Political Skill and Role Overload as Antecedents of Innovative Work Behavior in the Public Sector

We draw upon the theory of Conservation of Resources (COR) in positing political skill and role overload as influencing perceptions of either resource loss or conservation not previously studied in innovative work behavior. Based on a survey of 249 junior doctors in the United Kingdom, we found that role overload not only had direct positive effects on innovative work behavior but also negatively affects innovative work behavior, mediated through its effects on perceived organizational support. Political skill was positively associated with innovative work behavior, mediated through role-breadth self-efficacy. Our findings support a growing body of literature suggesting that engaging in innovative work behavior is a problem-focused coping strategy to deal with job demands and stressors. Current theorizing that job demands can have positive effects on innovative work behavior needs to be reconsidered given alternative negative effects suggested by COR.


Introduction
There is widespread agreement that organizations should actively support employee innovative work behavior (IWB) to respond to today's dynamic business environments (Getz & Robinson, 2003;Unsworth & Parker, 2003;Van de Ven, Polley, Garud, & Venkataraman, 2008). This is no less important for public sector organizations that are also under enormous pressures to innovate. Indeed, the increasing number of policy documents both 3 national and international attest to the gravitas attached to the pursuit of this goal (Audit Commission, 2007; Australian National Audit Office [ANAO], 2009; Organization for Economic Co-operation and Development [OECD], 2012). Alongside this burgeoning policy documentation has been an increasing body of empir-ical research on public sector innovation (Bason, 2010;Borins, 2001;De Vries, Bekkers, & Tummers, 2016). This recognizes that differing contexts associated with private and public sector organizations exert influence on capacities for innovation. Much of this literature suggests there are greater impediments to IWB in the public sector, given the nature of how public sector organizations work and are structured, as well as very different governance issues affecting them (Damanpour & Schneider, 2009;Fernandez & Moldogaziev, 2012). These include, for example, the lack of reward or incentives for employees to innovate, the costs of failure (particularly in terms of exposure to the media) should things go wrong, the lack of competitive pressure to innovate and strict agency regulation (Borins, 2001;Bysted & Jespersen, 2014). Bos-Nehles, Bondarouk, and Nijenhuis (2017), in a case study of IWB in the Netherlands fire services, high-lighted how strong formalization to secure quality in public services combined with strict agency control and 2014. Of these, 54 studies (approximately 30%) focused on innovation at the individual level. These highlighted individual characteristics such as employee autonomy, commitment, and creativity, as key factors involved in the generation and adoption of innovation. They concluded, however, that most studies lacked a clear theoretical underpinning and that more work was needed to understand what was the "publicness" of public sector innovation that distinguished it from the private sector; that is, what factors distinguish and promote 4 IWB in the public sector that might be different in some way to that in the private sec-tor. We take up this challenge by providing new insights into factors associated with IWB in a public sector setting. We make two significant contributions to the literature in this area. First, we adopt the concept of publicness as defined by Bozeman and Bretschneider (1994), to guide the selection of antecedents we believe to be of particular significance to IWB in public sector organizations. Next, we draw upon Conservation of Resources (COR) theory (Hobfoll, 2001) to further develop our understanding of the antecedents to IWB. Specifically, we posit and examine relationships between political skill and IWB based on the notion that this individual characteristic should support resource conservation and gain as suggested by COR theory. We also examine the effects of role overload, a form of job demands not previously explored in relation to its effects on IWB and which is widely reported as a particular concern in public sector organizations. We address the following research question: 1. Research Question 1: How do political skill and role overload contribute to IWB among public sector employees? 5 Our findings further our understanding of how job design and individual characteris-tics play a significant role in determining IWB in the public sector. We conclude by highlighting practical strategies for enhancing IWB in the workplace arising from our findings.

Political Skill and Role Overload: A Job Resource and Job
Demand Significant to Public Sector Work Environments Bozeman and Bretschneider (1994) proposed three dimensions of "publicness" associ-ated with government ownership, government funding, and degree of government oversight/regulation. All three dimensions led them to define "publicness" as "a char-acteristic of an organization which reflects the extent the organization is influenced by political authority" (p. 197). These organizations lack profit incentives and instead experience considerable political oversight with the involvement of many different interest groups (Rainey & Bozeman, 2001). Consequently, we highlight the extent to which publicness gives rise to organizational politics as one characteristic that may have implications for IWB in these settings. Although all organizations experience organizational politics to varying degrees, it is its particular salience in public sector organizations that sets it apart in comparison with those in the private sector. In highly politicized work environments such as these, far greater emphasis is placed on political skills to get work done (Raffel, Leisnik, & Middlebrooks, 2009) whilst employees in the public sector have been found to have a higher level of political skill compared with those in the private sector (Sharma & Hussain, 2013).
Political skill is part of the broader construct of social effectiveness. It captures those abilities an individual uses to observe the social environment, interpret the actions of those around them, and posit a strategy in response to influence others (Ferris et al., 2005b;Harvey, Stoker, Hochwater, & Kacmar, 2007). A number of studies have shown 6 that being adept with such skills can lead to positive benefits or outcomes for individuals (Andrews, Kacmar, & Harris, 2009;Shi, Johnson, Liu, & Wang, 2013) and has been suggested in particular to be a form of job resource (Kimura, 2014;Li, Sun, & Cheng, 2017).
We also highlight the concept of role overload as a form of job demand or situa-tional constraint more typically found in public sector organizations. Role overload has been identified as a significant role stressor, and studies have found it to be associ-ated with a range of negative outcomes including lower performance and burnout (Brown, Jones, & Leigh, 2005;Jones, Chonko, Rangarajan, & Roberts, 2007).
Role stressors are aspects of job design that describe environmental demands that exceed and/or strain employee coping resources (Lazarus & Folkman, 1984). Although indi-viduals in any organization may experience role overload, there is evidence that this is a major problem that occurs across the range of organizations within the public sector such as health care (Roslan, Noor Hazlan, Nor Filzatun, & Azahadi, 2014), social care (Skills for Care, 2015), and local government (Pieterson & Oni, 2014). One explana-tion as to why role overload maybe particularly problematic in the public sector is the relative inexperience of managers in managing downsizing compared with their pri-vate sector counterparts. It has been suggested that this results in the process taking far 7 longer and more drawn out, perhaps because of the emphasis placed on redeployment to protect the employment of staff (Ashman, 2015).
This results in staff in the depart-ments affected having to undertake additional responsibilities on top of what they currently do. Other explanations might lie in the high levels of staff turnover or staff shortages in some areas of the public sector. This results in current staff to take on more work to provide cover (Skills for Care, 2015).

IWB: A COR Perspective
Defined as, ". . . the intentional introduction and application within a job of ideas, processes, products and procedures that are new to that job and which are designed to benefit it . . ." (West & Farr, 1990, p. 9), IWB is recognized as more than being cre-ative, but instead captures a domain of behavior where individuals are involved in idea generation, idea promotion, and idea realization as a set of discontinuous activities (Yuan & Woodman, 2010). Research on IWB has taken place alongside that investi-gating creativity. Although conceptually sharing some overlap in that both involve generating novel ideas, the former is a broader construct in that it also includes the application of these ideas in practice. IWB thus also involves promoting creative ideas and engaging in implementation-based activities (Choi & Chang, 2009;Somech & Drach-Zahavy, 2013).
For example, convincing work colleagues to implement new ways to do things can prove challenging, especially if these colleagues are adverse to change. Janssen (2003) argued that workers who try to 8 pioneer new ideas to change the status quo, inevitably places them in conflict with their coworkers. He suggests that conflict arises because workers wish to avoid the uncertainty and anxiety that change provokes. However, Janssen (2000Janssen ( , 2004 also demonstrated that job demands might actually increase employees'innovative behavior. He found that job demands prompted IWB because employees try to develop new ways to manage these demands. From this perspective, IWB may serve as a problem-focused strategy in times of stress.
Both individual and organizational factors appear to predict IWB.
Individual fac-tors have highlighted a range of personal characteristics and traits (Woods, Mustafa, Anderson, & Sayer, 2018) including intrinsic interest (Yuan & Woodman, 2010), problem-solving style (Scott & Bruce, 1994), and knowledge-hiding (Cerne, Hernaus, Dysvik, & Skerlavaj, 2017). Organizational factors such as supervisory support and leadership style have also been identified (Bos-Nehles et al., 2017;Yuan & Woodman, 2010). Indeed, recent research suggests that supportive relationships with managers and coworkers may mediate relationships between high involvement HR practice and IWB (Prieto & Perez-Santana, 2014). A further area of interest, from a HRM perspective, are findings showing aspects of job design also to have effects.
Recently, a COR perspective (Hobfoll, 2001) has been applied as a means to further our understanding of why particular individual and contextual factors may lead to IWB. COR theory posits that individuals must invest resources to recover from losses and protect against resource loss. In essence, it is a motivation theory in that it describes what drives individuals to maintain existing resources and seek to acquire new ones when faced with stress. Those with greater resources are better able to do this, whereas those with lesser resources can find themselves in a spiral of resource loss that can lead to emotional exhaustion or burnout (Freedy & Hobfoll, 1994). Individuals can then respond to this resource loss (or stress) through either using their remaining resources in an effort to recover what has been lost or withdrawal to conserve what remains. Resources can come in many forms that include objects, conditions, personal resources, and energy resources (Halbesleben & Bowler, 2007;. Montani, Dagenais-Desmarais, Giorgi, and Gregoire (2018) applied a COR per-spective suggesting that mindfulness could enhance IWB for employees experiencing negative feelings. They found that low activated negative affect positively predicted IWB when mindfulness was high. These findings suggest that mindfulness, by enhanc-ing attentional quality, is a mechanism that enables an individual to redirect personal resources when personal resources (such as negative affect) are low. Chen and Huang (2016) similarly suggested that personal engagement assisted in minimizing resource loss and found it to positively predict IWB. A study by Stock, de Jong, and Zacharias (2016) examined colleague and supervisor support and customer aggression as con-textual factors associated with either supporting resource gain 10 or hindering resource loss. They found partial support for their model in that customer aggression negatively predicted IWB. Kiazad, Seibert, and Kraimer (2014) also examined IWB as a strategy used by individuals in response to psychological contract breaches in organizations. They argued that psychological contract breach could be perceived as a loss of valued resources and that individuals will respond to that loss through engaging in IWB. Their findings supported the propositions of COR, in that employees were found to engage in IWB to acquire additional resources. We build on this work in similarly applying a COR perspective to posit that political skill (an individual factor) and role overload (a situational factor) have both direct and indirect effects on IWB.

Theory and Hypotheses
Political Skill and IWB IWB emphasizes building coalitions and finding key sponsors to get ideas translated into action (De Jong & Den Hartog, 2010;Scott & Bruce, 1994). We posit that politi-cal skills are a personal characteristic that assist individuals in gaining new resources that facilitate innovation. This includes gaining support from colleagues to implement ideas and in overcoming others' resistance to new ideas. Those with political skills 11 have been found to possess greater networking abilities important for strong relational ties, within which social capital and other internal resources are embedded (Li et al., 2017;Munyon, Summers, Thompson, & Ferris, 2015;Zhang et al., 2010). They also confer advantages in both relationship building and persuading others to implement new ideas (Hochwarter, 2012). Consequently, employees with political skills often report experiencing far less strain in response to job stressors (Kimura, 2014;Zellars, Perrewe, Rossi, Tepper, & Ferris, 2008). Empirically, Janssen (2005) has previously found perceived influence to positively predict IWB. Given that political skills involve personal influence behavior, we would similarly expect a positive relationship between political skills and IWB. More recently, Kalra, Agnihotri, Chaker, Singh, and Das (2017) drew upon social influence theory in positing employees' creative performance was positively associated with their political skills. They suggested that political skills enable individuals to influence others in such a way that information is shared with them which enhances the opportunity to identify more creative solutions and capture novel insights from others (Miao & Wang, 2016). We therefore hypothesize as follows: Hypothesis 1 (H1): Political skill will have direct effects on IWB.
While our arguments above suggest significant relationships between political skill and IWB might be expected, we extend our theorizing further by positing that political skills may also have indirect effects on IWB mediated specifically through role -breadth self-efficacy. Role-breadth self-efficacy refers to a specific form of self-effi-cacy which captures an employee's "perceived capability of carrying out a broader and more proactive set of work tasks that extend beyond prescribed technical require-ments" (Parker, 1998, 12 p. 835). It refers to a motivational state that captures an indi-vidual's beliefs about their abilities to undertake a range of tasks beyond prescribed technical requirements. Political skills enhance role-breadth self-efficacy because they enable individuals to develop a greater sense of control. This arises out of their ability to better understand and influence others around them (Ferris, Treadway, Brouer, & Munyon, 2012;Perrewe, Ferris, Frink, & Anthony, 2000). Individuals with political skills possess greater self-confidence that they can cope with job stressors. This is because they are able to secure more tangible resources from those they influence, thus providing them with a greater sense of mastery (Perrewe et al., 2004). Their high lev-els of social astuteness give them a much better knowledge of the workplace which again strengthens their experience of mastery and sense of control. A few empirical studies have also found political skill can positively predict self-efficacy beliefs (Jawahar, Mews, Ferris, & Hochwater, 2008;Semadar, Robbins, & Ferris, 2006).
The significance of role-breadth self-efficacy for predicting IWB draws upon Bandura's (1986) social learning theory. This posits that individuals need to be confi-dent that they can overcome challenges and believe that their proactive behavior will meet with success (S. L. Martin, Liao, & Campbell, 2013;Hwang, Han, & Chiu, 2015). Whereas individuals can possess narrow self-efficacy expectations related to undertaking specifically defined tasks, role-breadth self-efficacy reflects a far broader 13 level or generality of self-efficacy, suggested as necessary for engaging in proactive, IWB (Sonnentag & Spychala, 2012). This is important, since engaging in IWB is often met with resistance or cynicism and can involve high social costs (Crant, 2000;Parker et al., 2010). Expectancy theory also posits that individuals are more likely to engage in particular behaviors the more they assume will be successful (Vroom, 1964).  & Fritz, 2007;Parker, Williams, & Turner, 2006). IWB is recognized as a discretionary form of behavior that requires individuals to go beyond what is normally expected from fulfilling their job requirements. We should therefore expect role-breadth self-efficacy to similarly predict IWB. This leads to our second hypothesis: Hypothesis 2 (H2): Political skill will be positively associated with employee innovative behavior mediated through role-breadth selfefficacy.
Role overload and IWB. One of the more important developments in the study of the effects of job demands on IWB, are findings that job demands can have both negative as well as positive effects. Whereas job hindrances such as job insecurity have been found to generally have negative outcomes in terms of an individual's motivation and wellbeing, job challenges such as role overload appear to have more mixed outcomes. More specifically, whereas role overload can be thought of as causing resource loss, somewhat counter-intuitively it has a positive effect on employee motivation to enact behaviors to restore the loss.
Based on COR theory, we suggest that role overload prompts an 14 individual to search for strategies to enhance their personal resources.
In this sense, role overload provides the motivation for an employee to search for alternative means or ways to deal with their workplace. A key response to stress such as this, is for an employee to engage in creative ideation and develop new responses to manage the additional stress brought to bear (Fritz & Sonnentag, 2009). Bunce and West (1996) found that employ-ees who participated in a training program that was designed to enhance their IWB improved their response to occupational strain. Based on previous research examining job demands, we suggest perceptions of role overload should have a positive effect through prompting a problem-focused coping strategy (Hammond et al., 2011;Janssen, 2000). IWB thus captures the cognitive and behavioral resources an employee utilizes to mitigate job demands (P. Martin, Salanova, & Peiro, 2007). We therefore hypothe-size as follows: Hypothesis 3 (H3): Role overload will have direct, positive effects on IWB.
Previously, Sonnentag and Spychala (2012) highlighted the complex relationship between job stressors as situational constraints that might have both positive and nega-tive effects on employee IWB. They found that situational constraints (measured as 15 insufficient tools or information to accomplish job role) was negatively associated with role self-efficacy. This was because situational constraints undermined their belief that they could successfully accomplish their tasks and undermined their confi-dence to carry our broader proactive behaviors (or enactive mastery). Although having this negative effect on role self-efficacy and thereby undermining proactive work behaviors, they also demonstrated that job stressors had a direct positive effect on proactive work behavior based on control theory (Carver & Scheier, 1998). That is, situational constraints stimulate search for proactive solutions to overcome the job stressor. We too, suggest that whereas role overload as a job stressor will have positive direct effects on IWB, it might also have a negative effect due to its influence on per-ceptions of organizational support.
In this instance, we suggest that role overload will be perceived by individuals as the organization failing to show sufficient regard for their well-being. This will nega-tively affect their perceptions of perceived organizational support (POS). POS has also been found to diminish the aversive effects of strains and stressors in the workplace because it indicates the availability of material aid and emotional support (Riggle, Edmondson, & Hansen, 2009). Beyond the buffering effects on stressor-strain rela-tionships, POS has also been theorized to provide Hypothesis 4 (H4): Role overload will be negatively associated with POS.

Procedure
We chose health care as an area of the public sector in which to base our study. Junior doctors are expected to undertake clinical leadership roles where engaging in IWB is considered key to the sustainability of ongoing public funding of health care in the United Kingdom (Walshe & Davies, 2013). We contacted the medical deanery, respon-sible for the postgraduate education and training of junior doctors in the south of England to facilitate data collection. The deanery agreed to send an email to all junior doctors registered with them to request their participation in the research via an online questionnaire. We were aware that junior doctors are often required to work long hours which can deter participation in research studies. We therefore offered two £50 gift vouchers to those taking part in a prize draw. We sent details of the questionnaire and requests to participate in the research to 2,027 individuals listed as junior doctors with the deanery. We received 249 completed responses or a response rate of 12.3%. 17 The Sample The characteristics of the junior doctors completing the questionnaire were as fol-lows: 62.2% (155) were females and 36.5% (91)  (3) aged between 18 and 24 years, and 2.0% (5) aged between 45 and 49 years. Finally, just 1.2% (3) were aged 50 years or over.

Measures (Items for all scales are shown in the appendix)
Political skill. We measured political skill using six items measured by Ferris et al. (1999) that utilized a 5-point Likert-type scale (1 = disagree strongly to 5 = agree strongly), α = .80.
Self-reported innovation. IWB was assessed by nine items developed by Janssen (2000) and based on Scott and Bruce's (1994) scale for individual innovative behavior in the workplace. Three items refer to idea generation, three items refer to idea promotion, and three items refer to idea realization. Individuals rated how often they engaged in these behaviors in the workplace. The response format ranged from 1 = never to 7 = always, α = .84.
Control variables. We included age, gender, and years of experience as control vari-ables in our analyses as these have been shown previously to be associated with IWBs (Axtell et al., 2000;Janssen, 2005). 19

Data Analysis
We used AMOS v21 to undertake structural equation modeling to test our hypothe-sized relationships. Prior to commencing hypothesis testing, we conducted a confir-matory factor analysis (CFA) to establish the discriminant validity of the measures used in our study (Anderson & Gerbing, 1988). We compared three separate estimated models using our data. We first examined our data for com-mon method bias effects.

Addressing Common Method Bias Concerns
Given that we collected all our measures from the same source at the same time, we undertook a number of strategies to control for common method bias. First, we fol-lowed recommendations by Podsakoff,20 MacKenzie, Lee, and Podsakoff (2003) in the design of our survey. We attempted a psychological separation in completing measures in the survey by asking additional questions regarding junior doctors' experience in the region. We divided the survey into differing sections with detailed introductions to read at the start of each section. In this way, respondents were required to refocus their concentration at key stages when completing the survey. We also asked for opinions to be typed into free text boxes before respondents moved on to complete new scales. Finally, we used various numbered rating scales for our measures.
We followed this with a number of statistical tests to identify the extent to which common method bias affected our data. In the first step, we employed Harman's (1976) one-factor test. We entered items from each of our variables into SPSS and performed an exploratory principal components factor analysis, constraining to a one-factor solu-tion. This resulted in one factor accounting for 24.71% of the total variance. This is well below the 50% cutoff value that is often suggested as acceptable.
Next, we examined the extent of common method bias in AMOS by creating a latent common factor with all our variables included in the model. We constrained all the regression weights to the common factor to equal a then squared the unstandardized results to arrive at the percentage of common method bias. This result indicated only 1.7% due to common method variance. The next step we followed was to compare the standardized regres-sion weights obtained when running our measurement model with and without the common latent factor.
Differences greater than 0.2 indicate problems with common method bias. None of our estimates showed differences of this magnitude indicating no significant problems with common method bias. We therefore proceeded with our anal-yses for our full structural model without the need to retain the common latent factor.

Results
We began initial tests using simple raw score correlation analyses.
Inter-correlations are shown in Table 1 below. 22 Here, we can see that political skill was positively and significantly associated with both role-breadth self-efficacy and IWB. Role-breadth self-efficacy was also posi-tively associated with innovative behavior.
Role overload was negatively associated with POS, and role overload was also positively associated with IWB. There is also a positive and significant relationship between POS and innovative behavior. Together these give primary indications of potentially significant relationships among a number of variables included in our study.
Next, to test our hypotheses, we followed the bootstrapping procedure set out by Preacher and Hayes (2004). This involved us first testing the direct effects of role overload and political skill on IWB with the hypothesized mediators present in the model. This was then compared with the results obtained when only direct effects were present. In each case, we included our three control variables in our structural model. This corresponds to the Baron and Kenny (1986) approach for testing mediation. The results of this first analysis showed the path from political skill to IWB had a standard-ized regression weight of .17 (p < .05) in the absence of the mediator (role-breadth selfefficacy) but that this changed to a value of .04 (p = n.s.) when we included the indirect as well as the direct pathway. By contrast, we found the pathway from role overload to IWB was significant with a standardized regression weight of .14 (p < .05) when the mediator was present. Whereas the value changed to .09 (p = n.s.) in the absence of a mediator. According to the Baron and Kenny (1986) approach then, this would suggest a full mediation effect for the effects of political skill on IWB mediated through role-breadth self-efficacy, whereas the relationship between role overload and IWB is direct and not mediated through POS. We next proceeded with further analyses as recommended by Preacher and Hayes (2004). Specifically, we used the boot strapping procedure which has greater control over Type I and II errors. We set AMOS to undertake 5,000 resamples to arrive at the direct and indirect effects. were also significant. This is in contrast to the result we obtained relying only on the Baron and Kenny (1986) approach. We also found the direct pathway from role overload to IWB to be positive and significant (β = .14, p < .05, 95% CI = [0.003, 0.29]). Hypothesis 3, predicting direct effects of role overload on IWB was therefore supported. The positive result indicating a relationship between role overload and IWB mediated through POS also provides support for Hypothesis 4. Our full findings are presented in Figure 1. 25 Finally, we tested two further alternative models to that which we theorized. In the second model, we examined the extent to which the variables we examined predicted political skill. We posited that POS could lead to increased innovative behavior which in turn would predict greater political skill. In the same model, we also posited that role overload would be negatively associated with role-breadth selfefficacy and that role-breadth self-efficacy would be positively Importantly, we failed to find a significant relationship for POS as a moderator in the model (β = -.004, p = n.s.). Taken together then, we are confident that our findings offer a good explanation for our hypothesized relationships.

Discussion
We extend the literature on public sector innovation by highlighting the significance of political skill and role overload as factors associated with IWB. Our findings are consistent with previous research in showing a positive and direct relationship between job demands (in this instance role overload) and IWB. Previous studies have found general measures of job demands (Janssen, 2000(Janssen, , 2005P. Martin et al., 2007) (Janssen, 2005;T. W. G. Ng & Feldman, 2010 In this instance, employees may respond to role overload by looking for more innovative ways to perform their jobs. This could include considering how to delegate tasks or to follow procedures differently. This is consistent with the notion that role overload may cause stress, but that this may have positive effects on problem-solving ability, which is linked to IWB. Previously, research has also found that prob-lemsolving ability can moderate stress-job performance relationships (Young, 2012). 27 However, we also demonstrate the complexity of the relationship between job demands and IWB in showing a negative, indirect relationship mediated by POS. Role overload, although explained as inducing a learning-focused coping strategy resulting in IWB, can also have detrimental effects on IWB. Role overload was found here to be negatively associated with POS. Organizational support theory (Eisenberger et al., 1986) suggests that employee's innovative behavior arises as a result of felt obligation and the need to reciprocate in response to the care, support, and attention shown to the employee by the organization. Engaging in IWB is thus a form of social exchange.
We demonstrated that role overload negatively affects POS. Rhoades and Eisenberger (2002) argue that employees make attributional processes regarding how fairly they are treated by the organization and the level of support the organization shows them. Employees consider that stressors such as role overload can be controlled and will attribute this stressor as indicating a lack of care and support. A number of studies have previously found negative relationships between work stressors and POS (Villanueva & Djurkovic, 2009). Our finding here is thus consistent with previous theorization in understanding antecedents that might negatively affect POS.
Research in the area of stressors and creativity has also found that the relationship is far from straightforward ( Van den Broeck et al., 2010).
It can be related to how stress inducing the stressor is (Byron, Khazanchi, & Nazarian, 2010) or the extent to which additional job support or resources are available at the same time (P. Martin et al., 2007). In relation to role overload, we find this job demand can have both negative and positive effects on IWB as predicted by alternative COR and social exchange theoretical explanations. Given the much stronger and significant effects were found for the impact of POS on IWB compared with role overload, our find-ings would suggest caution 28 in necessarily assuming that the positive relationships found between job demands and IWB in previous studies are necessarily a good thing if they come at the expense of perceptions of organizational support.

Future research investigating the relationships between job demands
and IWB should there-fore also examine the effects on POS to gain a clearer picture as to any potential positive or negative effects of job demands. This is also supported by other recent research demonstrating POS as a mediator between work stressors and outcomes (Ahraemi, Barak, & Michalle, 2015;Marchand & Vandenberge, 2016). It should also be noted that we did test an alternative model where POS moderated the rela-tionship between role overload and IWB. Recently, Xu and Yang examined the effects of POS as either a mediator or moderator between job stressors and burnout.
They found significant effects for POS as a mediator but not as a moderator. They suggested that this might be explained because POS only captures a general measure of support received by the organization. Whereas support might only show a buffer-ing effect when the form of support more closely matches the coping necessary for a specific stressor. It may be the case that other more specific forms of support may act as more specific moderators here.
A further theoretical contribution is the positive relationship we found between political skill and IWB mediated through role-breadth selfefficacy. Farr and Ford 29 (1990) suggested that "Since change and innovation in a work role may involve both uncertainty about future outcomes as well as possible resistance from others affected by the change, the individual who does not possess a reasonable amount of self-efficacy faces considerable barriers" (p. 67). We suggest that the importance of role-breadth selfefficacy to IWB attributed here, is further evidence in support of COR theory as explaining IWBs in the public sector. Self-efficacy represents an individu-al's belief that they are able to cope with the demands associated with specific situ-ations and contexts drawing upon the resources available to them. As such, self-efficacy is seen as a central construct in COR theory that is thought to provide a generalized, protective function against the loss of personal resources (Hobfoll, 2001). Political skill has been theorized to provide self-affirming information to individuals regarding their competence, as well as a greater sense of personal control and argued to enhance an individual's self-confidence or self-efficacy beliefs (Ferris et al., 2007). Given that both self-efficacy and political skill are social-cognitive constructs, it is understandable that both should reflect aspects of individual control and therefore be positively related (Ferris, Perrewe, & Douglas, 2002).
It is due to this positive effect on self-evaluations of competence and control that others have argued explains why political skill can inactivate stress factors (Perrewe et al., 2004). We have shown here that political skill should be thought of as an instrumen-tal resource that arguably not only protects individuals from resource depletion but can also help them to acquire new resources. For example, political skill offers a capability to draw upon contacts, call on favors, and use influence to achieve what is needed.
Our findings have particular significance for understanding innovation in the public sector broadly and health care organizations more specifically. Despite some significant advances in our 30 understanding of innovation processes within health care, authors often stop short of prescriptive frameworks due to mixed find-ings found in the literature (Greenhalgh et al., 2004;Rye & Kimberly, 2007;Walshe, Harvey, & Jas, 2010). Instead, there is increasing recognition of the role organizational context plays in influencing innovation processes (Berwick, 2003). In particular, organizational factors and individual characteristics and motivations are identified as key factors implicated in determining the adoption, diffusion, and implementation of innovations (Caccia-Bava, Guimaraes, & Harrington, 2006;Van de Ven et al., 2008;Williams, 2011). These are argued to be key elements that influence the "absorptive capacity" of organizations (Zahra & George, 2002), defined as the extent to which new knowledge is identified, distributed, and trans-lated. Walshe and Davies (2013), in a recent review of health research policy and innovation in England in the National Health Service (NHS), suggested that research should be directed to understand more about how absorptive capacity can be developed. At the individual level, and certainly as regards supporting IWB, our findings offer support for the theory of COR as a useful lens to consider how specific contextual circumstances found in the public sector, might account for why some employees are better able to engage in IWB compared with others.

Practical Implications
There is some evidence that HR systems are associated with organizational level inno-vation (Allen, Adomza, & Meyer, 2015;Jiang, Wang, & Zhao, 2012). What seems less clear is how organizations should go about supporting IWB (Unsworth & Parker, 2003). There is unlikely to be one intervention or HR practice that will trigger innovation as increasingly we recognize the importance that organizational context consider-ations play in how innovations are adopted and implemented (Berwick, 2003;Choi, 2004). However, supporting individuals to undertake innovative behavior is recog-nized as a key aspect of the innovation process. Arguably then, employee development programs that increase the political skill and awareness of those working in public sector organizations (such as junior doctors) should enhance their capacity for promot-ing innovative behavior (Kimura, 2014). Elsewhere, it has been suggested that politi-cal skill can be integrated within a range of HR activities including selection, appraisal, and training (Bing, Davison, Minor, Novicevic, & Frink, 2011).

Limitations and Future Research
A limitation of our study is that all of our variables were self-reported and collected at the same time. The results from both our CFA and attempts to address common method bias alleviate concerns to a considerable extent. It should also be noted that a number of authors have argued that other ratings of IWB may not be as effective as selfrat-ings, the argument being that much of the process involves extensive problem-solving and reasoning that others (e.g., such as managers) are not able to observe (Janssen, 2000;Montani et al., 2018). However this aside, research has found consistency between self and other ratings of IWB (Janssen, 2000;Monteta, Amabile, Schatzel, & Kramer, 2010). Nevertheless, other sources, such as supervisees or coworkers, might be used to assess political skills. We should also stress that although our hypothesized relationships demonstrated a better fit to our data than when we examined an alternative model, we have not demonstrated causality. Longitudinal studies are needed instead. A further consideration is that both political skill and self-efficacy are social-cognitive constructs. Jawahar et al. (2008) argued that these constructs were conceptu-ally distinctive and that while self-efficacy beliefs were important for regulating 33 behavior in the task domain, political skills performed a similar function in the inter-personal domain, and as such was far more contextual.
While we obtained a moderate correlation here between these variables (0.25**), there does nevertheless appear to be potential scope for overlap between them. Not least since both are essentially motivational constructs. This may suggest our finding that role self-efficacy is a mediator between political skill and IWB could be spurious. Future research can be directed toward investigating the extent to which political skill contributes to greater variation in IWB over and above that of role self-efficacy and vice versa, to rule out any possi-ble redundancy in these two constructs as regards IWB.
Next, not all resources are necessarily instrumental in all settings or contexts (Hobfoll & Lilly, 1993), and we might expect to find some degree of variation across organizations within the public sector. We collected data from junior doctors. Health care organizations are widely recognized as major political arenas. The competition for resources by professions and treatment specialities, concerns for efficiencies from tax payers, and quality or safety concerns from patients combine to create highly polit-icized work environments. These stakeholders as well as staff in the organization are all likely to have vested interests in changes to the way things are done, or suggested innovations. This may suggest that in such environments the effects of political skill on IWB are particularly pronounced. There are also particular characteristics associ-ated with our sample of public sector workers that may preclude the generalizability of our findings beyond this professional group. Junior doctors experience considerable intensity of work stressors arising from high workloads, shift systems, and working long hours (Goehring, Bouviert, Kuniz, & Bovier, 2005). These particular public sec-tor workers then might be considered as experiencing particularly high levels of job demands beyond those 34 typically found. Stress among health professionals has been found to be particularly high, with one study showing 28% of workers to be showing above threshold symptoms compared with 18% of workers more generally in the United Kingdom (Firth-Cozens, 2006). Research has also reported that doctors tend to be less extraverted than the general population (Clack et al., 2004). Individuals with less proactive and extraverted personalities respond to job demands with greater use of withdrawal behaviors rather than through mastery or control strategies (Singh, Burke, & Boekhorst, 2016). This may mean that the effects of political skills and the strength of their relationship with role self-efficacy may have been weaker than might be the case among other public sector workers. Our sample might also be thought of as char-acterized by a high level of general mental ability compared with other groups of public sector workers (Shen & Comrey, 1997). This might suggest buffering effects regarding the direct effects of role overload on IWB because high general mental abil-ity would enable these workers to draw upon a wider repertoire of problem-solving skills in their response to this job challenge (Tadic, Bakker, & Oerlemans, 2015). Future research should therefore aim to replicate our findings in other work settings across the public sector as well as conduct comparative studies with populations drawn from the commercial sector.
Finally, we investigated political skills and role overload as antecedents of IWB that we argued are particularly significant in the public sector. This was informed by ideas 35 of "publicness" as described by Bozeman and Bretschneider (1994).
This work has tended to differentiate these sectors in terms of the nature of ownership and funding, and places a focus on the political nature of public sector organizations. However, other important differences have been highlighted in the literature. These include differences in human resource management practices, ethics and values, organizational commit-ment as well as leadership styles (Andersen, 2010;Berman, West, & Cava, 1994;Goulet & Frank, 2002;Nutt, 2000). Given that human resource management practices have been found to affect IWB as mentioned earlier, differences in HRM practices seems a key area to investigate in terms of their effects on IWB. Similarly, many orga-nizations in the public sector continue to retain many characteristics associated with bureaucratic/administrative organizational cultures associated with more risk-adverse or less entrepreneurial leadership styles (Currie & Lockett, 2011). Again, this might be an area for further research that distinguishes IWB between these two sectors.

Conclusion
Following calls for more research to focus on factors that might specifically affect public sector innovation, we investigated the role of political skill and role overload as potential antecedents of IWB in a public sector organization. We suggested that both these individual and situational factors are likely to be of particular significance in this context. Drawing upon the theory of COR, we posited and found support for a positive and indirect effect of political skill on IWB. We similarly found a significant and posi-tive relationship between role overload (as a form of job challenge) and IWB. This latter finding supported the notion that role overload can prompt IWB as a means to seek out new 36 resources to deal with this challenge. However, these positive effects need to be tempered in light of the negative relationship we found between role over-load and POS. These findings suggest a far more complicated relationship between job challenge and IWB that requires further exploration in future research.

Scale Items
Political skill (Ferris et al., 1999). 1. I find it easy to envision myself in the position of others. 2. I am able to make most people feel comfortable and at ease around me. 3. It is easy for me to develop good rapport with most people. 4. I understand people well.
5. I am good at getting others to respond positively to me. 6. I usually try to find com-mon ground with others.
Role-breadth self-efficacy (Parker, 1998 Contacting people outside the company (e.g., suppliers, customers) to discuss prob-lems. 9. Presenting information to a group of colleagues.
10. Visiting people from other departments to suggest doing things, differently.
Transforming innova-tive ideas into useful applications (idea realization). 8. Introducing innovative ideas into the work environment in a systematic way (idea realization). 9. Evaluating the utility of innovative ideas (idea realization).

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to 38 the research, authorship, and/or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: The authors would like to thank the NHS Thames Valley Leadership Academy for funding this research.