Job Autonomy,Role Ambiguity,and Procedural Justice: A Multi-Conditional Process Model of Change-Oriented Organizational Citizenship Behavior in Public Organizations

Job Autonomy, Job Performance, and Change-Oriented OCB

Although it is a condition of organizational performance that employees reliably execute the tasks they have been assigned, organizations are complex open systems that additionally require supportive employee behaviors seldom specified in formal job descriptions (Katz, 1964). OCB denotes discretionary employee behaviors that support the organization’s established norms and processes (Organ, 1988). Change-oriented OCB is similar to the classic OCB types introduced by Denis Organ and colleagues in that it is discretionary and benefits the organization. However, whereas traditional OCB denotes behaviors that support current operational methods, change-oriented OCB challenges these and attempts to change and improve them (Li & Xie, 2022). Because of this challenging character, the conditions that drive change-oriented OCB differ from those associated with traditional OCB and include a baseline level of employee job autonomy (Campbell, 2022).

Job autonomy captures the extent to which an employee enjoys control over resources, time, work methods, and job expectations (Dysvik & Kuvaas, 2011; H. Kim & Stoner, 2008; Muecke & Iseke, 2019; Sorlie et al., 2022). Management scholars have long suggested that increasing employee autonomy allows employees to act upon their localized knowledge to improve work processes and performance. In centralized organizations, where employees have little scope to determine how their work efforts are expended, a necessary condition for change-oriented OCB is lacking (Cerit, 2017; Lee, 2019; Raub, 2008), and, in the public sector, when employees have more scope to craft their jobs, they will engage in change-oriented OCB (Chen et al., 2023). Fernandez and Moldogaziev (2012, p. 161) argue that looser controls allow employees to “tinker with existing elements and practices and reconfigure them in new ways,” which links job autonomy directly with change-oriented OCB (see also Tran et al., 2021). Job autonomy is also linked to concepts that bear a family resemblance to change-oriented OCB such as creativity, service innovation, and innovative behavior (Cho & Song, 2021; Dhar, 2016; Lynggaard et al., 2018; Wang & Cheng, 2009). From an organizational psychology perspective, job autonomy can enhance motivational factors including work engagement, job meaningfulness, job ownership, organizational identification, and organizational commitment (Garg & Dhar, 2017; Hackman & Oldham, 1976; Hameed et al., 2022; H. Kim & Stoner, 2008; Kuvaas et al., 2016; Sorlie et al., 2022). Self-determination theory (SDT) (Deci & Ryan, 2008; Muecke & Iseke, 2019) holds that individuals have an inherent need for autonomy, and thus the provision of job autonomy can strengthen intrinsic motivation (Dysvik & Kuvaas, 2011; Peng et al., 2015). A key job resource (Bakker & Demerouti, 2007), job autonomy signals to employees that the organization condones the adoption of broader job roles (Grant & Parker, 2009), and, in a public sector context, autonomy can further allow employees to act on their public service motives (Lynggaard et al., 2018), which Campbell and Im (2016) link to change-oriented OCB. In short, autonomy functions as both a practical and psychological condition of change-oriented OCB, allowing employees to “break out of a routine and to find the best solution in the process of trying out new ideas” (Lee, 2019, p. 33).

Job autonomy affords the opportunity to pursue change. However, even within the scope of existing operations, there are opportunities for employees to make small decisions about how to best allocate their time and effort. Consequently, the link between autonomy and formal job performance has been extensively documented, with meta-analyses reliably linking the two factors across different work contexts (Christian et al., 2011; Muecke & Iseke, 2019; Spector, 1986). Work consists of making many small decisions that draw on the knowledge and experience of the employee, the quality and accuracy of which are a determinate factor in producing high-performance. Even in machine bureaucracies, where processes are regimented, mechanized, and based on a centrally-produced blueprint, employees retain discretion about how their effort is expended, and discretion will be an even greater contributor to performance when operations call for judgment. All work processes, standardized or creative, require flexibility (Kuvaas et al., 2016, p. 4), and autonomy allows employees to make connections between different behaviors and job results (Dedahanov et al., 2019), leading to more consistent job goal attainment (Peng et al., 2015). In an influential review, Grant & Parker (2009, p. 343) write that autonomy “facilitates enactive mastery experiences,” not only helping to develop task controllability but also allowing the employee to acquire new job-related skills. Again, autonomy is linked to job performance via enhanced job motivation (Joo et al., 2010), self-efficacy (Grant & Parker, 2009; Saragih, 2011), and, among public employees, job engagement (Hassan, 2014). In contrast, when organizational control is excessive, the basic psychological need for autonomy is lacking, and the sustained expenditure of effort linked to high task performance will not materialize (Deci & Ryan, 2008). Autonomy thus facilitates the production of psychological factors that predict job performance (Donkor, 2021; Rich et al., 2010).

Because employee job performance is a critical factor for organizational performance, scholars often focus on its predictors. However, an employee’s job performance has consequences for their overall experience of and orientation toward the organization. Both the theory of planned behavior (Ajzen, 1991) and affective events theory (Weiss & Cropanzano, 1996) suggest that specific experiences at work shape employee work attitudes and behaviors along with the more stable structures of the organizational work context. For instance, performance ratings can affect job satisfaction (Ghasemy et al., 2022; Varma & Malhotra, 2022) and turnover intentions (Zimmerman & Darnold, 2009). Although job performance and affiliative OCB have been studied as independent job behaviors (for instance, in Chiaburu et al., 2017), a critical difference between change-oriented OCB and its affiliative cousin is that it is difficult. I argue that the opportunities to change organizational processes afforded by job autonomy will not lead to change-oriented OCB if a baseline of job mastery and formal performance has not been obtained. In other words, formal job performance is likely to predict change-oriented OCB.

High performers possess the necessary insight into the work context to identify concept, tool, or process-based performance deficiencies. Such opportunities for change are unlikely to be detected by those who struggle to meet baseline performance standards. Higher performers are also more likely to experience such work context deficiencies given that they are using the tools available to them at a level closer to their performance potential. Although change-oriented OCB can be a source of performance improvement for the employee that implements it (Campbell & Im, 2016), and therefore be an antecedent of formal performance, it is higher performing employees who possess the requisite skill, knowledge, experience, and record of success necessary to challenge existing job practices and processes. In general, employees who meet formal performance requirements are more likely to engage in additional discretionary behavior to benefit the organization (MacKenzie et al., 1998). However, this relationship is likely to be more prominent in the case of change-oriented OCB given that successfully pursuing it requires a level of skill, knowledge, and control that traditional OCB does not. In contrast, although negative performance feedback may encourage some employees to search for methods to improve, the empirical literature suggests that low-performance often leads to withdraw, psychologically and physically, from the workplace (Zimmerman & Darnold, 2009). In short, relative to high performers, weak performers lack the skill, insight, and confidence necessary to introduce performance-enhancing change. Based on this reasoning, I argue that higher performance is likely to be associated with higher levels of change-oriented OCB. ¹

Role Ambiguity, Job Autonomy, and Job Performance

Job autonomy is a job resource that impacts employee well-being and performance. A given work context, however, consists of both resources and demands, and these contextual factors are known to interact (or buffer one another) in the production of a range of relevant attitudes and behaviors (Bakker & Demerouti, 2007). Role ambiguity, which denotes an employee-level informational deficiency of critical job-relevant knowledge (House & Rizzo, 1970), is a contextual job demand (Jung, 2014; Zhang et al., 2022) that can shape the impact of job autonomy in public sector organizations. Role ambiguity can result from incompatible expectations or conflicting demands, which are prevalent in the public sector (S. Kim & Wright, 2007; Stazyk et al., 2011), and Jong (2016) suggests that the importance of quality performance information increases with job autonomy. When job autonomy is low, higher levels of role ambiguity lead to stress, burnout, and diminished job satisfaction (Jong, 2016; H. Kim & Stoner, 2008), conditions that are not conducive to performance. Given that role ambiguity in public organizations is not only a direct antecedent of performance (as well as a range of other performance antecedents, such as work motivation, Campbell, 2016, or commitment, Stazyk et al., 2011), but is known for its determinantal effect on other performance antecedents, I argue that role ambiguity will attenuate the link between job autonomy and job performance.

Subordinates possess knowledge of job conditions that their superiors do not, and job autonomy allows them to expend their time and effort in ways that leverage this knowledge and thereby improve performance. However, even if an employee enjoys job autonomy, if they lack sufficient knowledge about goals, job standards, or the scope of their responsibilities and priorities, autonomy will not translate into performance and may even undermine it. Sufficient information is necessary to transform autonomy into performance, and employees enjoying job autonomy but afflicted by role ambiguity lack the informational basis upon which to make sound decisions about the allocation of time and effort. Consequently, to the degree that role ambiguity prevails, the effect of job autonomy on performance will weaken. Although a well-designed bureaucratic structure can communicate the organization’s goals, priorities, and standards to employees (Olsen, 2006), these factors can conflict (Pandey & Wright, 2006). Job autonomy allows employees more scope to improve their work habits (Gao & Jiang, 2019), however, if they simultaneously experience low levels of certainty about the goals and standards relative to their work processes, it becomes less likely that such autonomy will translation into higher levels of job performance.

Procedural Justice, Job Performance, and Change-Oriented OCB

Due to its inherent difficulties, formal job performance should act as a preparatory phase in the production of change-oriented behavior. However, change-oriented OCB is not only difficult; because change-oriented OCB involves challenging existing processes and norms, those who pursue change-oriented initiatives may be open to criticism or sanction if their efforts fail (Campbell, 2022; Dedahanov et al., 2019). As a disruptive force, change-oriented OCB can lead to interpersonal conflict (Campbell, 2020), which is in contrast to affiliative types of OCB that facilitate group cohesion. Moreover, any given change initiative may fail, consuming resources but undermining performance. Employees will thus weigh the risks, costs, and benefits of making suggestions (Liang et al., 2012), and the decision to engage in change-oriented OCB will be based not only on opportunity and competence but also an evaluation of the probability of sanction or ostracism should an attempt at change fail. Indeed, the desire to avoid blame for failure undermines the pursuit of change (Timeus & Breaugh, 2020).

Procedural justice captures a perception that fairness is an objective property of organizational decision-making processes (De Clercq et al., 2021; Konovsky, 2000) and those affected by decisions have the means to influence them (Thibaut & Walker, 1975; Tyler, 1989). Whereas behaviors linked to formal performance pose little risk to employees, procedural justice is relevant to the link between job performance and change-oriented OCB due to the latter’s inherent risk. If public organizations lack policies that provide a feeling of safety and security, employees are less likely to offer new ideas to improve performance or fix problems (Brunetto et al., 2022; Iqbal et al., 2022) and, without fairness, employees will be reluctant to improvise (De Clercq et al., 2021). Change-oriented OCB is facilitated by psychological safety (Iqbal et al., 2020), and procedural justice can provide high-performance employees with the confidence that they will not be unduly prosecuted for pursuing (failed) change (Fernandez & Moldogaziev, 2012). Because high-performance employees are more likely to consider engaging in change-oriented OCB, I predict that perceptions of organizational justice will be more relevant in their case. Despite possessing the competence and knowledge necessary to pursue change, performant employees will be less likely to breach the boundaries defined by their formal performance arrangement and duties in contexts that are perceived to lack procedural justice. In such contexts, employees will simply maintain the status quo, given that there will be little opportunity to defend their behavior and prevent unjust punishment in the case that their change-oriented OCB initiative fails. ²

The logic of the inter-construct relationships suggests several additional hypotheses. Subjective performance occupies an intervening position between job autonomy and change-oriented OCB, suggesting that it plays a mediating role. Relatedly, the moderating roles of both role ambiguity and procedural justice on the mediation paths suggest that these factors will independently condition the indirect effect. This system of hypotheses represents a multi-conditional process model of change-oriented OCB. The auxiliary hypotheses are stated as follows.

2020 Korean Public Employee Perception Survey

The Korean Public Employee Perception Survey (KPEPS), collected by the Korean Institute of Public Administration (KIPA), is a large survey of central and local government employees conducted annually in South Korea. The 2020 version used here was collected from August 12 to September 30, 2020. The target population for the survey consisted of 23,703 public officials working in general positions in the headquarters of 46 central government ministries and agencies and 51,136 officials working across 17 metropolitan cities and provinces as of December 31, 2019. The target sample size was set to 4,000 responses (2,000 from central government and 2,000 from local government organizations), with a target sampling error rate of ±2%. A stratified sampling approach based first on organization type and second on employee number was used. Within the sample clusters, participants were randomly selected and sent an email with a link to the survey, or provided with a QR code-based link if email was unavailable due to security or other reasons. A total of 4,339 survey responses were collected. About 40% of the sample is female, and the average employee is in their 40s and with a four-year university degree. About 63% of the sample consists of employees with a job grade of 9 to 6, which denotes lower-level administrative workers. The remaining responses are drawn from grades 5 through 2, which consist of higher-ranking civil servants who have more managerial and policy-making responsibilities. More detailed information about the survey methodology is available (in Korean) from KIPA at https://www.kipa.re.kr/site/kipa/main.do.

Measures

Change-oriented OCB denotes “voluntary and constructive efforts, by individual employees, to effect organizationally functional change with respect to bow work is executed within the contexts of their jobs, work units, or organizations” (Morrison & Phelps, 1999, p. 403). Morrison and Phelps’s (1999) original scale used 10 items to measure the construct, 9 of which were modified slightly by Vigoda-Gadot and Beeri (2012) for use in the public sector. The content of the questions refers to performance-oriented work process modification behaviors (for instance, “This person often tries to change how his or her job is executed in order to be more effective”) or addressing work process deficiencies through change (for example, “This person often tries to correct a faulty procedure or practice”). Using a Korean sample, Campbell (2015) uses a modified 4-item subset of the original scale, which also emphasizes performance-oriented change and error correction. I use 4 items from the KPEPS that capture these two key change behaviors, namely, developing new work methods to improve effectiveness and efficiency and correcting faulty work processes. The items, like those in the other scales, are listed in Table 1. Although the 4 items are not word-for-word translations of the original scale, they have reasonable face and content validity as a measure for change-oriented OCB. However, here I note that ideally, the pre-validated version of the scale should be used, which is a limitation of this study. Further remarks below are provided about the validity and psychometric properties of the scale. Cronbach’s alpha for the 4-item scale is .82.

OPEN IN VIEWER

Table 1.

Survey Items.

Change-oriented organizational citizenship behavior (α = .82)

I try to create and apply a new and original way of doing things.

I develop new ideas to solve problems that arise while performing my work.

In pressing cases, I think problem solving takes precedence over compliance.

I actively strive to improve unreasonable factors in my work.

Job autonomy (α = .82)

I can choose the methods and procedures relevant to my work.

I can adjust the speed and deadlines of my work.

I can participate in the establishment of work performance evaluation indicators and standards.

Job performance (α = .89)

I acheive the performance required by my job.

I acheive the job performance related my and other organizations, and from stakeholders.

Role ambiguity (α = .90)

The scope of responsibility for my duties is unclear.

The priorities for my duties are unclear.

It is unclear how to perform my duties.

Procedural justice (α = .86)

In our organization, decisions are made in a fair way.

Employees may request explanations or additional information about decisions made in our organization.

Employees can challenge to decisions made in our organization.

Job autonomy captures “the extent to which a job allows the freedom, independence, and discretion to schedule work, make decisions, and select the methods used to perform tasks” (Hackman & Oldham, 1974; Morgeson et al., 2005, p. 399). Scholars in public administration have measured job autonomy in various ways (Jong, 2016; H. Kim & Stoner, 2008; Tummers et al., 2018), with items closely mirroring these conceptual dimensions. The 3 items measuring job autonomy here likewise reflect these three dimensions of the construct, capturing the perceived freedom of the employee to adjust both the scheduling as well as the method of completing their work. Moreover, the third item in the measure has particular relevance to performance management-based job design, specifically referencing the freedom of the employee to participate in the formulation of performance standards. The 3 items have an alpha of .82.

Motowidlo (2003) defines job performance as “the total expected value to the organization of the discrete behavioral episodes that an individual carries out over a standard period of time” (p. 39). When job tasks are simple, employee performance can be measured objectively (Bellé, 2013), via a subjective assessment by superiors or a third party, such as clients (Schwarz et al., 2020), or by the focal employee (Leisink & Steijn, 2009). Although piece rate-based estimates are accurate and objective, performance in the public sector is complex, and seldom are such measures available. In contrast, supervisor ratings are presumably less tainted by bias resulting from a desire to maintain a positive self-image, however, they are also subject to halo bias (Bellé et al., 2017). A recent study suggests that employee self-reports of performance are more biased when their target is broad (for instance, creative performance) than when the evaluative criteria focus on formal job requirements (Janssen & van der Vegt, 2011). In this study, I use a 3-item self-reported measure of job performance (α = .89), which captures the perceived achievement of job-mandated objectives. This self-reported measure of performance is a limitation of the study, which I discuss further after a presentation of the results.

Role ambiguity is a foundational construct in organizational behavior studies. The concept captures a “lack of clarity and predictability of the outcomes of one’s behavior” (House & Rizzo, 1972, p. 471), and is generally viewed as a job-relevant informational deficit. The 3 items I use to measure role ambiguity reflect this deficit in terms of basic job knowledge, job responsibility scope, and job priorities, and are like items developed by House and Rizzo (1972), as well as those used by Campbell (2016) and Pandey and Wright (2006). The alpha for the scale is .90.

Organizational procedural justice research originally grew from legal psychology and concerns perceptions of fairness in the exercise of authority (Blader & Tyler, 2003; Tyler & Lind, 2002). Measures of procedural justice emphasize fairness and transparency in terms of decision-making (J. Kim & Siddiki, 2018). I use 3 items that capture the fairness and openness of decision-making, as well as the perceived ability of the focal employee to challenge decisions. The alpha for the scale is .86.

Factor Analysis and Variance Diagnostics

A principal components analysis (PCA) with all items reports five distinct factors with eigenvalues above 1. Each item in the set loads onto its predicted factor strongly (lowest loading equal to .67 and the average loading of .83). The proportion of variance explained by the first factor is about, 34%. I also performed a second, more conservative PCA including only the items measuring autonomy, performance, and change-oriented OCB. This test suggests that a common factor explains only about 42% of total item variance. This figure is below the 50% threshold commonly used to diagnose common method bias.

Descriptive Statistics and Correlations

Table 2 shows descriptive statistics and correlations for the variables in the study. Composite variables in the table were created from the unweighted mean of all scale items. All variables are measured on a 5-point scale and means range from 3.12 in the case of job autonomy to 3.75 for job performance, with the mean of role ambiguity a lower 2.67. All variables are correlated in the expected direction at p < .001, with the highest correlation of .47 between job performance and change-oriented OCB.

OPEN IN VIEWER

Table 2.

Descriptive Statistics and Correlations.

#	Variable	Mean	SD	1	2	3	4
1	Change-oriented OCB	3.36	0.63
2	Job autonomy	3.12	0.77	.27
3	Job performance	3.75	0.63	.47	.25
4	Role ambiguity	2.67	0.79	−.15	−.32	−.27
5	Procedural justice	3.07	0.77	.29	.43	.27	−.34

Note. Scales have a minimum of 1 and maximum of 5. All correlations significant at p < .001.

Confirmatory Factor Analysis

A confirmatory factor analysis was performed using the lavaan (version 0.6-9) package in R (Rosseel, 2012). In the model, each item was specified to load onto the relevant construct, and there were thus five latent variables in the model. A variety of fit indices can be used to assess the fit of latent variable models, and I selected four that are common in the empirical public administration literature (for instance, Borst et al., 2019, or van Loon, 2017). The comparative fit index (CFI) and Tucker–Lewis index (TLI) indicate good fit with values above .90 and excellent fit at .95 or higher. The root mean square of approximation (RMSEA) indicates good fit below .08 and excellent fit below .05 (Cangur & Ercan, 2015; Hu & Bentler, 1999). The standardized root mean square residual (SRMR) suggests good fit at 0.10 and excellent fit at 0.05 or lower. Goodness of fit indices indicate excellent fit for the 5-factor model (CFI = 0.984, TLI = 0.980, RMSEA = 0.039, SRMR = 0.029). (I note that the chi-square test is statistically significant; however, this test is sensitive to sample size and is thus rarely used in the quantitative public administration literature (Borst et al., 2019).)

Standardized factor loadings for all items in the job performance, job autonomy, role ambiguity, and procedural justice scales were above .750. Three of the items in the change-oriented OCB scale were also above .750, however, 1 item had a lower loading of .444. Removing this item marginally improved the fit of the whole model, however, I decided to retain this item to increase the underlying variability of the scale. However, I also implemented the full structural model with latent interaction terms again with this item removed and found no substantive differences in the direction or statistical significance of the coefficients. The average variance extracted for the constructs ranges from .557 in the case of change-oriented OCB to .747 in the case of role ambiguity, all of which are above the .500 cutoff suggested by Hair et al. (2014). Composite reliability was also above .800 for each construct, which is comfortably above the conventional .600 threshold.

I compared the five-factor model with three alternative four-factor models which combined each of the three strongest correlated variables into single factors (change-oriented OCB and job performance, job autonomy and procedural justice, and role ambiguity and procedural justice, respectively). Each of these alternatives had worse fit statistics than the five-factor model and all had fit indices that were below the conventional cutoff points for assessing model fit. Together, these tests suggest that the five-factor model best captures the underlying structure of the data and that it is suitable for SEM.

Structural Equation Modeling With Latent Interaction Terms

The model proposed in this study depicts a sophisticated system of both mediated and moderated effects. For estimating indirect effects, SEM is favored relative to the traditional independent regressions approach popularized by Baron and Kenny (1986) as it allows latent variables to be specified as endogenous and estimates all paths simultaneously (Gunzler et al., 2013; Iacobucci et al., 2007). Also, unlike traditional regression, SEM accounts for measurement error, which is important when testing for interaction effects (Schoemann & Jorgensen, 2021). Testing interaction effects in latent variable models is more complex than in standard regressions, however, and organization scholars often forgo latent interaction effects in favor of simpler (through more biased) regression-based estimates (Cortina et al., 2021). However, there are a variety of strategies available to test interactions in a structural model (Sardeshmukh & Vandenberg, 2017). To create the interaction terms, I use the matched-pair product terms approach developed by Marsh et al. (2004), which provides a good balance between robustness, information utilization, and implementation practicality. Items were orthogonalized using residual centering before creating the observed product terms (Little et al., 2006). In this way, I can estimate the measurement and structural model simultaneously and thereby account for measurement error at the construct level, while also testing the interaction effects.

As the inclusion of the interaction terms artificially increases degrees of freedom (and thus inflates model fit), I estimate fit for the base measurement and structural model based on a model implementation without them but with all direct paths specified (i.e., paths A, B, and C in Figure, 1, as well as links between role ambiguity and job performance and between procedural justice and change-oriented OCB). Fit statistics suggest that the model fits the data well (CFI = 0.982, TLI = 0.978, RMSEA = 0.040, SRMR = 0.039). Next, I generated a set of residual-centered matched-pair product terms to create latent interactions for job autonomy and role ambiguity and second for job performance and procedural justice. To automate this process, I used the semTools R package (Schoemann & Jorgensen, 2021).

Figure, 2 shows the standardized path coefficients, related p values, and r-squared values for the two endogenous factors produced by the SEM analysis. ³ Parameters are consistent with hypotheses 1 through 5: both job autonomy and job performance are positively related to change-oriented OCB (with standardized coefficients (β) equal to .121 and .478, respectively, p < .001), and autonomy is likewise related to job performance (β = .232). Second, the path coefficient is negative for the interaction between role ambiguity and job autonomy (β = −0.152, p < .001) and positive for the interaction between procedural justice and job performance (β = .059, p < .05).

OPEN IN VIEWER

Figure 2.

Results of structural equation modeling with latent interaction effects.

The logic of the results suggests that job performance mediates the relationship between autonomy and change-oriented OCB (hypothesis 6). To evaluate this empirically, bias-corrected confidence intervals based on 2,000 bootstrap draws were estimated for the indirect effect (with each of the moderators held at their mean). These did not contain zero. Given that job autonomy has a direct and statistically significant relationship with change-oriented OCB, this statistic suggests partial mediation.

Bootstrapped bias-corrected confidence intervals were next estimated for the indirect effect at both low and high levels of both role ambiguity and job autonomy. Surveying the coefficients and confidence intervals of these estimates suggests that the effect of job autonomy on change-oriented OCB strengthens are role ambiguity decreases and procedural justice rises. To aid and simplify interpretation, I graphed the raw estimates and confidence intervals for the nine different cases. ⁴ As Figure 3 shows and consistent with hypotheses 7 and 8, the indirect effect is weakest when role ambiguity is high and procedural justice is low (position 1 on the x axis) and highest when role ambiguity is low and procedural justice is high (position 9). Interpreting the rate of change between different states suggests that role ambiguity plays a locomotive role relative to procedural justice. In any case, when role ambiguity is low or at its mean, the indirect effect is significant (the confidence intervals do not contain zero). If role ambiguity is high (positions 1 through 3 on the x axis), the lower confidence intervals fall below or at zero, suggesting that autonomy will have little indirect impact on change-oriented OCB in these cases.

OPEN IN VIEWER

Figure 3.

Indirect effect based on the joint status of moderators.