Role perceptions, collaboration and performance: insights from identity theory

ABSTRACT This paper uses identity theory to explain the link between public servants’ role perceptions, collaboration and performance. A longitudinal survey (N = 522) shows that the more public servants rely on a Weberian role perception, the more they focus on internal collaboration in performing their work, whereas the more public servants adopt a networking role perception the more they use external collaboration in order to enhance their performance. Further, an entrepreneurial role perception influences performance directly. This study empirically shows that public servants with various role perceptions all engage in forms of collaborative behaviour that benefit their performance.


Introduction
In the social sciences, there is a long running debate on how individuals relate to and work within organizations (Abdelnour, Hasselbladh, and Kallinikos 2017;Sluss, Dick, and Thompson 2011).Many studies focus on how social structures within organizations influence or limit individual behaviour, while others indicate that individuals act independently and make their own choices for the way they behave (Abdelnour, Hasselbladh, and Kallinikos 2017).In this study, we focus on the latter, the more active involvement of individuals by focusing on the concept of role perceptions.The term role perception refers to how individuals define their work role, such as how broadly they perceive it, what types of tasks, goals and problems they see as relevant, and how they believe they should deal with these (Parker 2007, 406).In this way, role perceptions enable individuals to address the varying demands of the complex and constantly changing organizational environment (Burke and Stets 2009).
Several studies have shown that role perceptions are important antecedents of performance (e.g.Clark, Zickar, and Jex 2014;Parker 2007).That is, the way employees give meaning to their role has an impact on how they behave (Burke and Stets 2009).However, the literature offers only limited understanding of how this relationship CONTACT Joëlle van der Meer j.j.s.vandermeer@essb.eur.nlworks (Deslatte and Swann 2020;Parker 2007).In response, the aim of this study is to examine the relationship between role perceptions and performance in greater detail.We focus on public servants working in a constantly changing public sector context where a variety of government reforms have affected the way public servants identify themselves in their role (Aschhoff andVogel 2019 Van der Steen, van Twist, andBressers 2018).As these government perspectives have also changed the way that public servants should collaborate to solve complex policy problems (Kruyen and Genugten 2020;Campbell 2016), we view collaborative behaviour as a mechanism through which public servants' role perceptions affect performance.
Traditionally, public servants were internally focused, following the formal boundaries of a bureaucratic organization and working according to rules and protocols in order to serve the common good (Hendrikx and Gestel 2017).This traditional perspective has long shaped public servants' role perception as being bureaucrats who follow rules and protocols in interpreting their role (Van der Steen, van Twist, and Bressers 2018).With the introduction of New Public Management (NPM) and then New Public Governance (NPG), public servants have been increasingly expected to connect with the external environment and to collaborate with non-state actors to enhance performance (Kruyen and Genugten 2020;Campbell 2016).Public servants who rely more heavily on the NPM perspective can be expected to see themselves as entrepreneurs who want to realize public values, while public servants who rely more on the NPG perspective should identify themselves as networkers who connect various stakeholders in a network ( Van der Steen, van Twist, and Bressers 2018).
Identity theory is used in this paper to understand how role perceptions are related to performance mediated by internal and external collaboration.A central argument of identity theory is that the way individuals act is to a large extent determined by how they identify with their role and that this therefore influences their behaviour (Hogg Michael, Terry, and Katherine 1995).In line with identity theory, we would expect that the way public servants perceive their role to have implications for the way they interact with others to boost their own performance (Hogg Michael, Terry, and Katherine 1995).We would expect that the more public servants perceive their role as that of a traditional bureaucrat, the more they will focus on internal collaboration to enhance their perceived performance (Campbell 2016).Conversely, the more public servants rely on entrepreneurial or networking role perceptions, the more they will engage in external collaboration to enhance performance.Public servants who rely on the latter two role perceptions can be expected to spend more time on moving between organizations, on developing new solutions to problems with a range of actors and on negotiating with them (Considine and Lewis 2003;Deslatte and Swann 2020).
Following this, our research question is as follows: To what extent is the relationship between a public servant's role perception and their performance mediated by internal and external collaboration?
The research question will be answered by studying public servants employed in a single large municipality in the Netherlands.A two-wave longitudinal survey (N = 522) has been carried out (February and April 2021).
This study adds to previous research in two ways.Firstly, most studies on employees' role perceptions and their effects have so far been executed in the private sector despite the role perceptions of public servants seeming highly relevant: understanding employees' role perceptions is seen as particularly important in a changing work context, and public servants operate in a complex, changing environment (Parker 2007).Although the role of public servants has often been discussed by public administration scholars (e.g.Rhodes 2016;Gestel et al. 2019), empirical evidence on the effects of public servants' role perceptions is scarce.Most studies that focus on the role of public servants are descriptive in that they identify, rather than test, the effect of different role perceptions (e.g.Selden, Brewer, andBrudney 1999 Van der Steen, van Twist, andBressers 2018).In this study, we contribute to the literature by more strongly linking public servants' role perceptions to their behaviour by examining the mechanism through which public servants' role perceptions affect performance.
Secondly, in this study, both internal and external forms of collaboration are used as mediating variables.With the rise of NPM and NPG, increasing attention has been given to external collaboration and, as such, an expanding stream of public administration studies has been focusing on public servants' collaboration with external parties (Brandsen, Steen, and Verschuere 2018).However, not all public servants will have an external focus.Public servants who to a great extent still perceive their role as bureaucratic are more likely to focus on internal collaboration.Therefore, studying both internal and external collaboration in relation to public servants' role perceptions will provide a more detailed insight into broader aspects of collaboration by public servants.
This paper is structured as follows.Below, in the theoretical section, we discuss the literature on public servants' role perceptions, collaboration and performance.Then, based on these insights, we formulate hypotheses.Following this, we elaborate on the research method.Subsequently, we present our analysis, followed by a discussion of the findings and the drawing of conclusions.

Background on role perceptions
Within the field of organizational studies, it has long been recognized that employees hold different interpretations of their role.Porter and Lawler (1968) for instance proposed that performance is dependent on the extent to which employees possess appropriate role perceptions for their job.Further, Wrzesniewski and Dutton (2001) argue that employees differ in how they see their jobs in achieving meaning and identity in the workplace.
Role perceptions are subjective, meaning that two workers with very similar jobs can have different perceptions of their role (Grobgeld et al. 2016).These perceptions are shaped by the environment and by personal attitudes, values and past experiences (Robbins and Judge 2017).Furthermore, it is important to note that role perceptions are cognitive beliefs rather than affective reactions such as job satisfaction and affective commitment.These beliefs can change in response to changes in the external environment or changes within the person (Grant and Hofmann 2011).As such, role perceptions are distinct from concepts such as personality (Parker 2007).There are different dimensions to, or types of, role perceptions, and which is the most relevant dimension or type of role perception will depend on the specific behaviours that one is trying to understand within a given context (Parker 2007).
In the public administration literature, the role of public servants has frequently been discussed, for instance by Downs (1967) and Selden, Brewer, and Brudney (1999), and more recently by Rhodes (2016) and Gestel et al. (2019).From these contributions, three important role perceptions seem to dominate: the bureaucrat, the entrepreneur and the networker (Brandsen and Honingh 2013;Rhodes 2016;Selden, Brewer, and Brudney 1999;Van der Meer, Vermeeren, andSteijn 2022 Van der Steen, van Twist, andBressers 2018).
Firstly, the bureaucratic role perception seems to reflect the traditional public servant, as described by Max Weber, someone who is neutral, impartial, predictable, safeguards procedures and is loyal to political authorities in order to serve the public good (Weber 1946).Public servants who rely on a bureaucratic role perception have a strong internal focus and follow the formal boundaries of their bureaucratic organization.Consequently, Hendrikx and Gestel (2017) refer to them as 'guardians' because public servants who rely heavily on a traditional role perception have the expertise to serve the public interest and to take unbiased decisions selflessly.Hence, we define the bureaucratic role perception as reflecting public servants who focus on rules and protocols in interpreting their role and who aim to provide standardized public services (Van der Steen, van Twist, and Bressers 2018).
The rise of NPM saw the introduction of public entrepreneurship, performance indicators, contracting out and a stronger user-orientation with the goal of helping public organizations become more dynamic and increase their efficiency.Subsequently, a degree of public entrepreneurship, characterized by risk taking and proactiveness, was promoted, whereby public servants should increasingly collaborate with multiple actors to improve the public sector (Deslatte and Swann 2020).Hence, the second role perception that we distinguish reflects public servants who see their role as being entrepreneurial.These public servants consider citizens to be customers and want to realize public values in an entrepreneurial way (Van der Steen, van Twist, and Bressers 2018).
Recently, a further shift towards NPG can be witnessed in which public servants are now expected to deliver public services in a network of external actors (Gestel et al. 2019).Hence, the third role perception we distinguish is the networking role perception.Networkers view their public organizations as a partner in collaborative networks, rather than as the main developer and executor of policies.They actively connect with stakeholders, and regard themselves as the 'linchpin' between various partners (Tummers and Knies 2016).
Macro-level public sector reforms have thus affected the role of public servants, resulting in various role perceptions.To date, most studies that have focused on the role of public servants are descriptive in that they only identify, rather than test, the effect of different role perceptions (De Graaf 2011).In this study, we more strongly link role perceptions to their behaviour by examining the mechanism through which public servants' role perceptions affect performance.

Identity theory: linking public servants' role perceptions to collaboration and performance
Although several studies have shown that role perceptions are an important predictor of performance (e.g.Clark, Zickar, and Jex 2014;Parker 2007), the literature provides only limited understanding of how this relationship works.It is argued that the way employees give meaning to their role has an impact on how they behave, therefore influencing their performance (Burke and Stets 2009).In this study, we assume that collaborative behaviour is the mechanism through which role perceptions influence performance (Deslatte and Swann 2020).Collaboration is generally defined as 'a process whereby two or more parties work with each other to achieve mutually beneficial outcomes' (Miles, Miles, and Snow 2005, 40).Traditionally, collaboration by public servants was mainly internally oriented -within the bureaucratic organization -but with the influence of NPM and NPG, public servants are increasingly expected to collaborate with external parties to deal with complex policy problems that go beyond the capacity of single public actors (Blijleven and Van Hulst 2021;Costumato 2021).Empirical evidence suggests that both internal and external collaboration positively influence performance (Whitford, Lee, and Su Jung 2010;Mitchell, O'Leary, and Gerard 2015;Provan and Milward 1995).Internal collaboration is expected to improve performance when public servants gain access to a broad pool of expertise, resources, knowledge and information, and this is shared with others within the organization (Whitford, Lee, and Su Jung 2010;Campbell 2016).Whitford, Lee, and Su Jung (2010) showed that different types of internal collaboration can have different effects on performance, and showed that collaboration within a single work unit has the strongest effect.Therefore, we further differentiate internal collaboration into collaboration with direct colleagues within one's formal team and collaboration across teams.External collaboration is expected to improve performance when public servants bring different perspectives and new resources to create solutions that go beyond the limited perspective and capabilities of a single organization (Lee and Hung 2021).
By applying identity theory, we seek to explain how role perceptions are, through internal and external collaboration, related to performance.Identity theory is concerned with explaining an individual's role-related behaviours (Hogg Michael, Terry, and Katherine 1995).The main concept within identity theory is 'the self', which is assumed to emerge from the interaction between the individual and society (Burke and Stets 2009;Stryker and Burke 2010).The self is composed of multiple role identities that provide meaning for the self, 'not only because they refer to concrete role specifications, but also because they distinguish roles from relevant complementary or counter roles' (Hogg Michael, Terry, and Katherine 1995, 256).Role identities are hierarchically arranged within the self, and the role identities that are positioned at the top are more likely to be invoked in a particular situation than role identities placed at the bottom (Hogg Michael, Terry, and Katherine 1995).Hence, through a process that is labelled self-verification, individuals engage in specific behaviours because they want to signal to others and to themselves what they find important (Stets and Burke 2005).
Identity theorists highlight that individuals can have multiple coexisting role identities.The self-verification process determines which of those role identities become relevant for one's behaviour provided that there is a match between an identity and behaviour (Sluss, Dick, and Thompson 2011;Burke and Stets 2009).The existence of multiple roles has also been noted by public administration scholars who argue that public organizations are influenced by multiple government perspectives, referred to as the layering of perspectives (Kruyen and Genugten 2020;Meyer et al. 2014).Due to this layering, public servants can simultaneously hold different role perceptions and can adhere to specific ones to lesser or higher degrees.In this article, we thus argue that macro-level public sector reforms are closely linked to roles at the individual level in organizations because these reforms influence public servants' interests, actions and understandings of the self (Aschhoff and Vogel 2019;Meyer et al. 2014).Through their interaction with their organization, individuals create a perception of their own role that can be categorized into bureaucratic, entrepreneurial and networking role perceptions.Further, the stronger employees rely on bureaucratic, entrepreneurial or networking role perceptions, the stronger they will want to express to others and to themselves who they are by engaging in related activities that benefit their performance.
Based on the above, one could expect public servants who rely most on a bureaucratic role perception to engage more in internal collaboration to enhance their performance, whereas public servants who rely more on entrepreneurial or networking role perceptions will engage more in external collaboration to enhance their performance.Public servants who interpret their role as more as that of a bureaucrat, will view collaboration as largely determined by their organization's formal divisions and hierarchy (Campbell 2016).These public servants are expected to be internally focused, and to rely on rules and protocols to provide guidance when collaborating.This is reflected in the following hypothesis: Hypothesis 1a: Internal collaboration positively mediates the relationship between a bureaucratic role perception and performance.
In contrast, when public servants identify themselves more with entrepreneurial or networking role perceptions, their patterns of behaviour will not be primarily determined by the organization's formal structure of processes but, rather, by what is necessary to solve, with others, complex problems (Campbell 2016).Public servants with a strong entrepreneurial or networking role perception will increasingly collaborate across organizational borders (Deslatte and Swann 2020).These public servants see the relevance of interacting with the broader environment to solve complex problems (Campbell 2016;Considine and Lewis 2003).Deslatte and Swann (2020) further indicated that public managers with a strong entrepreneurial orientation were more likely to engage in external collaboration with the aim of enhancing performance.
Further, the literature notes some differences regarding the entrepreneurial and networking role perceptions in relation to external collaboration (Van der Meer, Vermeeren, and Steijn 2022).Public servants who rely more on the entrepreneurial role perception will behave in line with management expectations.The extent of the external collaboration will therefore be determined within fixed frameworks aimed at achieving results (Deslatte and Swann 2020).Conversely, public servants who rely more on the networking role perception will collaborate because they want to go out and bring different parties together.These public servants are looking for tailor-made solutions which can go beyond the fixed NPM frameworks (Van der Meer, Vermeeren, and Steijn 2022).Although several authors have studied these different role perceptions of public servants, little is known about how these public servants actually collaborate.Consequently, we do not distinguish between the entrepreneurial and networking role perceptions when it comes to external collaboration.This leads to the following hypotheses: Hypothesis 1b: External collaboration positively mediates the relationship between an entrepreneurial role perception and performance.
Hypothesis 1c: External collaboration positively mediates the relationship between a networking role perception and performance.
The above arguments thus suggest that it is relevant to study whether the collaborative behaviour of public servants who rely on the different role perceptions differs in how it affects their performance.A conceptual model reflecting the above assertions is presented in Figure 1.

Procedure and respondents
We collected data through two questionnaires separated by a three-month gap.The first questionnaire was distributed in February 2021 (T0), and the second in April 2021 (T1).The first questionnaire measured public servants' role perceptions, internal collaboration, in terms of collaboration with direct colleagues and across teams, and external collaboration.The dependent variables, performance, was then measured in the second questionnaire.We differentiated between various organizational levels and assessed both individual and team performance.As such, this study uses longitudinal data where the independent and mediating variables are measured at T0 and the dependent variables at T1.The advantage of this lagged approach, over crosssectional designs, is that it provides a time sequence which is essential for establishing causality in the relationship between role perceptions and performance.This approach to data collection should also minimize common source bias (George and Pandey 2017).Nevertheless, due to unavoidable restrictions, both the independent variable and the mediating variable were measured at T0.We consider this to be acceptable because, following identity theory, we consider role perceptions to be closely linked to collaborative behaviour since the way individuals act is largely determined by how they identify their role.
Both questionnaires were distributed during the second COVID-19 lockdown in the Netherlands, when only schools and day-care centres were reopening at the beginning of February.All other public servants were compelled to work from home during the research period.The targeted respondents were public servants working for a large Dutch municipality and therefore mostly working from home.Employees were invited by email, distributed by the municipality, to participate in an employee panel to provide information on a variety of work-related topics.While the municipality organized this panel, our research team collaborated closely and were able to integrate our measurements into the study.
Invitations were sent to all public servants (13,287) who had an employment contract with one of the nine core departments of the municipality rather than hired externally.In total, 1360 employees signed up for this panel.The first questionnaire had 739 responses and the second 1279 responses.This difference can be explained by the fact that the second questionnaire was more heavily promoted by the organizers of the panel study.In total, 681 respondents (response rate of 50.1%) completed both questionnaires and only these respondents were included in this study.Further, respondents with missing data for any of the variables to be analysed were removed from the sample, along with two respondents who indicated that they never collaborated with others in any of the three suggested forms.This resulted in a dataset of 522 respondents who had completed both questionnaires with valid responses for all the key variables studied.
Our sample broadly reflected the overall panel members' characteristics in terms of age, tenure, function and department, although in the overall panel there were slightly fewer employees with a supervisory position (10.4%).Our sample also broadly reflected the municipality's broader workforce, although the dominant age range there was 56-60 (15.7%) suggesting our sample was slightly biased towards younger employees.Moreover, the department of Public Space Management (24%) was underrepresented in our sample, while Urban Development (12%) and Administrative and Operational Support (22%) were slightly overrepresented.
The respondents were guaranteed complete confidentially and anonymity, and the data were managed in accordance with the Dutch Personal Data Protection Act.Further, this research has been approved by the research school's ethics committee.

Measures
The items in all the scales are listed in Appendix 1.The questionnaire included scales that were validated and used in previous research.

Dependent variables: team performance and individual performance
To distinguish different levels of performance, both individual and team performance were assessed.It is often argued that team performance consists of 'more than simply adding up the individual performance of each team member' (Kozlowski and Klein 2000, 17).Rather, team performance is seen as emerging out of the complex interplay between individual performances and contextual factors such as interpersonal relationships among group members, group performance strategies and the nature of tasks (Han and Williams 2008).
Individual and team performance were measured using public servants' selfreported performance ratings.In the public administration literature, using employees' perceptions to measure performance is widely considered to provide valid information (Kim 2005;Whitford, Lee, and Su Jung 2010).A five-point Likert scale ranging from 'totally disagree' (1) to 'totally agree' (5) was used with a range of response items.Individual performance and team performance were both measured at T1.
To measure individual performance, we used the five items that make up the task performance scale of the Individual Work Performance Questionnaire (Koopmans et al. 2014) that was developed to compare measures of self-evaluated performance across different types of jobs.Here, a satisfactory Cronbach's alpha (.82) was achieved.
Team performance was measured using six items developed by Brewer and Selden (2000) and Kim (2005) and a satisfactory consistency was again achieved (α=.79).We followed the recommendation of Kim (2005) and comprehensively evaluated team performance by including the following performance-related values: efficiency, effectiveness and legitimacy.For team performance, a 'don't know' response option was included.

Role perceptions
Public servants' role perceptions were measured at T0.We transformed the statements that Van der Steen, van Twist, and Bressers (2018) developed for their Q methodology to cluster public servants in terms of different government perspectives, into survey questions.The bureaucratic role perception is measured with the item 'I particularly see myself as a public servant who implements the rules in the interest of the inhabitants of the municipality'.The entrepreneurial role perception is measured with the item 'I particularly see myself as a public servant who is enterprising and wants to realize public values'.Finally, the networking role perception is measured with the item 'I particularly see myself as a public servant who, as a networker, connects different stakeholders'.A five-point Likert scale ranging from 'totally disagree' (1) to 'totally agree' (5) was used.In this way, respondents could identify with multiple role perceptions, reflecting the public administration literature on the possibility of layering multiple government perspectives (Kruyen and Genugten 2020).

Mediator: Collaboration
Collaboration was measured at T0.The respondents were asked how many times they collaborated internally or externally in any of the following three forms: collaborating by exchanging information, knowledge and experiences; collaborating by aligning processes; and collaborating by working with others to create a product, service or programme.Internal collaboration was further differentiated into collaboration with direct colleagues within their own formal team and collaboration across teams.In addition, collaboration across teams was further broken down into collaboration with colleagues from other teams within their own department and collaboration with colleagues from other departments (α=.68).External collaboration was measured with a single item.All items were measured on a six-point Likert scale ranging from 'never' (1) to 'multiple times a week' (6).

Control variables
Six control variables were included.Age was treated as a continuous variable and ranged from 21 to 67.Tenure was similarly a continuous variable, ranging from 0 to 48 years.We coded supervisory position as a dummy variable (1= supervisory position).We also included team size, which was subdivided into four categories (1= < 10 people, 2 = 10-20 people, 3 = 21-30 people, 4= > 30 people).Salary grades were used to measure job level.Salary levels were subdivided into five categories (1= levels 1-3, 2= levels 4-6, 3= levels 7-9, 4= levels 10-12, 5= level 13 and above (an 'I don't know/ don't want to share this' option was also available)).Finally, we dummy coded whether respondents had direct contact with clients in their work (1= direct contact).

Analysis
We tested our hypotheses using a Structural Equation Model (SEM).This statistical methodology allows the full conceptual model to be simultaneously tested in a single analysis.We followed a two-step approach as suggested by Anderson and Gerbing (1988).The first step consisted of a confirmatory factor analysis (CFA) to test measurement reliability and validity.The second step tests the structural model (SEM).Both the CFA and the SEM steps were conducted using AMOS version 26.

Descriptive statistics and correlations
Table 1 presents the descriptive statistics of the main variables.On average, public servants scored 3.86 on the bureaucratic role perception scale, 4.15 for the entrepreneurial role perception and 4.09 for the networking role perception.Furthermore, on average, public servants rated the performance of their team as 3.86 on a scale from 1 to 5, while the average score for their own individual performance was 3.74.The most frequent form of collaboration indicated was collaboration with direct colleagues, and the least frequently reported was collaboration with external parties.Table 2 presents the correlations between the variables.This shows that the bureaucratic role perception positively correlates with the entrepreneurial role perception (.270) and that the entrepreneurial role perception positively correlates with the networking role perception (.400).However, there was no significant correlation between the bureaucratic and networking role perceptions.This indicates that public servants' perceptions of the three roles on the one hand can be considered as distinct, but on the other hand are interconnected, which fits with the literature on the layering of government perspectives.Furthermore, individual performance positively correlates with the role perceptions of bureaucrat (.201) and entrepreneur (.196).Team performance also positively correlates with these role perceptions (bureaucrat (.134) and entrepreneur (.152)).However, neither individual nor team performance correlate with the networker role perception.
In addition, we found several correlations between the different public servants' role perceptions and types of collaboration.The more public servants identified themselves as a bureaucrat, the more positively this correlates with collaboration with direct colleagues (.115).Further, collaboration across teams correlates positively with both entrepreneurial (.126) and networking (.220) role perceptions.External collaboration correlates positively with the role perceptions of both entrepreneur (.177) and networker (.263).

The measurement model
The overall factor structure of the variables was first examined using a CFA to reveal the factor structure of the study's constructs.The overall fit of the models was evaluated using chi square (CMIN/DF), a goodness of fit index (GFI), the adjusted goodness-of-fit index (AGFI), a comparative fit index (CFI) and a root mean square error of approximation (RMSEA).In order to improve the fit, we controlled for the errors associated with items 3 and 4 of team performance (both measuring efficiency) and items 5 and 6 of team performance (both measuring legitimacy).The calculated fit values of the resulting measurement model were: 2.12 (CMIN/DF), 0.96 (GFI), 0.93 (AGFI), 0.96 (CFI) and 0.05 (RMSEA), all of which are indicative of a good model fit.
To further assess the validity of the constructs, we calculated the factor loadings of individual performance and team performance.Each indicator significantly loaded on to the appropriate factor, and all loadings were above the 0.40 threshold (ranging from 0.41 to 0.85).
Finally, we calculated the Average Variance Extracted (AVE) for individual and team performances to indicate the convergent validity of the constructs, and the standardized factor loadings are presented in Table 3.The results show that the AVE of individual performance is above 0.50, indicating satisfactory convergent validity.The AVE of team performance (0.41) was below this threshold.However, Fornell and Larcker (1981) argue that an AVE above 0.40 can still be considered acceptable provided the composite reliability (CR) is above 0.60.In our study, the CR of team performance was 0.80, so the construct validity can be regarded as adequate.Another argument for not excluding any of the team performance items is that, in measuring team performance, we distinguished between three different performance-related aspects: efficiency, effectiveness and legitimacy (Kim 2005).Thus, although the factor loadings of the two items measuring legitimacy (items 5 and 6) are slightly low, it would be wrong to ignore legitimacy since it is relevant alongside efficiency and effectiveness in a public sector context (Kim 2005).

The structural equation model
In the second step, in a structural equation model, we added the control variables and the various potential causal paths between the variables in order to test our hypotheses.We also considered whether there was a correlation between individual performance and team performance as the literature shows that these are strongly related (Kozlowski and Klein 2000).
As this study includes mediation effects, we employed a bootstrapping method (Hayes 2009).We created 1000 samples (with replacement) from the available observed sample.Using such samples, one can establish robust estimates of the expected value and the variability of the statistics (Hox 2002).The overall fit of the structural equation model met the acceptance criteria with fit values of 2.55 (CMIN/ .817Individual performance 5 .617 Team performance 0.80 0.41 Team performance 1 .794 Team performance 2 .809Team performance 3 .639 Team performance 4 .598 Team performance 5 .507 Team performance 6 .411 DF), 0.92 (GFI), 0.89 (AGFI), 0.90 (CFI) and 0.06 (RMSEA), all of which indicate that the model has a good fit.
Figure 2 shows the significant relationships between the variables (i.e.those with a statistical significance level of <.05).The numeric values presented are the standardized regression coefficients (β).The results for the control variables can be found in Table 4, and for the indirect effects in Table 5. Hypothesis 1a posited that internal collaboration positively mediated the relationship between having the role perception of a bureaucrat and performance.The mediation analysis shows that the path from the bureaucrat role perception, through collaboration with direct colleagues, to team performance is indeed significant (β = 0.148 p = 0.011).Hypothesis 1a is thus supported.
Hypothesis 1b stated that external collaboration positively mediates the relationship between the role perception as an entrepreneur and performance.However, we did not find any significant mediating effect for the entrepreneurial role and therefore Hypothesis 1b is rejected.
Hypothesis 1c posited that external collaboration positively mediates the relationship between a networking role perception and performance.However, the results show that, counterintuitively, external collaboration negatively mediates the relationship between the networker role and individual performance (β = −0.029,p = 0.002).Hypothesis 1c is thus rejected.
Further, we found direct effects of both the bureaucrat and entrepreneur role perceptions on performance: when public servants mostly identify themselves as either an entrepreneur or a bureaucrat, this positively influences both individual and team perceived performance.
Turning to the influence of the control variables, we found that age negatively influences external collaboration, while both having direct contact with clients and a more senior job level positively influences external collaboration.The results also showed that having a supervisory position positively influenced Whereas age negatively influenced team performance, having a supervisory position had a positive influence.Finally, having a more senior job seemed to have a negative effect on self-assessed individual performance.

Discussion
The aim of this study was to examine the relationship between role perceptions and performance through the mechanism of collaboration.Various government reforms have affected the way public servants identify themselves in their role, and have also changed the way public servants are expected to collaborate in order to solve complex policy problems.We used identity theory to explain how public servants with different role perceptions engage in various forms of collaborative behaviour because they see these as beneficial for their performance.
We carried out a two-wave longitudinal study, and the results indicate that the relationship between a bureaucratic role perception and perceived team performance is partially mediated by collaboration with direct colleagues.Further, the relationship between a networking role perception and perceived individual performance is partially mediated by external collaboration.A perceived NPM role influences performance directly.
Our research expands existing research in four ways.
Firstly, with this study, we showed that identity theory is particularly relevant when trying to understand how public servants' role perceptions, through the mechanism of collaboration, relate to their perceived performance (Burke and Stets 2009).Through the process of self-verification, we found that those public servants who identify themselves as having a bureaucratic role will want to engage in internal collaboration to express to others and themselves what they find important, whereas public servants who perceive themselves in a networking role want to engage in external collaboration as a means to enhance their performance.Public servants who largely see themselves as having an entrepreneurial role may consider working efficiently and achieving results as more important aspects of their identity than collaborating with others, explaining the direct effect found on performance (Kruyen and Genugten 2020).
Secondly, our research extends existing studies that have identified various public servant roles (e.g.Selden, Brewer, andBrudney 1999 Van der Steen, van Twist, andBressers 2018) by showing the effects of different role perceptions on their behaviour.We have shown that the way public servants define their role affects their performance by directing their efforts towards specific behaviours (e.g. in how and with whom they collaborate).Although we did not make a theoretical distinction between entrepreneurial and networking role perceptions when it comes to external collaboration, our empirical findings do indicate several differences between them.Additional research could usefully further differentiate between the entrepreneurial and networking role perceptions and their effects.Moreover, there is a need to further investigate the unexpected negative relationship between a networking role perception, through the mediation of external collaboration, on perceived performance.One plausible explanation is that public servants are not always able to achieve the goals that they see as relevant to the perceived work role with which they identify.
Thirdly, this study extends our understanding of how individuals and organizations relate to each other by showing that individuals can draw on different role perceptions to deal with the complex and constantly changing organizational environment.Rather than adopting an organizational perspective, we focused on individuals who, within the organization, make their own choices as to the way they behave, and we show that these perceptions influence their behaviour.At the same time, we realize that individual behaviour is embedded within the broader organizational context and in our research, we did not take the organizational structures into account.Consequently, we would encourage research to further investigate the interaction between individuals and their organizations by taking the organizational context more into account (e.g.hierarchy, formalization, and centralization) (Abdelnour, Hasselbladh, and Kallinikos 2017).
Fourthly, our study expands existing research by emphasizing the importance of studying both internal and external aspects of collaboration.Although in the context of NPM and NPG, the involvement of various external non-state actors in the creation and production of public services has been studied extensively in the public administration field (Brandsen, Steen, and Verschuere 2018;Meier and O'Toole 2010), these studies do not fully capture the totality of the behaviour of public servants with different role perceptions.Our study shows that not all public servants are externally focused, and that those who adhere to the more traditional bureaucratic perception of their role collaborate mostly with colleagues from within their formal team.In this respect, the findings of Blijleven and Van Hulst (2021) are important by showing that the tensions experienced by public servants who engage with the public mostly surface in their interactions with other colleagues who do not prioritize collaboration with external parties.Our study empirically shows that role perceptions are an important element in explaining whether public servants consider collaborating with non-state actors to be part of their identity.In this way, our study strengthens the relevance of the role perception concept (Parker 2007).On this basis, we recommend future studies to adopt a broader concept of collaboration to capture the collaborative behaviours of public servants who understand their roles in different ways.
Finally, it is important to note that our study provides empirical evidence that public organizations are, as has been claimed, influenced by multiple government perspectives, known as the layering of perspectives (Meyer et al. 2014;Pollitt and Bouckaert 2011).As a consequence, public organizations have become a 'mixed zone' of public servants with different role perceptions (Burke andStets 2009 Van der Steen, van Twist, andBressers 2018).This finding accords with those of Aschhoff and Vogel (2019) and of Meyer et al. (2014) who used identity theory to show the individual-level consequences of public management reforms, resulting in multiple identities.In our study, we focused on various role perceptions, but did not examine how these role perceptions relate to each other within an individual.We nevertheless found that the different role perceptions were interrelated, suggesting a layering of perspectives.More research is needed to clarify how individuals combine different role perceptions and whether this creates role conflicts when public servants have more than one role perception (Meyer et al. 2014).
Despite the contributions of this study, there are some limitations that affect the analysis and results of this study.Firstly, our study is limited to public servants in a single Dutch municipality which affect the generalizability.Therefore, additional research in other organizations is needed before the results of this study can be more widely generalized.This is particularly relevant since our findings suggest that the organizational context might have an important influence on public servants' role perceptions.Hence, future research should explore whether these findings are consistent across other countries, other organizations and departments within organizations.
Secondly, this study has been executed during the very specific circumstances of the COVID-19 pandemic.The spread of the virus made working from home mandatory rather than voluntary for public servants.This might have particularly influenced the work of those public servants who primarily perceive a networking role since we found an unexpected negative relationship between holding such a perception and perceived performance.Although the specific COVID-19 context in itself provides valuable information, it imposes limitations when seeking a broader perspective through generalization.Future research should therefore investigate whether our results are valid in other contexts for instance in a working environment where virtual working is voluntary.
Thirdly, we used subjective perceptions of both individual and team performance rather than objective performance indicators.Although there is evidence of a strong correlation between perceptual and objective measures of performance, there are inevitably doubts when perceptual performance measures are used (Kim 2005).A suggestion for further research would therefore be to replicate our study by using objective performance indicators when comparing outcomes.In addition, when measuring individual performance, our focus was on task performance and this operationalization could have been too limited, especially for public servants who rely heavily on a networking role perception.It would therefore be worth expanding the operationalization of performance, for instance by including extra tasks, initiatives, innovative ideas and cooperating with others (Koopmans et al. 2014).
Finally, it should be noted that both the independent variable and the mediating variable were measured at the same time.More research is needed to determine whether behaviour does indeed immediately follow from role perceptions.This could, for instance, be addressed in a diary study that could provide greater insight into the close relationship between role perceptions and behaviour.

Conclusions
Although several studies have shown that role perceptions are important antecedents of performance, there has been limited understanding of how this relationship works.To deepen our understanding, we studied the relationship between a public servant's role perception and their performance, mediated by internal and external collaboration.Based on our findings, we have concluded that the way public servants perceive their role affects their performance by directing effort towards specific behaviours.As such, we empirically show how, in the context of macro-level public sector reforms, public servants' individuallevel role perceptions result in different forms of collaborative behaviour and performance.
Our findings have important implications for public organizations in that they suggest that they should take account of the fact that public servants execute their work based on a range of role perceptions that they act upon.Paying attention to this mix of public servants can become a strength provided public servants' perceptions are recognized and acknowledged.

Table 3 .
Factor loading, CR and AVE.