was .90.
Work Performance. We measured work performance using a six-item scale adapted from a 10-item measure to assess work effort (three items) and work quality (three items; Kuvaas, Buch, Dysvik, & Haerem, 2012) on a 5-point Likert-type scale (1 = Strongly disagree, 5 = Strongly agree). Examples are "I try to work as hard as possible" and "The quality of my work is top notch." The internal consistency for work effort and work quality was .72 and .76, respectively.
Conflict Avoidance. We used conflict avoidance as a marker variable to control for common method variance (CMV), and we measured it using four items adapted from Bresnahan, Donohue, Shearman, and Guan (2009) on a 5-point Likert-type scale (1 = Strongly disagree, 5 = Strongly agree). A sample item is "I avoid conflict if at all possible." The internal consistency of the scale was .67.

Analyses
We analyzed the data in three phases. First, we conducted confirmatory factor analysis on the seven measures to examine their construct validity. We used individual items as observed indicators. The hypothesized seven-factor measurement model consisting of leader behaviors (initiating structure, consideration, and intellectual stimulation), mediating variables (self-leadership and creative climate), and work performance (work effort and work quality) fit the data well (χ 2 [381] = 742.184; χ 2 /df = 1.95, p ⩽ .001; comparative fit index [CFI] = .94; Tucker-Lewis index [TLI] = .93; root mean square error of approximation [RMSEA] = .05; standardized root mean square residual [SRMR] = .05). As the data came from a single questionnaire, CMV may have occurred. We therefore tested for CMV in our sample. Second, we conducted linear regression analyses, with work effort and work quality as the dependent variables. We controlled for gender and tenure to evaluate the contributions on work performance. Finally, to examine our mediation hypotheses, we relied on structural equation modeling (SEM) and used Mplus to test the indirect effects (Muthén, Muthén, & Asparouhov, 2016). We used bootstrap estimates to evaluate the significance of the indirect effects.

Test of CMV
To test for CMV, we followed the marker procedure by Williams, Hartman, and Cavazotte (2010). Specifically, the marker variable approach requires that marker variables are theoretically unrelated to the variables in a model. As noted, we use conflict avoidance as our marker variable. The literature on conflict avoidance is rather limited, and little empirical data have shown its effectiveness (Tjosvold, 2008). In addition, we find no conceptual models linking conflict avoidance with the variables in our study (i.e., leadership, self-leadership, creative climate, and work performance).
As Table 1 shows, the first model examined the CFA model of the seven variables and the marker variable in our study. The second model evaluated the Baseline model, in which the seven variables correlated with one another, but it also had an orthogonal marker variable, whose indicators had fixed factor loadings and fixed error variances. We used the unstandardized estimates from the CFA model as fixed values for the factor loadings and error variances for the marker variable indicators in the Baseline model. We also constrained the loadings between the substantive variables and the marker variable to be zero. The third model, the Method-C model, is identical to the Baseline model, in which the marker variable is orthogonal and the measurement parameters associated with its indicators are fixed. However, the Method-C model has an additional factor loading from the method marker variable to each indicator in the model, and all loadings are forced to be equivalent in values to appropriately reflect the assumption of the CMV model of equal method effects. A comparison of the Method-C model with the Baseline model provides a test of the presence of method variance associated with the marker variable.
The results show that the Method-C model fit significantly better than the Baseline model (Δχ 2 [1] = 8.85, p < .05). The fourth model, Method-U, is similar to the Method-C model, except that the factor loadings from the marker variable to the substantive indicators are freely estimated, thus reflecting that the assumption of unrestricted method variance in the marker variable is differentially related to the substantive variables. A comparison of the Method-C and Method-U models provides a test of the key differences between the CMV and unrestricted method variance models and the assumption of the equal method effects. The Method-U model fits significantly better than the Method-C model (Δχ 2 [29] = 63.49, p < .05), suggesting that the method effect did not affect the substantive variables in an equivalent way. Finally, the fifth model, Method-R, uses the obtained unstandardized factor correlations for our substantive variables from the Baseline model as fixed values in the Method-U model (as the Method-U model fit significantly better than the Method C-model). A comparison of the Method-U and the Method-R models provides the statistical test of the biasing effects of our marker variable on substantive relationships. The Method-R model did not fit significantly worse than the Method-U model (Δχ 2 [21] =.73, ns). Previous tests indicated that the marker variable effects were significant and represented significant effects in the Method-U model, but the results of the Method-U and Method-R models indicated that the effects of the marker variable did not significantly bias factor correlation estimates, even though CMV was present. In addition, we used the marker variable as a control in the regression and SEM. All paths maintained their level of statistical significance. Thus, the marker variable did not change the results. Taken together, the results from CMV testing suggest that method variance biases are not likely to confound the interpretations of the results from this study. The Method-U model, which controlled for CMV, provided the best fit to the data (χ 2 [481] = 896.20; χ 2 /df = 1.86, p ⩽ .001; CFI = .93; TLI = .92; RMSEA = .04; SRMR = .05); therefore, we used this model to test the hypotheses in our study.

Results
Descriptive statistics, coefficient alphas, and correlations for the variables used in the study appear in Table 2. All scale reliabilities exceeded .70, as recommended by Nunnally (1978). However, the marker variable conflict avoidance had a Cronbach's alpha of .67. Table 3 reports the R 2 differences and related F values for each step in the hierarchical regression analyses. As Table 3 and Step 2 show, partners' initiating structure did not predict employees' work effort or work quality, when the three leadership behaviors were present in the model. Thus, Hypothesis 1 is not supported. In support of Hypothesis 2, partners' consideration was positively related to both employees' work effort (β = .14, p < .02) and work quality (β = .15, p < .02). For the test of Hypothesis 3, partners' intellectual stimulation did not predict employees' work effort or work quality. Thus, Hypothesis 3 is not supported.
Hypotheses 4 and 5 were designed to test whether selfleadership and creative climate mediated the relationship between leadership behavior and work performance. As the two control variables did not affect any results in the full model (see Table 3), we did not include the control variables in the remainder of the analyses. Figure 1 presents the overall structural model with path coefficients for leadership behavior and work performance. The hypothesized model fit the data well (χ 2 [482] = 905.701; χ 2 /df = 1.88, p ⩽ .001; CFI = .93; TLI = .92; RMSEA = .05; SRMR = .05). The SEM model reports that consideration is positively related to creative climate (β = .23, p < .01) and that intellectual stimulation is positively related to self-leadership (β = .29, p < .01) and creative climate (β = .47, p < .01). Self-leadership had a positive relationship to work effort (β = .32, p < .01) and work quality (β = .32, p < .01), while creative climate showed a positive relationship to work effort (β = .27, p < .01), but not to work quality.

Work performance
Work effort Work quality Step 1 Step 2 Step 3 Step 1 Step 2 Step 3  Table 4 presents the total, indirect, and direct effects and bootstrapped estimates for each of the mediation hypotheses. Hypothesis 4 stated that self-leadership would positively mediate the relationship between (a) initiating structure, (b) consideration, and (c) intellectual stimulation and work performance. As Table 4 shows, the model provides partial support for Hypothesis 4, as self-leadership positively mediates the relationship between intellectual stimulation and work effort (β = .09, p < .03) and between intellectual stimulation and work quality (β = .09, p < .03). Hypothesis 5 stated that creative climate would negatively mediate the relationship between (a) initiating structure and work performance and positively mediate the relationship between (b) consideration and (c) intellectual stimulation and work performance. As Table 4 shows, the model provides partial support for Hypothesis 5, as creative climate positively mediated the relationship between consideration and work effort (β = .06, p < .03) and between intellectual stimulation and work effort (β = .12, p < .01). Furthermore, bootstrap analyses showed support for Hypothesis 5. As Table 4 shows, the indirect effect of consideration on work quality through creative climate was different from zero (.002, .104), as was the indirect effect of intellectual stimulation on work quality through creative climate (.000, .175).
We also investigated several alternative models that were less likely to fit the data but were plausible from a theoretical standpoint. In the first alternative model, we tested a model in which self-leadership and creative climate had a direct effect on work effort and work effort had a relationship to work quality. This model provided an adequate fit to the data (χ 2 [484] = 915.644; χ 2 /df = 1.89, p ⩽ .001; CFI = .93; TLI = .92; RMSEA = .05; SRMR = .06) but was not significantly better than the hypothesized model (Δχ 2 [2] = 9.943, p > .05). Overall, the hypothesized model was more consistent with the data.

Discussion
In this article, we used social learning theory (Bandura, 1977(Bandura, , 1986 and examined a theoretical model of the novel relationships between partners' leadership behaviors, via self-leadership and creative climate, and employees' perception of work performance, which we subsequently tested in the context of a PSF. We found that partners' consideration leadership behavior predicted both work effort and work quality. Furthermore, we found that partners who used an intellectual stimulating leadership behavior and encouraged employees to lead themselves elicited higher work performance from their employees. Finally, we also found that creative climate served as a mediator between consideration and work performance and between intellectual stimulation and work performance.

Theoretical Implications
This study provides three important theoretical contributions. First, we contribute to the literature on leadership behavior by examining the leadership behaviors from the partner's role in a PSF. To our knowledge, this study is the first to address this notion. Indeed, studies on leadership behavior in PSFs are limited in general. Investigating PSF partners is important as they have a somewhat different role than what traditional hierarchical leadership models indicate. That is, partners' leadership roles help set the scene to communicate to employees what is right or important in a work context and to influence climate perceptions and employee behavior (Bandura, 1977(Bandura, , 1986. In particular, their roles of challenging employees' self-leadership and creating a creative climate may be especially important in PSFs, and our results unpack the partners' efforts in this regard. Second, we combine Ohio State University's leadership behaviors and intellectual stimulation from transformational leadership theory. Our findings confirm that the three leadership behaviors have distinct implications for work performance (Derue et al., 2011;Ekvall & Arvonen, 1991;Judge et al., 2004;Piccolo et al., 2012;Yukl, Gordon, & Taber, 2002). Our findings also corroborate previous research indicating that consideration is important for leader effectiveness (Judge et al., 2004;Piccolo et al., 2012). By modeling consideration behavior, leaders induce in followers a focus on respect, approachability, and openness to input from others that results in better work performance. Partners' role-modeling of intellectual stimulation may also be important for work performance in PSFs but, according to our results, only as long as it works through employee self-leadership and creative climate.
Third, we use social learning theory (Bandura, 1977(Bandura, , 1986 and extend previous research by introducing a model that tests self-leadership and creative climate as mediators between leadership behaviors and work performance. We found that self-leadership has an indirect effect between partners' intellectual stimulation and employees' work effort and work quality. This finding implies that partners' rolemodeling of intellectual stimulation leadership behaviors is associated with greater mastery of self-leadership, which in turn leads to greater work performance. Our finding also gives empirical support to the model of superleadership (Manz & Sims, 2001) and extends previous research by Amundsen and Martinsen (2015) by identifying intellectual stimulation as an additional leadership behavior associated with self-leadership and work performance.
We also found that creative climate mediated both the relationship between consideration and work performance and that between intellectual stimulation and work performance. These findings imply that partners' role-modeling of consideration and intellectual stimulation leadership behaviors is associated with a greater creative climate, which in turn leads to a greater work performance. Our results are in line with social learning theory in that partners' leadership behavior influences climate perceptions, which in turn influence employee behavior (Bandura, 1977(Bandura, , 1986. This research finding contributes to extend previous studies on the links among leadership behavior, creative climate, and work performance (Ogbonna & Harris, 2000;Xenikou & Simosi, 2006).

Practical Implications
Our findings suggest that partners in a PSF have an important leadership role that contributes to work effort and work quality. The first practical implication is that leaders may need to think through their consideration leadership behavior. For example, leaders who show concern for their employees' well-being, express support, and display warmth achieve both greater work effort and higher work quality. Our results suggest that PSFs may benefit from training leaders to adopt consideration leadership behaviors to achieve better work performance. A second practical implication is that partners should focus on leadership behaviors that enhance self-leadership and a creative climate. Leaders who use intellectual stimulation should train their employees to apply self-leadership skills (Neck & Manz, 1996). In addition, leaders who use consideration and intellectual stimulation leadership behaviors should facilitate a supportive creative climate. Overall, our results indicate that self-leadership and a creative climate enhance work performance in a PSF.

Limitations and Future Research Directions
As with most studies, our study has several limitations. First, we collected data from a single source using one survey, which may have inflated estimates of correlations between the variables because of CMV. However, we used the marker variable approach (Williams et al., 2010) and did not find any biasing effect of CMV on the paths between our study variables. Ideally, we would have preferred to have information on objective measures, though Levy and Williams (2004) found that performance ratings assessed by supervisors are not necessarily less biased than self-reported measures. Nevertheless, although we did not find any biasing effects of CMV, the marker technique still has limitations (Richardson, Simmering, & Sturman, 2009;Williams & McGonagle, 2016). Therefore, to be on the safe side, future research should collect ratings of leadership, mediators, and work performance from different sources and different periods to test the relationships. Second, the cross-sectional design also limits the opportunity to exclude, for example, the possibility of higher work performance leading to more relational leadership behaviors. To assess causal directions, future research could try to replicate the findings in an experimental design or a longitudinal study. Moreover, data collection of each partner's organizational structure would have allowed us to test our research model as a multilevel model. In addition, future research should pay attention to the various types of leadership behaviors from different types of leaders and their influence on work performance. This is especially important because many PSFs have leaders with different roles. For example, research could examine the different leadership roles of partners and team leaders by testing boundary conditions that may moderate the mediating effects of self-leadership and creative climate on work performance, especially when consideration and intellectual stimulation are applied. Finally, we hope that our study will trigger further research and debate on partners' leadership behaviors in PSFs and how their behavior sets the scene for employees' self-leadership, creative climate, and work performance.

Self-Leadership
I establish specific goals for my own performance. I make a point to keep track of how well I'm doing at work. I work toward specific goals I have set for myself. I visualize myself successfully performing a task before I do it. Sometimes I picture in my mind successful performance before I actually do a task. When I have successfully completed a task, I often reward myself with something I like. Sometimes I talked to myself (out loud or in my head) to work through difficult situations. I try to mentally evaluate the accuracy of my own beliefs about situations I am having problems with. I think about my own beliefs and assumptions whenever I encounter a difficult situation.

Creative Climate
The climate in the company is basically positive and encourages new ideas. Employees in the company can bring up new ideas and opinions without quickly being criticized. The company allows you to solve problems and take actions that you think are most suitable in a given situation. There is a free atmosphere in the organization, where the seriousness of the task can be mixed with unusual ideas and humor. Different opinions, ideas, experience, and knowledge can be discussed in projects. The organization has a dynamic atmosphere.

Work Effort
I try to work as hard as possible. I intentionally expend a great deal of effort in carrying out my job.
I usually don't hesitate to put in extra effort when it is needed.

Work Quality
The quality of my work is top notch. I deliver higher quality than what can be expected from someone with the type of job I have. Others in my organization look at my work as typical high quality work.