Leader autonomy support in the workplace: A meta-analytic review

For many, the motivation and optimal functioning of employees is largely seen to be driven by leader behavior (Gilbert and Kelloway 2014). Thus, there is a growing literature devoted to how leaders can positively affect the motivation and behavior of individual workers (e.g., Barling et al. 2010). Early attention to the topic can be traced to Taylor’s “scientific management”, where subordinate motivation was thought to result from close supervision and controlling practices such as scientifically designed incentive systems (Taylor 1911). Literature later differentiated autocratic from democratic leadership, suggesting that these leadership styles generate different motivational states in followers (Coch and French 1948; Lewin et al. 1939). Autocratic leadership, to the extent that it engenders a climate of fear, was thought to activate fleeting motivational states that persist only while the leader remained physically present. The more consultative democratic style, in contrast, was thought to produce sustained motivated behaviors that persisted even in the absence of the leader. These two contrasting leadership styles set the foundation for later theorizing about the antecedents of intrinsic and extrinsic work motivation (Zaccaro et al. 2008).

More recently, few theories have generated as much scholarly attention on extrinsic and intrinsic forms of motivation as SDT (Ryan and Deci 2000, 2017). SDT is a broad theory of human motivation that concerns individuals’ innate growth tendencies and basic psychological needs, and focuses on the degree to which individual behavior is autonomously motivated or controlled. It has been applied across a variety of research domains, including education, health, sport, parenting, and organizations (e.g., Ng et al. 2012; Reeve 2015; Ryan and Deci 2017; Su and Reeve 2011).

A key focus in the organizational domain has been on the contextual factors that support employee self-determination and basic psychological needs (e.g., Deci et al. 2001; Van den Broeck et al. 2016). Job autonomy, broadly defined as the extent to which individual workers can self-govern how and when they perform the various tasks that make up their job (Hackman and Oldham 1976; Spector 1986), is often identified as a contextual antecedent of self-determination in the workplace (Johns 2006; Ryan and Deci 2017). In addition to the design of jobs (Barrick and Mount 1993), perceived autonomy can stem from interpersonal characteristics, including the leader’s motivating style, which can range from highly supportive to highly controlling (Deci et al. 1989; Gagné et al. 2018; Reeve 2015).

Leader autonomy support refers to a cluster of supervisory behaviors that collectively promote a climate of support and understanding within leader-worker relationships (Reeve 2015). An autonomy supportive style generally involves leaders acknowledging worker perspectives, encouraging self-initiation, offering opportunities for choice and input, communicating in an informational rather than a controlling manner, and avoiding the use of rewards or sanctions to motivate behavior (Baard et al. 2004; Hardré and Reeve 2009; Su and Reeve 2011). It is thought to foster more agentic and self-determined pursuits, as recipients perceive themselves to be the regulators of their own actions, fostering a heightened sense that behavior is internally directed rather than externally controlled (Deci et al. 2017). In contrast, a controlling leadership style involves leaders imposing external constraints on behavior with the intention of compelling individuals to produce specific outcomes (Ryan et al. 1983). A controlling style is often interpreted as prescriptive, inflexible, and rigid, pressuring the employee to think, feel, or behave in particular ways (Ryan and Deci 2017). Deviations from leader demands are often met with corrective or other punitive actions intended to restore behavior back to its desired course. Hence, the style signals to employees that the leader is the initiator of action, shifting the perceived cause of one’s behavior to an external source (Deci et al. 1989; Deci and Ryan 1987).

Interest in organizational applications of LAS first emerged in field studies in corporate settings. For instance, Deci et al. (1989) observed that when employees in a Fortune 500 firm perceived their direct manager as supporting their autonomy, their satisfaction with their supervisors, job satisfaction, and trust in the senior organizational leaders tended to be elevated. Since then, research on autonomy support in leadership has grown substantially, with autonomy supportive practices shown to predict positive work behavior (e.g., proactive and prosocial work behavior; Gagné 2003; Slemp 2017; Slemp et al. 2015), as well as employee well-being and work engagement (e.g., Deci et al. 2001; Moreau and Mageau 2012; Schultz et al. 2015; Williams et al. 2014). Other studies have explored work-based correlates of LAS, including job attitudes (e.g., Collie et al. 2016), performance (e.g., Baard et al. 2004; Braun et al. 2012), and turnover intentions (e.g., Gillet et al. 2013; Liu et al. 2011). But perhaps the most prevalent area of inquiry has explored how LAS relates to motivational processes, including basic psychological needs at work (e.g., Van den Broeck et al. 2016), and various forms of work motivation, to which we turn next.

Central to SDT is the distinction between forms of autonomous motivation, controlled motivation, and amotivation (see Fig. 1; Gagné et al. 2018; Howard et al. 2016; Ryan and Deci 2000). Autonomous motivation encompasses behaviors emanating from within the self, involving a sense of volition and choice. Intrinsic motivation is a prototype of autonomous motivation, and involves engaging in an action because it is interesting or enjoyable. Other autonomous forms of motivation are identified and integrated regulation. When acting through identified regulation, the individual recognizes the value of a behavior and is volitionally motivated to enact it. Integrated regulation occurs when identified values fit together and are congruent, such that the person can be wholeheartedly engaged. In contrast, controlled motivation involves behaviors that are performed due to causes perceived to be external to the self, and thus the volitional component of behavior is either partially or completely absent. A highly controlled form of behavior is represented by external regulation, in which one is motivated by externally administered contingencies such as tangible rewards or avoiding punishment. Slightly less controlled is introjected regulation, where an individual begins to internalize and value the external contingencies sought in external regulation. When introjected, the pressures on the individual are internal, and self-esteem is contingent on one’s behaviors. Beyond these categories of autonomous and controlled motivation is amotivation, which involves no desire to enact behavior. Amotivation has two broad sources or subtypes—amotivation due to lack of perceived control or efficacy; and amotivation due to lack of value or concern, both of which can be problematic in the workplace when the behavior needs to be enacted.

Open image in new window

An important part of this SDT framework is the process of internalization, which refers to a natural tendency for people to transform controlled motivation into more autonomous forms of motivation (Ryan and Deci 2000). Support for basic psychological needs for autonomy, competence and relatedness is seen as essential to fuller internalization, and among these supports, autonomy support is particularly central (Ryan and Deci 2017). Autonomy support includes such elements as taking the actor’s internal frame of reference (empathy), provisions of rationale and choice, and minimal use of controls to motivate. Within an autonomy supportive context, individuals are encouraged to value and assume responsibility for behaviors or goals (Deci et al. 1994). Thus, LAS should theoretically be more positively and strongly correlated with the more internalized motivational processes described by SDT (Gagné et al. 2015; Ryan and Deci 2017), as illustrated in Fig. 1.

Whereas LAS is likely to show positive and progressively stronger correlations with the more internalized motivational processes towards the right of Fig. 1, we expect that for externally regulated behavior, given its lack of internalization, correlations will generally be unrelated to LAS (Gagné et al. 2015; Ryan and Deci 2017). This is because LAS is more likely to exert its positive effects on individuals’ more autonomous and intrinsic forms of motivation, which have typically been uncorrelated with external regulation (Ryan and Deci 2000, 2017). In external regulation, one’s motivation is focused on factors external to the self, such as pay or incentives, whereas an autonomy supportive manager would be supporting a sense of ownership, interest and value, reflected in identified and intrinsic motives. Depicted as well in Fig. 1 is our expectation that LAS will, in contrast, show a significant negative correlation with amotivation. Since autonomy support fosters an environment where individuals experience willingness and choice, people should be more motivated to engage in their work, and amotivation is more likely to be absent. Thus, unlike external regulation which has typically been uncorrelated with highly autonomous motives, amotivation has typically been negatively associated with them, and we expect a similar a pattern in how LAS relates to these motivational types.

Notably, although many studies support these general patterns of relations (e.g., Collie et al. 2016; Gagné et al. 2015), others do not. For instance, correlations between LAS and introjected regulation range from strongly positive (r = .42; Nie et al. 2015) to moderately negative (r = − .18; Krieger and Sheldon 2015). The strong correlation of LAS and introjection in the Nie et al. study was accompanied by even stronger positive relations of LAS to intrinsic and identified motivations, as well as a strong negative relation with external regulation (r = − .38). It is possible therefore that the sample of Chinese teachers in this study were not only more autonomously motivated by higher autonomy support, but also felt greater internal pressure “not to disappoint”. Mixed findings can also be observed in other studies (e.g., Allen and Bartle 2014). Correlations between LAS and intrinsic motivation range from relatively small (r = .15; Roche and Haar 2013) to large (r = .56, Olafsen et al. 2015), with similar ranges for identified regulation. Correlations with external regulation are similarly varied, with some studies showing positive relations (e.g., Jones 2002; Oostlander et al. 2014) and others showing negative relations (e.g., Nie et al. 2015). By statistically aggregating previous findings through meta-analysis, we hope to provide clarification about the strength of the associations between LAS and the full range of motivational processes discussed in SDT.

The heterogeneity observed in prior studies between LAS and motivational processes, including internalization of work motivation or basic need satisfaction, also point to a need to understand study and participant characteristics that impact these associations. Meta-analysis is particularly well-suited to testing moderating factors (Rosenthal and DiMatteo 2001). While any number of moderators could be explored, depending on appropriate data being available, we make two theoretically derived predictions concerning possible factors that moderate correlations between LAS and its motivation-based correlates.

First, there is evidence to suggest that the effects of leader behaviors on employees may be affected by the physical and/or psychological distance between the leader and employee (Antonakis and Atwater 2002). The associations between transformational leadership and follower behavior, for example, has been found to be attenuated by increased social (e.g., Cole et al. 2009) or physical (e.g., Howell and Hall-Merenda 1999; Humphreys 2002) distance. Based on the available studies, we broadly define distance in terms of the supervisor and employee hierarchy within the organization, ranging from proximal (i.e., immediate supervisor) to more distal sources (e.g., senior organizational leaders). It is likely that the salience of leader behavior is heightened within proximal relationships because it is more commonly observed, and thus observed correlations may be stronger in these situations.

Second, prior research has suggested that perceived autonomy predicts well-being across cultures (e.g., Chirkov et al. 2003; Ryan 1995), suggesting that the need for LAS may be culturally universal. But studies have not directly compared the impact of LAS on psychological needs and motivation in employees across cultures. While culture can be classified in a variety of ways, a basic classification that has been used in the literature—and which we use here—is between individualist and collectivist cultures (Oyserman 2017).

While the moderators previously reviewed are theoretically driven, there are a variety of study-related factors that can also yield differences in the observed effects. Two factors that can bias results of meta-analyses are publication bias and imperfect construct validity in measurement instruments (Schmidt and Hunter 2015). If publication bias is present in a literature, the likely upshot is an upward bias in mean effect sizes and a downward bias in the variability across effect sizes due to “missing” small effect-size studies (Schmidt and Hunter 2015; Schmidt and Oh 2016). Hence, publication bias is known to falsely inflate the results of research literatures. To explore this possibility in the present meta-analysis, we test publication status as a possible moderator of observed effects, and we also explore evidence of small study bias (Borenstein et al. 2009).

In addition, several instruments have been used to measure perceived leader autonomy support in the literature, including the short- and long-form of the Work Climate Questionnaire (Baard et al. 2004), the Perceived Autonomy Support Scale for Employees (Moreau and Mageau 2012), and the Work Climate Scale (Deci et al. 1989), among other idiosyncratic measures (e.g., Lynch et al. 2005). Because correlations are attenuated by imperfect construct validity in measurement instruments (Schmidt and Hunter 2015), to the extent that any of these measures are less construct-valid than others, their yielded effects will likely be systematically biased downwards. Thus, the operationalization of autonomy support is important to consider as a possible moderator of observed effects and we explore this possibility in the present meta-analysis.

We had no a priori hypotheses for either publication bias or the operationalization of autonomy support, and thus present these moderators as exploratory results.

SDT posits that individuals experience optimal psychological functioning to the extent that three basic psychological needs for autonomy, competence, and relatedness are satisfied (Deci and Ryan 2000). Autonomy involves a sense of choice and freedom in one’s behavior, and entails the perception that one’s behavior is a function of one’s own interests and values rather than being controlled by forces or pressures external to the self. Competence involves feelings of mastery, attaining desired outcomes, and succeeding at challenging tasks. Relatedness involves an ability to develop meaningful relationships and connection with others.

Leaders play an important role in establishing and maintaining a social context that allows employees to feel free to pursue experiences that satisfy these three needs. SDT argues that, because autonomy supportive leaders attempt to understand the perspective of their employees, they are more likely to facilitate employee experiences of autonomy, competence, and relatedness (Ryan and Deci 2017). When operating within an autonomy supportive workplace, employees can engage in more self-directed behaviors, more freely address obstacles and challenges, and feel more support and connection (Deci et al. 2001). LAS may therefore be a potent social-contextual motivational precursor to fulfilling these three psychological needs (Baard et al. 2004; Gagné 2003). This is consistent with recent meta-analytic findings (Van den Broeck et al. 2016) that found positive associations with autonomy (ρ = .65; k = 13), competence (ρ = .38; k = 13), and relatedness (ρ = .39; k = 14).

Because the satisfaction of autonomy, competence, and relatedness tends to catalyze intrinsic enjoyment and/or value and satisfaction in activities themselves, need satisfaction is understood to be essential for the internalization of motivation. As noted by Gagné (2003), “People are more likely to be intrinsically motivated … when they can freely choose to pursue the activity (autonomy), when they master the activity (competence), and when they feel connected and supported by important people, such as a manager, a parent, a teacher, or team-mates (relatedness)” (p. 202). On this basis, prior work tends to position basic need satisfaction as a conceptual and empirical antecedent to autonomous work motivation (e.g., Collie et al. 2016; De Cooman et al. 2013).

SDT also predicts that need satisfaction results in individual well-being (Olafsen 2017). Numerous theories and definitions of psychological well-being exist, including ideas stemming from both hedonic (e.g., subjective well-being) and eudaimonic (e.g., meaning/purpose) traditions (Deci and Ryan 2008; Ryan and Deci 2001). While there is some evidence that LAS predicts both forms of well-being, it has not yet been established whether it relates more strongly to hedonic indicators, or feelings and experiences associated with eudaimonia, as few studies have examined both simultaneously. This is important because an understanding of the environmental conditions that support different types of well-being will allow for more targeted interventions (Kern et al. 2015).

Beyond well-being, the need satisfying experiences and internalization of motivation produced by LAS may also impact a variety of workplace outcome variables, including work performance, job satisfaction, work engagement, and various work-related behaviors (e.g., proactive or prosocial behavior) (Deci et al. 2017). When employees perceive that they are free to perform their work in their own way within an autonomy supportive context, they may be more likely to find that work engaging, possess more favorable evaluations of the job (job satisfaction), and proactively engage with their environment and others with whom they work (proactive and prosocial behavior, and work performance). Overall, the empirical data is largely consistent with these premises, showing positive relations between LAS and work engagement (e.g., Deci et al. 2001; Van Schie et al. 2015), positive job attitudes (e.g., organizational commitment; Chang et al. 2015; Collie et al. 2016) prosocial and proactive behaviors (e.g., Gagné 2003; Güntert 2015; Slemp 2017; Slemp et al. 2015), and performance (e.g., Baard et al. 2004; Braun et al. 2012).

To summarize, Deci et al. (2017) suggested a model of work motivation that posits how the SDT constructs relate with each other within the organizational literature. Figure 2 adapts their model specifically to focus on LAS. From this perspective, autonomy support is a social-contextual antecedent that fosters basic need satisfaction, which in turn leads to the internalization of work motivation, which leads to enhanced employee health (e.g., employee well-being and lower distress) and work outcomes (e.g., performance, job satisfaction, work engagement). While some studies have tested various parts of Fig. 2, no study to our knowledge has simultaneously tested the full model. Thus, our specific predictions regarding this proposed model are:

Open image in new window

Although several studies have examined correlates of LAS in workplaces, no attempt has yet been made to statistically aggregate the findings in this literature. Accordingly, the exact magnitude of its relationship with important variables, including the internalized work motivation processes described by SDT, as well as well-being, behavior, and other work-based outcomes have not yet been established. Meta-analysis can help address this need because it makes possible the systematic combination and statistical aggregation of single studies to obtain more stable estimates of overall associations, as well as moderators of those associations (Schmidt and Hunter 2015).

In conducting our meta-analysis, we had three objectives. First, we aimed to identify and quantify the relations between perceived LAS and a range of variables that have been tested in the organizational literature, presenting broad descriptive evidence of its most common correlate variables with estimates of the size of the corrected correlations. As illustrated in Fig. 2, studies have primarily treated autonomy support as a predictor. Because few studies have explored potential antecedents of LAS, particularly from the leaders’ perspective, including age, gender, tenure, organizational seniority, or relevant personality characteristics (e.g., agreeableness), we primarily focus on the theoretical outcomes associated with LAS.

Second, we aimed to test potential moderators of the observed correlations. We consider the distance between the employee and supervisor and compare individualist and collectivist cultures as potential influences. We also explore publication bias and how autonomy support was operationalized as potential study-level factors. Identification of moderators can help explain prior inconsistent findings in the literature, and may also help guide future research efforts.

Third, we aimed to test the hypothesized pathway model summarized in Fig. 2. Integrative reviews make it possible to bridge across samples, even when no one primary study has investigated all of the variables of interest (Becker 2009; Schmidt 2010; Viswesvaran and Ones 1995). Because the overall correlations connect the studies together, pathways can be tested, and theoretical predictions can be explored. This technique is important because it allows us to go beyond the individual meta-analysis results and assess the relative impact of the key theoretical mechanisms intended to explain the autonomy support to work motivation and work outcome associations proposed by Deci et al. (2017).