Promoting Autonomy to Reduce Employee Deviance: The Mediating Role of Identified Motivation

The organizational environment is purported to have a profound impact on how employees behave at work. In particular, the extent to which the work environment can foster autonomy in employees has been shown to predict several positive outcomes for employees and organizations. This research explores the associations between employees’ experiences of autonomy at work and organizational deviance. We also investigate the mechanisms underlying this association and the possible role of identified motivation as a mediator of this relation. Three studies conducted in a variety of settings, countries, populations and assessment methods showed that employees who experience more autonomy at work tend to engage in lower levels of organizational deviance. Two studies also showed that this relation was mediated by identified motivation. Thus, employees’ experiences of autonomy at work seemed to foster higher levels of identified motivation towards work, which in turn predicted lower levels of organizational deviance. The present results may help guide managerial training and promote organizational cultures that are respectful of employee autonomy, potentially reducing the costs associated with organizational deviance.


Introduction
Organizational deviance, also labelled as counterproductive work behaviors directed at the organizations (CWB-O; Bennett & Robinson, 2000), is extremely harmful to organizations, costing employers billions of dollars annually worldwide (Ones, 2002).Employee CWB-O consists of volitional acts that harm or have the potential to harm organizations (Spector & Fox, 2005).It encompasses different types of deviant behaviors such as theft, production deviance (e.g., intentionally working slowly, taking long breaks, cyber loafing), and withdrawal (e.g., absence, lateness).Past research suggests that work contexts, and specifically the extent to which employees experience autonomy at work (Fox, Spector, & Miles, 2001), may influence employees' propensity to engage in CWB-O.One possible explanation for this finding is found within Self-Determination Theory (SDT; Ryan & Deci, 2017), which suggests that employees who experience autonomy in their workplace are more likely to have internalized their organization's goals as their own, thus making CWB-O counterproductive for both the company and themselves.However, conflicting findings linking autonomy to higher levels of CWB-O have also emerged (Vardi & Weitz, 2003).In addition, previous research has only examined the autonomy-CWB-O relation using self-reported measures of deviant behaviors and has yet to explore the mechanisms underlying this relation.The present study thus investigated the relation between experiences of autonomy at work and CWB-O using a mix of self-reported and coded measures of deviant behaviors, and examined identified motivation at work (i.e., engaging in work because it is congruent with one's goals and values) as a possible mediator of this relation.

Determinants of CWB-O
Previous research points to many possible reasons as to why employees engage in CWB-O.Most research has focused on employees' personality traits (Berry, Ones, & Sackett, 2007) as possible predictors of deviant work behaviors.In contrast to research on stable personality factors, recent work has looked at contextual factors that may impact CWB-O, such as organizational justice and support, as well as managerial leadership (Jacobs, Belschak, & Hartog, 2014;Kessler, Bruursema, Rodopman, & Spector, 2013).Identifying contextual antecedents is crucial as they are generally under the control of organizations and can be used as levers against CWB-O.Available evidence suggests that organizations may also influence CWB-O by creating working environments that foster employees' experiences of autonomy (Fox et al., 2001;Roberts, Harms, Caspi, & Moffitt, 2007).Given the strong benefits associated with autonomy at work (Hardré & Reeve, 2009), a closer look at its effect on CWB-O was warranted.

Experiences of Autonomy at Work
Employee autonomy has been extensively studied within SDT (Ryan & Deci, 2000, 2017) .This theoretical framework posits that individuals thrive when their social environment supports the satisfaction of their basic psychological needs for autonomy, competence, and relatedness.Competence and relatedness refer to the need to feel that one has an impact on one's environment (competence) and shares significant relations with the people around him/her (relatedness).In contrast, the need for autonomy is defined as the need to feel a sense of agency and ownership of one's behaviors.For instance, employees who can choose the best way to perform a certain task experience more autonomy than those compelled to do their task in a specified way regardless of efficiency.Employees who experience higher levels of autonomy tend to display higher levels of well-being, engagement, motivation, and productivity at work (Baard, Deci, & Ryan, 2004;Hardré & Reeve, 2009;Moreau & Mageau, 2012).
In relation to CWB-O, two previous studies have hinted at the potential importance of autonomy in its prediction.The first study was conducted with a sample of students who also occupied external jobs.Results showed that when these employees perceived that they had some level of choice regarding their own work schedule, they tended to engage in less CWB-O (Fox et al., 2001).The second study investigated autonomy at work in members of the general population using a longitudinal design (Roberts et al., 2007).In this study, autonomy was measured by asking employees if they had a say in setting their own work hours and if they had a boss.Results showed that employees presenting higher levels of autonomy engaged in significantly less CWB-O.
Other studies have also documented a negative relation between managerial behaviors known to nurture employees' autonomy (Wang & Gagné, 2013) and CWB-O, but without directly measuring employees' experiences of autonomy.Specifically, these studies have shown that transformational leadership (Kessler et al., 2013) and supervisor support (Jacobs et al., 2014) are negatively linked to CWB-O.In contrast, one study observed an unexpected positive relation between autonomy and CWB-O in an Israeli sample (Vardi & Weitz, 2003).In this study, autonomy was assessed as the degree of choice and input in work activities, schedules, and evaluation.This last finding casts doubts on the exact nature of the relation between autonomy and CWB-O, suggesting that autonomy could also provide opportunities for engaging in more frequent CWB-O.Taken together, these studies suggest that employees' experiences of autonomy may influence their propensity to engage in CWB-O, though the direction of this relation is uncertain, warranting further investigation.H1.Experiences of autonomy at work will be negatively related to CWB-O.
To our knowledge, the explanatory mechanisms of the autonomy-CWB-O relationship have never been explored.However, SDT research has repeatedly shown that experiences of autonomy influence people's behaviors through their impact on motivation (Moreau & Mageau, 2013).In the work domain, autonomy-supportive practices contribute to high-quality work motivation (Gagné & Bhave, 2011), which in turn predicts positive outcomes for employees (Gagné et al., 2014).Drawing on this body of research, we postulated that work motivation may mediate the hypothesized relation between autonomy and CWB-O.

Work Motivation
Within SDT, work motivation has been investigated by contrasting different types of motivation, which have been classified as low or high in self-determination (Gagné et al., 2010).Non-self-determined or controlled types of motivation are external and introjected regulations (Ryan & Deci, 2000).External regulation refers to situations where the behavior is done solely for purposes external to the behavior itself.These purposes may include external incentives such as money, gratitude, or awards.With introjected regulation, external incentives have been internalized but the reasons underlying the behavior have not been endorsed by the individual.For example, introjected regulation occurs when behavior is driven by guilt, compulsion, or in order to maintain self-worth.In contrast, more autonomous types of motivation are identified regulation and intrinsic motivation (Ryan & Deci, 2000).Identified regulation occurs when the behavior has been internalized as having value in its own right and as being consistent with one's own goals.Finally, intrinsic motivation refers to situations where the object of the behavior is motivating in itself, usually because it provides enjoyment.
Past research has shown that more autonomous forms of motivation tend to be associated with adaptive behaviors and greater levels of well-being across life domains including work, education, sport, health, and exercise (Ryan & Deci, 2000), while controlled forms of motivation tend to predict poorer outcomes in these domains.Recently, research also showed differential relations across more specific motivation types and a variety of outcomes.For example, one study revealed that intrinsic motivation was the only type of motivation that was consistently positively associated with academic achievement over a one-year period, controlling for baseline achievement (Taylor et al., 2014).Koestner et al. (1996) showed that identified motivation towards politics was the only type of motivation that predicted actual voting behavior.Similarly, when students are motivated through identified regulation, they tend to perform better in school and to be more satisfied with their levels of academic performance (Burton, Lydon, D'Alessandro, & Koestner, 2006).At work, employees with higher levels of identified motivation (but also higher levels of intrinsic motivation) tend to show higher levels of optimism, job satisfaction, organizational commitment, and well-being (Gagné et al., 2010).
Taken together, these results suggest that identified motivation may be particularly important across multiple life contexts.In the work context, identified motivation should be a crucial predictor of engagement in work-related tasks (Gagné & Deci, 2005) because not all work tasks are enjoyable, yet they need to be done.It is also likely that identified motivation is a predictor of CWB-O specifically.When motivation for work is identified, employees have internalized the organizational goals as their own such that their organization's interests are more intertwined with their own.They are thus more likely to support these goals and abide by the organization's standards.This means that they would consider CWB-O, such as taking long breaks or leaving early, as being counterproductive not only for the organization's goal attainment but for their own goal attainment as well.This should make them less likely to engage in such behavior.In contrast, intrinsic motivation should not play a similar role in preventing CWB-O because, when facing difficult or uninteresting challenges, employees whose work motivation is solely intrinsic may actually consider CWB-O as a way of avoiding meeting such challenges.Employees whose motivation for work is more identified would instead be more likely to value these challenges, making CWB-O less tempting.
Given that autonomy is known to increase levels of identified motivation by facilitating the internalization of external values (Deci, Eghrari, Patrick, & Leone, 1994), and that identified motivation may reduce CWB-O by making these behaviors more at odds with the person's goals, we proposed that identified motivation may mediate the hypothesized link between employee autonomy and CWB-O.
H2.The relation between experiences of autonomy and CWB-O will be mediated by identified motivation.

The Present Research
The above hypotheses were tested using self-reported data from employees from two countries, Canada and Greece, as well as a mix of self-reported and behavioral data collected in a laboratory setting.The first and second studies relied on structural equation modeling (SEM) to test the relation between experiences of autonomy in the workplace and CWB-O among employees.The second study also verified the possible mediating role of identified motivation in the hypothesized relation between experiences of autonomy and CWB-O.Finally, the third and final study tested whether the relations identified in the first two studies could be replicated using an objective behavioral measure of deviance as an indicator of CWB-O, namely cheating in a task for one's own personal benefit.Bringing these three different studies together allowed us to test the robustness of the proposed associations among autonomy, identified motivation, and CWB-O across different samples, methodologies and measures.

Participants and Procedure
Participants were 144 employees (51.8% female) of an international food company, were aged between 25 and 63 years (M = 39.32 years; SD = 8.23 years), and had worked for this company for an average of 8.69 years (SD = 7.15 years).Employees occupied a variety of positions: (office support, 5.5%; technician, 12.2%; professional, 40.9%; senior professional, 26.8%; director, 10.4%; executive committee member, 4.3%).Recruitment took place at the company's Canadian head office.Researchers first sent an invitation email to all employees.Interested employees were provided with a link to an online questionnaire.To increase participation, researchers also visited the company's offices with laptops to provide opportunities to an even greater number of employees to complete the online questionnaire.Most of the sample (79.6%) completed the questionnaire in French, while the remaining participants completed the questionnaire in English.All employees who participated did so following an informed consent procedure.All scales were translated to French using a standard back-translation procedure (van de Vijver & Hambleton, 1996).

Experiences of Autonomy in the Workplace
Participants' experiences of autonomy in the workplace were assessed using the three positively worded items from the need for autonomy subscale of the Work-related Basic Need Satisfaction Scale (Van den Broeck, Vansteenkiste, De Witte, Soenens, & Lens, 2010; sample item: I feel free to do my job the way I think it could best be done'').Employees were asked to rate their level of agreement with each item on a Likert scale ranging from Does not agree at all (1) to Strongly agree (7), while thinking about their working environment in general.This scale had satisfactory scale score reliability (α = .75).

Employee CWB-O
Participants' CWB directed at the organization were assessed using a 6-item version of the organizational deviance scale (Bennett & Robinson, 2000; sample item: During the last year, I intentionally worked slower than I could have worked'').These items were selected from the original 12-item scale because they had the highest loadings on the organizational deviance factor (loadings between .61 and .68;Bennett & Robinson, 2000), and also because their prevalence over a one-year period was the highest in the validation study (rates between 54.1% and 78.5%; Bennett & Robinson, 2000, Study 2).As it is the case for the original scale, the items were answered on a frequency scale ranging from Never (1) to Every day (7).The original scale showed satisfactory scale score reliability as well as good convergent validity with other scales of property and production deviance and with constructs of physical and psychological withdrawal at work (Bennett & Robinson, 2000).The scale score reliability of the scale in this study was low but acceptable (α = .67).Furthermore, because this study relies on fully latent SEM models, the relations between the latent CWB-O construct and the other constructs are estimated while controlling for the measurement error related to this low level of scale score reliability.
The organizational deviance scale provides valuable information on employee CWB-O.In the present sample, 48 employees (34%) had never engaged in any of the deviant behaviors listed while 30 employees (21%) engaged in at least one deviant behavior more than once or twice per year.Furthermore, very few employees (N = 5; 3.5%) reported having engaged in one or more deviant behaviors at least on a monthly basis.

Results and Discussion
Means, standard deviations and correlations for all variables for all studies are presented in Table 1.The relation between employees' experiences of autonomy at work and CWB-O was examined through SEM using the robust maximum likelihood (MLR) estimator implemented in Mplus 7.2 (Muthén & Muthén, 2014), which is robust to non-normality.Both experiences of autonomy and CWB-O were specified as latent variables that captured the common variance across their respective items.The fit of the model was evaluated using the Chi-square statistic (χ²), the Comparative Fit Index (CFI), the Tucker-Lewis Index (TLI), and the Root Mean Square Error of Approximation (RMSEA).Values greater than .90and .95for both the CFI and TLI respectively indicate adequate and excellent fit to the data, while values smaller than .06and .08 for the RMSEA respectively suggest an excellent and acceptable model fit (Hu & Bentler, 1999).
According to these guidelines, the fit of the present model was adequate, χ² (26) = 39.57,p = .04;CFI = .93;TLI = .90;RMSEA = .06.The results showed that autonomy was significantly and negatively related to employee CWB-O, β = -.22,p = .03.Thus, in line with past research (Fox et al., 2001), Study 1 supports the idea that employees' experiences of autonomy at work is related to their CWB-O.The more employees experienced autonomy in their workplace, the less they tended to engage in CWB-O.).Questionnaires were distributed electronically to a list of organizations in Greece and then relayed to employees within these organizations.Interested employees were directed toward the online questionnaire and were asked, following informed consent, to complete the scales described below.However, the electronic system used for data collection purposes did not allow us to identify the organizational membership of participating employees for purposes of confidentiality.All scales were translated to Greek using a standard back-translation procedure (van de Vijver & Hambleton, 1996).

Experiences of Autonomy in the Workplace
In this study, experiences of autonomy in the workplace were assessed using three items taken from the same original scale as in Study 1, selected based on their factor loadings in Van den Broeck's ( 2010) study.Although this scale's items had the highest loadings in the original validation study, its scale score reliability was low in the present study (alpha = .53).This reinforces the need for this study to rely on a fully latent SEM model that corrects for measurement error when estimating relations among key constructs (Bollen, 1989).

Employee CWB-O
Participants' CWB-O was assessed using the translated version of the organizational deviance scale used in Study 1, which resulted in an acceptable estimate of scale score reliability (α = .76).

Identified Motivation
Participants' identified motivation at work was assessed using the Greek adaptation of the 3-item identified motivation subscale from the Motivation at Work Scale (Gagné et al., 2010; sample item: I put efforts in my job because this job fits my personal values; α = .65).This subscale, validated in English and French, has shown satisfactory scale score reliability in the validation paper (from .81 to .89) and has been related to job satisfaction and well-being (Gagné et al., 2010).
Results revealed that experiences of autonomy at work were strongly and positively related to identified motivation (β = .77,p < .001),which was in turn significantly and negatively related to CWB-O (β = -.30,p = .003;see Figure 1).In addition, inspection of the indirect effect of autonomy on CWB-O as mediated through identified motivation, showed that autonomy indeed indirectly predicted employee's levels of CWB-O (β indirect = -.23,p < .05).nt prepared d to count lose, on a the signal steps (i.e., ber).After below.At ad written f matrices than they actually did.This was possible because the original test sheet on which the participants completed the task was supposedly thrown away without a way to link it to its owner.However, following Shu, Gino, and Bazerman's (2011) procedure, a specific number in one of the matrices (not included in the solution of that matrix) differed for every participant and corresponded to participants' identification number.These specific numbers were also written, unknown to participants, on their answer sheet where they indicated how many matrices they claimed to have solved.Participants' actual performance could thus be compared with their own report of solved matrices.Deviant behavior was defined as participants saying that they had solved more matrices than they really had (presumably to get more money or to self-enhance).An index was created by subtracting the number of completed matrices as coded from the test sheet from the number of matrices participants claimed they had solved.Scores above the zero value on this index indicated deviant behaviors.

Experiences of Autonomy
We measured participants' experience of autonomy in the task with a 3-item scale adapted from other autonomy need satisfaction scales (sample item: "I had the opportunity to make choices"; Gagné, 2003;Sheldon & Gunz, 2009).Items were selected based on their relevance to the laboratory context.The scale score reliability of this scale was low but acceptable (α = .68).

Identified Motivation
Identified motivation was measured using the 3-item identified motivation subscale from the Situational Motivation Scale (SIMS; Guay, Vallerand, & Blanchard, 2000).The scale score reliability of this scale was satisfactory (α = .82).

Results and Discussion
Given that the behavioral measure of deviant behavior was an observed variable and because the sample size was smaller than for the other studies, we tested the full mediation model with a path analysis approach, using the MLR estimator available in Mplus 7.2.(Muthén & Muthén, 2014).The obtained model showed adequate fit to the data, χ² (1) = 1.55, p = .74;CFI = .98;TLI = .93;RMSEA = .07. Results showed that autonomy was positively related to identified motivation (β = .33,p < .001)and that, in turn, identified motivation was negatively related to deviant behavior (β = -.25, p = .02).In addition, the indirect effect of experiences of autonomy on deviant behavior, through identified motivation, was again shown to be significant (β indirect = -.06,p = .04).
The present results support the hypothesis that experiencing autonomy results in less deviant behavior.People who felt autonomous in an experimental situation engaged in the proposed task with higher levels of identified motivation.In turn, people who were motivated in a more identified way were less likely to engage in behaviors that violated the task rules.

Summary
Three studies consistently showed a negative link between experiences of autonomy and deviant behavior, supporting H1.Employees who experienced more autonomy tended to engage in lower levels of CWB-O.Likewise, students who experienced more autonomy during an experimental task were less likely to engage in deviant behavior.Two of these studies further showed that this relation is fully mediated by identified motivation, supporting H2.It thus seems that work environments that foster employees' experiences of autonomy may facilitate higher levels of identified motivation at work, which in turn predict lower levels of CWB-O.These results appear to be robust to variations in culture (Canada versus Greece), in the measurement of autonomy (each study used a distinct set of items), to type of deviant behavior assessments (questionnaires versus observed behaviors), and to type of participants (employees versus students), suggesting that autonomy and identified motivation may be consistent predictors of CWB-O.

Implications for SDT
Taken together, these results have implications for Self-Determination Theory and for research on organizational deviance.First, by showing that experiences of autonomy are negatively associated with CWB-O, this research provides further support to the proposition that autonomy is associated with optimal employee functioning (Gagné & Bhave, 2011).Given the fact that autonomy support has been shown to predict performance and productivity (Baard et al., 2004) in addition to being linked to work satisfaction and greater mental health (Moreau & Mageau, 2012), it seems likely that promoting employees' experiences of autonomy is indeed an essential part of effective management.

Implications for HRM
The current results are particularly important for human resource management given that managers are more likely to make attempts at compelling employees to be obedient and diligent rather than granting them autonomy (Hamel & Breen, 2013).In the specific context of CWB-O, past research has shown that some managers try to reduce employees' CWB-O by behaving in more controlling ways (i.e., abusive supervision; Lian, Ferris, Morrison, & Brown, 2014).However, unlike objects that need to be programmed or fixed, human optimal functioning is best fostered using autonomy support instead of control (Moreau & Mageau, 2013).Our research suggests that granting autonomy may be a potential managerial strategy for reducing employee CWB-O.Furthermore, if CWB-O elicits attempts at managerial control (Lian et al., 2014), and control in turn induces more CWB-O (Lian, Lance Ferris, & Brown, 2012), a downward spiral may occur.Managers engaging in more autonomy-supportive practices could instead prevent this unwanted outcome.Future research should investigate whether interventions aimed at increasing managerial support for autonomy could be effective to guide employees away from CWB-O in situations where these behaviors are already ingrained in the workplace.
Results also showed that autonomy may weaken the propensity to engage in CWB-O because people are more likely to internalize reasons for doing their work.As previous studies have shown, when employees have high identified work motivation, they are more optimistic in, committed to, and satisfied with their work (Gagné et al., 2010).They also experience greater well-being and show better performance in their work (Burton et al., 2006;Gagné et al., 2010).Promoting such high quality motivation thus seems a highly desirable outcome, and again this is done through fostering experiences of autonomy.

Limitations and Future Research
Despite shedding light on important psychological processes related to engagement in CWB-O, the present studies have limitations worth mentioning.First, all studies were cross-sectional in nature, meaning that no direction of effects can be clearly inferred.It is indeed possible that people who engage in less CWB-O are subsequently granted more autonomy in the workplace (Lian et al., 2014).Second, only self-reported experiences of autonomy were assessed, which provides a possibly inaccurate or biased assessment of affordances of autonomy in the workplace, usually coming from structures, job design, and managerial behaviors.In addition, sample sizes across the three studies were rather small, which may limit the stability of the findings.Replication of results with bigger and more diverse samples is needed.Abridged versions of the need for autonomy subscale of the Work-related Basic Need Satisfaction Scale ( Van den Broeck et al., 2010) were also used in Study 1 and 2, and these varied across the two studies, which limits the external validity of the findings.Third, although items from validated scales were used, four scales did not reach an alpha level of .70.However, these scales had few items and Cronbach's alphas are greatly reduced when the number of items is small.When alphas were calculated based on the assumption that they had 8 items using Streiner's (2003) prophecy formula, all were found to be fully satisfactory (≥ .70).Furthermore, because Studies 1 and 2 relied on fully latent SEM models corrected for measurement errors, these lower levels of reliability are unlikely to have played a role in the estimated relations.Finally, Study 3 targeted university students and not actual employees and the experimental task used to code deviant behavior was not work-related.This makes the comparison across studies more difficult.Yet, the fact that the similar negative relations between autonomy and deviant behaviors were observed across all studies despite these methodological and psychometric variations suggests that this association may be particularly robust.It also suggests that autonomy and identified motivation may be key predictors for deviance at large, and not only CWB-O.Furthermore, as previously mentioned, one study (Vardi & Weitz, 2003) has found that autonomy was positively associated with CWB-O.While the present findings, as well as past research, have shown otherwise, we have yet to understand if there are specific situations in which autonomy could predict more or less deviant behavior.Fox and Spector (2006) suggested that employees with greater autonomy in their work might occupy hierarchically superior positions and have greater access to critical organizational resources.Power, as an employee or leader characteristic, has been shown to predict CWB-O and even corruption (Fox & Spector, 1999).In the Vardi and Weitz (2003) study, experiences of autonomy were measured using Breaugh's (1985) work autonomy scale which includes a measure of employees' perceptions of work criteria flexibility.It is possible that employees who have more flexibility in the way that they are evaluated also have greater power, and that power was a confounding variable in Vardi and Weitz' study.Future research assessing autonomy should thus distinguish among the different sources of experiences of autonomy (e.g., making decisions, having power over resources, being able to solve problems, believing in the company's purpose, having flexible work conditions, etc.), as these constructs may have different outcomes.

Conclusion
In conclusion, the present series of studies provides clues as to how managers and organizations may be empowered to prevent occurrences of CWB-O in their employees.While past research has mainly focused on employees' personality as a determinant of CWB-O, the present findings suggest that employee CWB-O is also influenced by the level of autonomy they experience in their work environment and by their work motivation.

Table 1 .
Means, Standard Deviations and Correlations across Variables for all StudiesNote.Theoretical range for all scales is from 1 to 7, except for Study 3 "Deviant Behavior", which is an observed absolute.Correlations between latent variables are presented for Studies 1 and 2. Items assessing autonomy experiences vary across studies.All correlations are significant at the p < .05level.Participants were 130 employees (61.5% female) working in different public and private Greek companies.Most participants' age fell in the 25-35 years category (60.0%), while others were in the 36-45 (26.2%) or 46-55 (13.8%) years category.Employees had worked for their company for an average of 7.05 years (SD = 7.46 years) and occupied a variety of positions:(clerk, 32.3%; technician, 4.6%; professional, 46.2%; manager, 10.0%;  director, 6.9%