Job Resources to Promote Feelings of Pride in the Organization: The Role of Social Identification

Introduction The concept of resources, based on the Conservation of Resources theory (Hobfoll, 1988) from an individual perspective, has become increasingly popular in organizational research. In studies of organizational behavior, job resources have mostly been useful for understanding stress and strain (Halbesleben & Buckley, 2004), often with reference to the Job Demands and Resources model (JD-R model) (Bakker & Demerouti, 2007). This model suggests that work engagement is an important outcome related to resources. However, the relationships between job resources at different organizational levels and their associations with other positive dependent variables, such as positive group-based emotions, appear to have been overlooked in previous research. Halbesleben et al. define resources as ‘anything perceived by the individual to help attain his or her goals’ (2014, p. 6). The JD-R model identifies feedback, job autonomy, supervisor support, rewards, role clarity and participation as important job resources (see e.g., Bakker, Demerouti, & Euwema, 2005; Bakker & Demerouti, 2007). However, most JD-R studies have focused on negative outcome variables, such as burnout, ill health and repetitive strain. There has been insufficient research on how organizations provide or develop resources to help employees successfully complete their tasks and goals and thereby enhance positive group-based emotions (e.g., group-based pride) and performance (Bakker & Demerouti, 2007; Kraemer & Gouthier, 2014). Despite this, evidence suggests that group-based emotions are meaningful. For example, group-based emotional profiles have been shown to predict connections with the in-group and the avoidance of or confrontation with members of the out-group. Groupbased emotions are socially shared within the group and contribute to the motivation and regulation of intra-group and intergroup attitudes and behavior (Mackie, Devos, & Smith, 2000; Smith, Seger, & Mackie, 2007). One example of a positive group-based emotion that can be assumed to have desirable consequences for an organization and its employees is pride in the organization (organizational pride). Organizational pride is positively related to job satisfaction (Mas-Machuca, Berbegal-Mirabent, & Alegre, 2016) and has also been found to affect employees’ decisions to stay with their organizations (Appleberg, 2005). It also correlates with a reduction in employees’ stress levels, which in turn reduces their turnover intentions (Kraemer & Gouthier, 2014). According to the identity matching principle (Ullrich et al., 2004), job resources that encourage positive ORIGINAL ARTICLE


Introduction
The concept of resources, based on the Conservation of Resources theory (Hobfoll, 1988) from an individual perspective, has become increasingly popular in organizational research. In studies of organizational behavior, job resources have mostly been useful for understanding stress and strain (Halbesleben & Buckley, 2004), often with reference to the Job Demands and Resources model (JD-R model) (Bakker & Demerouti, 2007). This model suggests that work engagement is an important outcome related to resources. However, the relationships between job resources at different organizational levels and their associations with other positive dependent variables, such as positive group-based emotions, appear to have been overlooked in previous research.
Halbesleben et al. define resources as ' anything perceived by the individual to help attain his or her goals ' (2014, p. 6). The JD-R model identifies feedback, job autonomy, supervisor support, rewards, role clarity and participation as important job resources (see e.g., Bakker, Demerouti, & Euwema, 2005;Bakker & Demerouti, 2007). However, most JD-R studies have focused on negative outcome variables, such as burnout, ill health and repetitive strain. There has been insufficient research on how organizations provide or develop resources to help employees successfully complete their tasks and goals and thereby enhance positive group-based emotions (e.g., group-based pride) and performance (Bakker & Demerouti, 2007;Kraemer & Gouthier, 2014). Despite this, evidence suggests that group-based emotions are meaningful. For example, group-based emotional profiles have been shown to predict connections with the in-group and the avoidance of or confrontation with members of the out-group. Groupbased emotions are socially shared within the group and contribute to the motivation and regulation of intra-group and intergroup attitudes and behavior (Mackie, Devos, & Smith, 2000;Smith, Seger, & Mackie, 2007). One example of a positive group-based emotion that can be assumed to have desirable consequences for an organization and its employees is pride in the organization (organizational pride). Organizational pride is positively related to job satisfaction (Mas-Machuca, Berbegal-Mirabent, & Alegre, 2016) and has also been found to affect employees' decisions to stay with their organizations (Appleberg, 2005). It also correlates with a reduction in employees' stress levels, which in turn reduces their turnover intentions (Kraemer & Gouthier, 2014).
According to the identity matching principle (Ullrich et al., 2004), job resources that encourage positive

ORIGINAL ARTICLE
Job Resources to Promote Feelings of Pride in the Organization: The Role of Social Identification group-based emotions at the workgroup level should differ from those at the organizational level. For example, since perceptions of job resources vary, they cannot be expected to automatically generate pride at the group level. Therefore, although resources like job autonomy and supervisor support have been identified as antecedents to organizational pride (Kraemer & Gouthier, 2014), further research is necessary in order to determine which job resources affect group-based pride specifically. This is particularly important in complex organizations characterized by diverse structures and multiple operational groups (Ashforth & Johnson, 2001;Edwards & Peccei, 2010).
The main purpose of this study is to investigate the relationships between workgroup job resources and organizational job resources and how and to what extent they affect workgroup pride and organizational pride. Social identification is expected to play a part in this process as an intervening factor and contribute to the explanation of how pride varies in the workgroup and the organization. By understanding how job resources affects different identification processes and, ultimately, employees' positive emotions and behavior, organizations may be better able to decide where to focus their intervention efforts.

Pride in the organization and pride in the workgroup as group-based emotions
Pride is a particularly important emotion in the context of social behavior (Elfenbein, 2007;Tracy & Robins, 2007). Previous empirical research has identified two forms of pride in organizational settings: personal pride and collective pride. Personal pride is intrinsically motivated and relies on personal achievements, such as work quality, the sense of personal dignity or value and self-respect regarding personal work accomplishments (Lea & Webley, 1997), whereas collective pride is a desire to connect to an organizational group (Bouckaert, 2001).
According to Arnett, Laverie and McLane (2002), collective pride in an organizational context results from individuals' personal experiences of a particular organization. Collective pride partly arises from the belief that personal actions influence an organization's success and can be enhanced or diminished by one's own or other people's perceptions of it. Unlike the concept of personal pride, collective pride mostly results from relationships or affiliations, such as those formed in organizational memberships (Lea & Webley, 1997;Tracy & Robins, 2007). In order to further advance the existing literature, this study focuses on organizational pride and workgroup pride. Regardless of the organizational level it refers to, group-based pride is conceptualized as an emotional consequence of people's positive evaluations of shared experiences, interactions and events that are believed to are influence their group membership. This study makes an effort to distinguish between emotions relating to social identification (such as belongingness and satisfaction) and emotional reactions to feedback and pride in the group's accomplishments. Some factors related to workgroup pride and organizational pride are explored below.

Job resources related to group-based pride
Although the JD-R model acknowledges that job resources can be found structurally in the organization as well as at the interpersonal and individual levels (Bakker & Demerouti, 2007), the fact that it fails to clearly identify resources by their specific sources has been discussed and criticized (Halbesleben et al., 2014). A more mature model allows for job resources to be operationalized at numerous organizational levels (i.e., the individual, group, leader and organizational [IGLO] levels; see Nielsen et al., 2017). In line with this, according to the identity matching principle (Ullrich et al., 2004) a job resource from a particular level and source should more strongly relate to employees' emotions of pride at that level than job resources at other levels. Thus, job resources at different levels can impact group-based emotions differently at different organizational levels. For example, at the workgroup level, social support among co-workers has been identified as a group level resource that fosters the exchange of high quality interactions among individuals (Nielsen et al., 2017). Co-worker support (defined as the interpersonal transfer of instrumental or emotional resources between/among co-workers) is regarded as more informal and genuine than other types of support due to the general absence of unequal authoritative relationships among colleagues (Rousseau & Aubé, 2010). It has also been shown to be more valued and consistent over time than supervisor support (Ng & Sorensen, 2008). At the organizational level, it has been suggested that leader-member (supervisor-employee) exchange is an important resource that promotes employees' identification with and positive emotions towards their overall organization (He & Brown, 2013;Tangirala, Green, & Ramanujam, 2007). Social support from supervisors (defined as the interpersonal transfer of instrumental or emotional resources from the supervisor to the employee) has been shown to strengthen employees' emotional attachments to their employing organization (Rousseau & Aubé, 2010). From the employee perspective, supervisors are mainly the organization's representatives (Sluss et al., 2012) and are expected to play a key role in promoting the employees' positive feelings towards the organization. Other organizational resources, such as employee satisfaction with general conditions (e.g., employment security) and job autonomy, are tested in this study as independent variables that are assumed to be related to organizational pride. Job autonomy has received considerable attention in the organizational literature (Nielsen et al., 2017) and is defined as 'the degree to which the job provides substantial freedom, independence, and discretion to the individual in scheduling the work and determining the procedures to be used in carrying it out' (Hackman & Oldham, 1976, p. 395). In a recent meta-analysis study (Nielsen et al., 2017), job autonomy is conceptualized as an organizational level job resource in that it is directly related to job design. In other words, it is the organizational efforts that provide employees with autonomy. Autonomy is associated with more opportunities for employees to influence their environment and withdraw from undesirable circumstances (Naus, van Iterson, & Roe, 2007), although how and to what extent they do so is ultimately decided by the organization. To date, only Kraemer and Gouthier (2014) have examined this study's assertion that job autonomy positively relates to organizational pride.

Social identification as a potential mediator between job resources and group-based pride
In their study on work identification, Miscenko and Day (2016) concluded that the most acknowledged approach to social identification with organizations is Mael and Ashforth's (1992) approach, which defines organizational identification as 'the perception of oneness with or belongingness to an organization, where the individual defines him or herself in terms of the organization(s) of which he or she is a member' (Mael & Ashforth, 1992, p. 104). This definition takes a strictly cognitive position based on social identity theory and self-categorization theory. However, since its initial conceptualization, social identity theory has also included affective dimensions, such as the emotional significance and feelings of belongingness that an individual attaches to membership (Tajfel, 1972(Tajfel, , 1978Tajfel & Turner, 1979; see also Albert et al., 1998). Similarly, Ashforth (2001) suggests that one can think or feel one's way into identification, meaning that cognition and affect reinforce identification. Thus, the conceptualization and measurement of the identification construct in organizations is a topic that is actively discussed by scholars. This study views cognitive (self-categorization, i.e., placing oneself into a specific social group) and emotional (affective attachment, i.e., the individual's sense of the importance of a group for his or her overall sense of well-being) components as incorporated in the measurement of social identification. This conceptualization acknowledges that a cognitive dimension and an affective dimension can together form one social identification factor (see Ashforth, 2001;Dimmock, Grove, & Eklund, 2005;Stoner, Perrewé, & Hofhacker, 2011).
Organizational identification has traditionally been primarily regarded as a holistic construct that assumes that employees identify with their overall organizations. However, Ashforth and colleagues (e.g., Ashforth & Johnsson, 2001;Ashforth, Harrison, & Corley, 2008) have thoroughly discussed the salience of numerous social identification processes in organizational contexts. According to the identity matching principle (Ullrich et al., 2004), social identification that involves a particular focus or foci should correlate most strongly with outcomes that have a similar focus or foci (for example, workgroup resources, such as co-worker support, would be more strongly related to workgroup-based emotions as outcomes). In the present study, this identity matching principle is applied to the independent variables assumed to be related to social identification, which means that job resources (i.e., the independent variables) associated with a given level should relate to social identification and its outcomes at the same level (Ellemers & Rink, 2005).
It has been argued that the foci-specific antecedent outcome process is mediated by social identification. For example, Lavelle, Rupp and Brockner (2007) argue that justice/trust emerging from three specific foci (the organizational, the supervisory and the co-worker) is related to various identification processes and that citizenship behavior is consequently related to these particular aspects. Although their data are cross-sectional, social identification is shown to potentially mediate the relationship between perceived organizational support and organizational involvement within a dual organizational identity context (Edwards & Peccei, 2010; see also, Restubog et al., 2008;Vora & Kostava, 2007). According to social identity theory (Tajfel, 1978;Tajfel, & Turner, 1979) and intergroup emotions theory (Mackie, Devos, & Smith, 2000;Smith, Seger, & Mackie, 2007), the strength of an individual's social identification has a pronounced and central effect on group-based emotions such as pride (Arnett, Laverie, & McLane, 2002). Given the fact that group-based emotions depend on an individual's group identification levels, in the sense that those who more strongly identify with the group report more positive and less negative group-based emotions (Smith, Seger, & Mackie, 2007), these studies suggest that social identification can be viewed as a mediator between different types of antecedents and positive employee outcomes. Social identification with a group can also lead to emotional reactions in response to situations and events that can influence in-group membership even when individuals are not directly or personally involved in a given situation (Mackie, Devos, & Smith, 2000). Thus, social identification can boost or play down the impact of job resources on group-based emotions such as group-based pride. This process can differ depending on the focus of the identification in organizations. However, the literature on the effects of specific job resources at different organizational levels is scarce (i.e., workgroup and organization), especially with regard to foci-specific group-based emotions and the intervening role of foci-specific social identification.
Lastly, it should be pointed out that some aspects of group-based pride are related to social identification such as the employees' emotional attachments to and involvement in organizational groups. However, the major distinction between social identification and group-based pride is that social identification comprises both cognitive and emotional elements related to belonging (Dimmock, Grove, & Eklund, 2005;Stoner, Perrewé, & Hofhacker, 2011), whereas group-based pride is an emotion that is primarily evoked by the shared experiences and interactions among group members (Arnett, Laverie & McLane, 2002;Smith, Seger & Mackie, 2007).

Aim
The aim of this study is to investigate the relationship between job resources at two organizational levels and their relationships to group-based pride as an emotional reaction related to the workgroup and the organization respectively. Target-specific social identification is assumed to play a part in these relationships as an intervening factor that can help to explain the variation in pride. Here, both direct and potential mediating effects are investigated.

Hypotheses
Although some have argued that social identification has both cognitive and affective dimensions, it has also been suggested that group-based pride, while related to social identification, can be distinguished from it. The perception of belongingness is argued to have positive emotional components, whereas group-based pride is said to be the consequence of interaction and accomplishment. Following the identity matching principle (Ullrich et al., 2004), this study suggests that the processes function separately at the two levels. The study's first two hypotheses are as follows: H1: Workgroup identification and workgroup pride are distinct but correlated constructs. H2: Organizational identification and organizational pride are distinct but correlated constructs.
Furthermore, it is suggested that the processes that lead to group-based pride at these two levels have distinct job resources related to those levels. Again, following the identity matching principle (Ullrich et al., 2004), the study suggests that resources associated with the workgroup level (co-worker support) positively relate to workgroup pride, and that resources associated with the organizational level (supervisor support, satisfaction with work conditions and job autonomy) positively relate to organizational pride. With this in mind, we propose the following two hypotheses: H3: Perceived co-worker support can explain additional variance in workgroup-based pride over and above the effects of the organizational resources. H4: Perceived supervisor support, job autonomy and satisfaction with work conditions can explain additional variance in organizational pride over and above the effects of the workgroup job resource.
Finally, in order to examine a foci-specific job resources identification outcome model as recommended by previous studies (Edwards & Peccei, 2010;Vora & Kostava, 2007), social identification at both levels is suggested to potentially and partially mediate the relationship between the foci-specific job resource and foci-specific pride (i.e., the direct and indirect effects are expected to be statistically significant at both levels). Following this, we propose the following two hypotheses: H5: Workgroup identification partially mediates the relationship between perceived co-worker support and workgroup pride. H6: Organizational identification partially mediates the relationship between perceived supervisor support, job autonomy, satisfaction with work conditions and organizational pride.

Sample
Data was obtained from public sector employees working in two middle-sized municipalities in Sweden, each of which employs some 10,000 people. Sweden's 290 municipalities are responsible for a larger share of public services than those in many other countries and have decision-making powers (SALAR, 2016). Three-quarters of Sweden's municipal activities are directly related to demographic characteristics and are determined by the number of residents, age and health status. Although the municipalities are in the public sector, they are also major employers, are subject to national labor laws and are responsible for the well-being of their staff. One in five employees in Sweden, just over 800,000 individuals in 2016, are employed by a municipality. A majority of females employed by municipalities work in the health care and education sectors (SALAR, 2016).
In order to obtain a sample for the study, 42 supervisors from different public sector divisions in the two municipalities were contacted via e-mail. Twenty-two of the supervisors responded to say that their employees would be allowed to participate (the supervisor response rate was 52%). Altogether, 450 municipal employees completed and submitted the questionnaire. Questionnaires with missing data for more than one relevant item were omitted from the analysis (14 cases), yielding in a final sample of 436 respondents. The average age of the respondents was 46.10 years old (SD = 10.92, range = 22-66 years) and 72% were female. Approximately 34% worked in the educational sector, 27% in the health care sector, 25% in the social services sector, 8% in administration, and 4% in the technical sector. The remaining 2% of the sample did not fit into any of these sectors, although they were included in the analysis. The size of the sectors in the sample reflected the organizations total proportion of the sectors. The majority of individuals in the sample (90%) were full-time employees.

Procedure
Because no current analogous measures of workgroup pride and organizational pride existed at the time of the study, our first step was to develop scales to measure both constructs. The preliminary instrument development began with a qualitative phase and included four explorative focus group interviews (two groups from each participating organization with six employees in each focus group), as recommended by Nassar-McMillan et al. (2010) and Vogt, King and King (2004). The interviews lasted approximately 45 minutes and included open questions about the employees' perceptions of pride at work. These questions were designed to ensure that the group discussion guided the development of the instrument, rather than vice versa (see Rowan & Wulff, 2007). All the interviews were transcribed, and the interview content was used to form categories. The thematic vertical analysis was complemented with a narrative analysis of each focus group interview. To further increase validity, two (or occasionally three) people did all the categorizations, both independently and collaboratively. This way of establishing validity is frequently recommended in qualitative research and is described by Kvale (2008) as dialogical intersubjectivity. The focus group interview analysis resulted in three separate categorizations: personal pride, workgroup pride and organizational pride. Only items aimed at measuring workgroup pride and organizational pride were constructed. As recommended by Lynn (1986), all the items were then sent to content-validity researchers and members of the target population, who were asked to individually evaluate the items for clarity and relevance to the target constructs. This resulted in eight items (four workgroup pride items and four organizational pride items), which then were included in the final questionnaire.
The final questionnaire was administered in person during team meetings. Before the questionnaires were administered, information about the study and the invitations to participate were sent to team supervisors via e-mail. The e-mail content invited teams to participate by completing the questionnaires during regular working hours. The completed questionnaires were collected by the researchers, who then processed and administered the survey. In most cases, the respondents completed the questionnaires at their regular team meetings after having been informed orally and in writing (on the questionnaire's cover sheet) that their participation in the survey was voluntary and that their responses would remain confidential.

Materials
The self-report questionnaire consisted of 112 items, all of which aimed to measure the job resources at the workgroup and organizational levels, intervening social identification variables and group-based pride. Many of the items had been used in previous studies and were adapted for this study and translated into Swedish. 1 Gender, age, the number of years in the organization (M = 14.84, SD = 10.80) and the number of years in the workgroup (M = 5.82, SD = 6.25) were included as control variables. For all the items, the respondents were asked to rate their extent of agreement on a scale from 1(=strongly disagree) to 5(=strongly agree).

Independent variables
Perceived co-worker support was measured using a mean value of responses to five items from the Swedish Job Demands, Social Support, Control and Competence Scale (Hovmark & Thomsson, 1995). Two sample items were: 'I get sufficient information from my co-workers in order to do my work' and 'My co-workers usually help me if I run into problems at work.' Cronbach's alpha was 0.86.
Perceived supervisor support was measured by a mean value of the responses to six items from the abovementioned scale (Hovmark & Thomsson, 1995). Two sample items were: 'When struggling with practical difficulties at work, I rely on my immediate supervisor to help me' and 'I get sufficient information from my immediate supervisor in order to do my work efficiently.' Cronbach's alpha was 0.86.
Satisfaction with work conditions was measured by a mean value of the responses to four slightly modified items from the General Nordic questionnaire for psychological and social factors at work (QPS Nordic; Lindström et al., 2000). Two sample items were: 'I am satisfied with my employment security.' and 'I am satisfied with my prerequisites to work professionally.' Cronbach's alpha was 0.74. Job autonomy was measured by a mean value of responses to three items from the QPS Nordic (Lindström et al., 2000). Two sample items were: 'I have the freedom to decide what to do on my job' and 'It is my own responsibility to decide how my job gets done.' Cronbach's alpha was 0.84.

Intervening variables
Workgroup identification was measured using a scale based on three items from the Multi Identification Scale modified to address the workgroup foci (MDIS; Stoner, Perrewé, & Hofhacker, 2011). The three items were: 'If asked if I belong to this workgroup, I would say 'Yes'', 'I consider myself a member of this workgroup' and 'When something bad happens to this workgroup, I personally feel hurt.' Cronbach's alpha was 0.78.
Following Brislin (1986), before the questionnaire was administrated, a bilingual expert translated these items from English into Swedish and another bilingual expert back-translated them into English. Two experts then compared the original English versions to the back-translated versions to ensure that the Swedish version was accurate.
Organizational identification was measured using a threeitem scale adapted from the MDIS modified to address the organizational foci (Stoner, Perrewé, & Hofhacker, 2011). The three items were: 'If asked if I belong to this organization, I would say 'Yes'', 'I consider myself a member of this organization' and 'When something bad happens to this organization, I personally feel hurt.' Cronbach's alpha was 0.75. Items were translated from English to Swedish following the procedure described above regarding workgroup identification.

Dependent variables
Workgroup pride was measured by four items developed according to the procedure described above. The workgroup pride items were: 'I feel proud of my workgroup even if it is criticized by an outsider', 'When my workgroup is successful I feel pride', 'I feel proud when someone externally praises my workgroup' and 'I feel proud when telling others about my workgroup.' Cronbach's alpha was 0.87.
Organizational pride was measured by four items developed according to the procedure described above. These items were: 'I feel proud when someone externally praises my organization', 'I feel proud when telling others about my organization', 'I feel proud when someone externally praises my organization' and 'I feel proud when telling others about my organization.' Cronbach's alpha was 0.90.

Statistical analyses
Preliminary analyses were performed to ensure that there were no violations of the assumptions of normality, linearity, multicollinearity or homoscedasticity. The factor structure of the variables was then investigated using confirmatory factor analysis (CFA) in Amos 21 (Arbuckle, 2011). Chi-Square (χ 2 ) was used to compare the models. As χ 2 is sensitive to sample size, the assessment also considered other indices of fit (Byrne, 2010), such as the Root Mean Squared Error of Approximation (RMSEA), Common Fit Index (CFI), Adjusted Goodness-of-Fit Index (AGFI), and Goodness-of-Fit Index (GFI). Hu and Bentler (1999) have suggested that RMSEA < 0.06 is a cut-off criteria indicative of very good fit. However, given the exploratory nature of this study, a less conservative criterion, RMSEA < 0.08, was deemed necessary for indicating moderate levels of model fit (Berntson et al., 2016;Browne & Cudeck, 1993;MacCallum, Browne & Sugawara, 1996).
As the data was collected using a single instrument (one questionnaire), a common method bias (CMB) test was performed to determine whether this influenced the results of the measurement model. A Harman's single factor test revealed that one single factor explained 27% of the variance, indicating that CMB was not a serious concern (because it is clearly <50%; see Podsakoff et al., 2003).
Hayes' (2012) conditional process model using the PROCESS custom dialog box tools added to SPSS (version 22) was computed with bootstrapping to test the mediation hypotheses. PROCESS is a computational procedure that inter alia implements mediation analysis. Bootstrapping involves the drawing of a large number of sub-samples from a main sample and performing the analyses on all of the sub-samples. In this case, the number of sub-samples was set to N = 5000, as suggested by Hayes (2009). To interpret the results, the statistical significance of a path coefficient or an indirect path is indicated by the bootstrap confidence interval. The null hypothesis is rejected when the confidence interval excludes zero. Statistical significance of an indirect path is then assessed using the critical ratio test (the ratio of the path estimate over its standard error), where a CR > 1.96 denotes a significant path estimate at the 5% α level. Bootstrapping has been preferred over Sobel's product of coefficients test (Sobel, 1982) and the traditional causal steps approach (Baron & Kenny, 1986) because it is superior in power and has the ability to control Type 1 errors (Shrout & Bolger, 2002). Table 1 presents the descriptive statistics, including correlations and the internal consistency coefficients (Cronbach's alphas) of the scales.

Results
In order to ensure that the measures relating to social identification and group-based pride at both levels assessed distinct but correlated constructs (see hypotheses 1 and 2), a series of confirmatory factor analyses (CFAs) were performed. Several models were compared in which the numbers of latent factors varied. In the first model, all of the items were loaded onto one general latent factor. The second model consisted of all four constructs as independent factors. In model three, workgroup identification and workgroup pride were combined, as were organizational identification and organizational pride. The fourth and last model was identical to the second, the only differences being that workgroup identification and workgroup pride were separate yet correlated constructs and that organizational identification and organizational pride were also separate and correlated constructs. As Table 2 shows, model 4 outperformed the other models indicating that the four constructs were independent, although identification and pride remained correlated at each level. Thus, the results of the CFAs support hypotheses 1 and 2.

Workgroup resources and pride
In order to test hypothesis 3 (whether the workgroup level resource perceived co-worker support explains additional variance in workgroup pride over and above the effects of resources at the organizational level), a hierarchical multiple regression analysis was performed with workgroup  pride as the dependent variable (see Table 3). The results in the first step showed that only gender, job autonomy and perceived supervisor support significantly predicted workgroup pride. In the second hierarchical step, when perceived co-worker support was added, the beta values of job autonomy and perceived supervisor support declined, but remained significant. The second analytical step explained incremental variance in workgroup pride, and thus, hypothesis 3 was supported. In the third and final step when workgroup identification was added to the model, the significant beta values declined, indicating that workgroup identification could function as a potential partial mediator for the workgroup foci.

Organizational resources and pride
A hierarchical multiple regression analysis was performed with organizational pride as the dependent variable to test whether organizational resources (satisfaction with work conditions, job autonomy and perceived supervisor support) and organizational identification added to the amount of explained variance (see Table 4). In the first hierarchical step, only gender and age significantly predicted organizational pride. Gender remained significant in all three steps, indicating that women experienced more organizational pride than men. The second hierarchical step added the organizational resources (i.e., job autonomy, perceived supervisor support and satisfaction with work conditions). This increased the total amount of explained variance in organizational pride, and thus supported hypothesis 4. The organizational resources results indicated that satisfaction with work conditions such as job security and pay appeared as the strongest predictor. Including organizational identification in the third and final step revealed a pattern similar to that of the workgroup foci, and significantly explained additional variance in organizational pride and decreased the significant beta values of the preceding steps. Using Cohen's effect size index (Cohen, 1988), all of the coefficients in the two regression analyses had medium-sized effects (f2 < 0.50). The results of the final step indicated that organizational identification could potentially mediate the effects of the organizational foci.

The potential mediating role of workgroup and organizational identification
To test the mediation hypotheses, Hayes' (2012Hayes' ( , 2013 conditional process was performed in combination with bootstrapping. In this approach, mediation is tested by assessing the effect sizes of the indirect effects and their bias-corrected bootstrap confidence intervals based on 5000 bootstrap samples. When a confidence interval does not contain zero, mediation is present (Hayes, 2012(Hayes, , 2013. The results from a simple mediation analysis conducted for each resource variable separately using ordinary least squares path analysis are presented below. Figure 1 shows that perceived co-worker support indirectly influenced workgroup pride through its effect on workgroup identification. As can be seen in Figure 1, perceived co-worker support was positively related to workgroup identification (b = .64), and workgroup identification was positively related to workgroup pride (b = .37). A bias-corrected bootstrap confidence interval for the indirect effect (b = .24) was entirely above zero (0.184 to 0.293). In addition, there was a significant direct effect on the relationship between perceived co-worker support and workgroup pride (b = .27). This indicated that the job resource related to the workgroup has a positive association with feelings of pride in the workgroup and that these feelings are further strengthened by workgroup identification.
The following three figures present the job resources related to organizational pride. In Figure 2, perceived supervisor support is positively related to organizational Table 3: Standardized regression coefficients (β), R 2 Change (∆R 2 ) and adjusted R 2 regarding study variables predicting workgroup pride.
identification (b = .26), and organizational identification is positively related to organizational pride (b = .56). A bias-corrected bootstrap confidence interval for the indirect effect (b = .14) is entirely above zero (0.087 to 0.213) and there is a significant direct effect on the relationship between perceived supervisor support and organizational pride (b = .21). Figure 3 demonstrates that job autonomy is related to organizational identification (b = .19) and that organizational identification is strongly related to organizational pride (b = .57). The bias-corrected bootstrap confidence interval for the indirect effect (b = .11) is entirely above zero (0.047 to 0.174) and there is a significant direct effect on the relationship between job autonomy and organizational pride (b = .26).
Finally, Figure 4 demonstrates that satisfaction with work conditions is positively related to organizational identification (b = .25), and that organizational Table 4: Standardized regression coefficients (β), R 2 Change (∆R 2 ) and adjusted R 2 regarding study variables predicting organizational pride.

Model 1 2 3
Step 1 Control factors Gender -.12** -.20* -.14** Age . Note. * = p < .05, ** = p < .01, *** = p < .001. Predictors: gender (women = 1), age, number of years employed by the organization, number of years in the workgroup, perceived co-worker support, workgroup identification, satisfaction with work conditions, job autonomy, perceived supervisor support, organizational identification.  identification is related to organizational pride (b = .53). The bias-corrected bootstrap confidence interval for the indirect effect (b = .13) is entirely above zero (0.081 to 0.200) and there is a significant direct effect on the relationship between satisfaction with work conditions and organizational pride (b = .33). The three models above indicate that the job resources related to the organization are positively related to feelings of pride in the organization and that these feelings are further strengthened by organizational identification.

Discussion
The main aim of this study was to investigate the relationship between job resources at two organizational levels and their relationships to group-based pride as an emotional reaction related to the workgroup and the organization, respectively. It was argued that social identification played a part in this process as an intervening factor that contributes to the explanation of variation in pride.
Workgroup identification and workgroup pride were found to be two separate yet correlated constructs (hypothesis 1), as were organizational identification and organizational pride (hypothesis 2). Thus, hypotheses 1 and 2 were supported. Following the identity matching principle (Ullrich et al., 2004), hypothesis 3 and 4 addressed whether the job resource(s) associated with a particular level would relate to group-based pride at the corresponding level. At the workgroup level, the job resource perceived co-worker support explained additional variance in workgroup pride (beyond the contributions of resources at the organizational level), confirming hypothesis 3. Similarly, hypothesis 4 was confirmed: At the organizational level, the job resources perceived supervisor support, job autonomy and satisfaction with work conditions explained additional variance in organizational pride (beyond contributions of the workgroup resource perceived co-worker support). These results support the notion of separate processes at the two levels, thereby indicating that resources were related to different foci and in turn to feelings of groupbased pride on the same foci. The study's hypotheses 5 and 6 addressed whether social identification at both levels would function as potential partial mediators between the foci-specific job resource(s) and foci-specific pride. Both of these hypotheses were supported and are in line with previous research, suggesting that when social identification is salient, group concerns are also salient and result in group-based emotions associated with that identification (Iyer & Leach, 2008;Yzerbyt et al., 2003). This paper makes the following contributions to existing literature on pride in organizations. First, the fact that different levels of resources function by linking independent variables and dependent variables at each level offers a theoretical contribution to the JD-R model. A further contribution to the JD-R model is that group-based emotions were tested as possible dependent variables of the process. Second, the phenomena of positive group-based emotions in an organizational setting, such as groupbased pride, is a relatively unexplored yet fertile territory for expanding understanding about social processes at work. This study's findings verify that employees' perceptions of their job resources at the workgroup level are more strongly related to workgroup pride than to organizational pride, and that the job resources at the organizational level are more strongly related to organizational pride than to workgroup pride. Clarifying these processes  may help organizations to decide where best to focus their intervention efforts (see Nielsen et al., 2017). This study also suggests at least two possible additions to social identity theory as it applies to work settings. (1) Although organizational identification has mostly been studied as a unidimensional concept (Miscenko & Day, 2016), this study supports the notion that employees identify with at least two organizational foci (workgroup and organization), which is in line with findings on multiple foci of identification (Ashforth & Johnson, 2001;George & Chattopadhyay, 2005). (2) The study also shows that employees seem to identify less strongly with their organizations than with their workgroups, which supports previous findings that proximal targets are likely to achieve relatively more identification (Van Knippenberg & van Schie, 2000). This supports the idea that a sense of in-group membership may be more easily triggered with respect to the supporting interaction within the immediate workgroup than with the more emotionally distant organization (Ashforth & Rogers 2012;Riketta, 2005;Van Knippenberg & van Shie, 2000).
Although several studies have argued in favor of a foci-specific antecedent identification outcome process, they have mainly addressed behavioral or attitudinal outcomes, such as organizational involvement (Edwards & Peccei, 2010) or organizational citizenship behavior (Lavelle, Rupp, & Brockner, 2007). Research on emotional outcomes at the group level in this three-step process is scarce, although intergroup emotions theory has addressed the latter part of this process by focusing on the direct link between social identification and group-based emotions (Mackie, Smith, & Ray, 2008). Viewing social identification as a potential mediating factor between the relationships of job resources and affective dependent variables at the group level, as done in this study, increases the possibilities for theory development and model analysis. For example, it was found that when foci-specific job resources are mediated by social identification with a particular foci, they not only influence behavior or attitudes (as previous studies have indicated [Ashforth, Harrison, & Corley, 2008]), but they also influence group-based emotions, such as group-based pride in the same foci. This indicates that different types of job resources seem to influence how and which organizational foci employees socially identify with, and consequently, the foci of their group-based pride as well.
Lastly, a discussion about the conceptual difference between organizational pride and affective organizational commitment is warranted. Even though affective commitment mainly refers to a general sense of emotional attachment to the organization (Meyer & Allen, 1991), it is suggested that organizational pride is a more specific potential emotional consequence of interaction and actual accomplishment and not solely the broad emotional state that affective organizational commitment is conceptualized as. Group-based emotions, such as organizational pride, are also socially shared within important in-groups (Smith, Seger, & Mackie, 2007), a phenomenon that has not been highlighted in the research on affective organizational commitment. Thus, an organizational accomplishment that stimulates shared organizational pride might promote even greater interaction, cohesiveness and influence in a way that makes interpretation of a future event even more likely to result in organizational pride (see Smith, Seger, & Mackie, 2007). However, the suggested distinction between organizational affective commitment and organizational pride is a subject that merits further critical considerations.

Limitations
This study is limited by its cross-sectional design, which means that inferences about causal effects cannot be drawn from the found relationships. Using multi-phase longitudinal data would allow for a clearer understanding of the phenomena. Also, longitudinal multi-phase mediation models would permit the examination of several mediation questions that cannot be asked using the cross-sectional mediation model, such as whether a potential mediated effect is stable over time. Our design cannot help us in sorting out which causal sequences are plausible and which are not, as the data were collected at a single point in time. Workgroup and organizational identification could, thus, be confounding variables and not mediators, even though theory seem to reinforce the legitimacy of our assumptions of the variables' potential functioning. Multi-phase longitudinal models would therefore shed light on temporal-precedence or causal-ordering assumptions by quantifying mediation relations among variables over time. Although commonly used in research (see Rucker et al., 2011 for an overview), the simple crosssectional mediation modeling presented above may oversimplify the complex dynamics of how job resources can influence the two types of group-based pride. In line with Taris and Kompier (2006), simple cross-sectional mediation analyses cannot unambiguously separate the relationships among the concepts in question, at least not in the absence of more information about the correct causal order. Second, although the size of the sectors in the sample reflected the proportions of the organizations' sectors, this study's sample was relatively small and therefore should not be generalized to larger municipal organizations in Sweden. This is because larger Swedish municipal organizations may differ in terms of their complexity and internally diversified structures compared to the studied organizations (Larsson, 2015). Third, following Bakker and Demerouti (2007), the study has been restricted to a given and limited set of predictor variables that may not be relevant for all job types. This means that certain occupations may require other combinations of resources than those used in this study to predict workgroup pride or organizational pride more accurately. Fourth, because the scales measuring workgroup pride and organizational pride were specifically developed for this study, they would need to be tested on other populations in order to establish convergent, discriminant and, ultimately, construct validity. Even though focus groups were used to enhance content validity, the content validity of a particular assessment device is always conditional (Haynes, Richard, & Kubany, 1995). Further, because the model indices in the CFA only showed moderate levels of fit, testing the scale in other populations would also allow for a better understanding of how model indices could be improved.

Practical Implications
This study's findings have a number of practical implications for management practices, in that they refine the understanding of the specific influences of job resources on two organizational identification foci and their potential group-based emotional consequences. Social identification is based on shared values and goals, which means that employees are likely to behave in ways that support the particular organizational foci that are most salient to them. Despite this, in many organizations the role of relational processes is downplayed and emphasis is put instead on superordinate goals of budget or performance management. Nevertheless, the social processes and the groupbased emotions they evoke (such as pride), most probably have direct consequences for attitudinal and behavioral outcomes (e.g., Hornsey & Hogg, 2000;Tyler & Blader, 2001). For example, the management policies in the public sector in which administrative units are held accountable and responsible for the results, not only regarding their costs but also their incomes, make it increasingly difficult for employees to identify with (Ashforth & Johnson, 2001;Höjer & Forkby, 2011) and feel pride for the overall organization. This may fuel workgroup identification. Although it may improve the workgroup climate, this strong lowerorder identification could be detrimental to the overall organization if it creates invisible boundaries, territorial thinking or increased internal competition by impairing cooperation between different groups of employees (Ashforth & Johnson, 2001). Another downside of strong workgroup identification is that it may be too excessively exclusive and could threaten the employees' desires for a general sense of belonging at higher organizational levels (Brewer & Gardner, 1996). Thus, organizations could benefit by counterbalancing strong lower-order workgroup identification with enhanced higher-order organizational identification. For example, as found in this study, organizational identification was related to organizational pride, which could encourage employees to fulfill organizational goals and to cooperate with other members of the organization to realize organizational values and norms, thus surpassing their lower-order workgroup goals and values. When provided with opportunities to engage with and become attached to their organizations at a social and emotional level through job resources such as supervisor support, job autonomy and fair working conditions, employees are more likely to establish cognitive and emotional bonds with their organizations and experience positive group-based emotions like organizational pride. However, the downside of employees investing emotionally in their organization is that they may be less willing to change employers, which could cause organizations to become less dynamic over time.

Suggested Future Research
The finding that employees may have multiple social identities in the workplace suggests that future studies should longitudinally examine other group-specific job resources as antecedents of social identification in organizational contexts and emotional outcomes linked to these possible multiple identities. Moreover, when investigating social identification in organizations, a more fruitful approach would be to consider social identification as a potential mediating factor rather than an outcome, as recent studies have done (Edwards & Peccei, 2010;Lavelle, Rupp, & Brockner, 2007). Finally, future research that measures identification with a variety of organizational foci and corresponding group-based emotions over time would allow for the examination of the temporal changes in both group identifications and group-based emotions. Using multi-phase longitudinal analyses would help researchers to elaborate if the particular mediation models presented in this study fit the data better than competing models. This would strengthen the model's ability to draw causal conclusions. Positive group-based emotions also warrant further investigation, particularly regarding how they connect with other positive outcomes for the organization.

Additional File
The additional file for this article can be found as follows: • Appendix. Item loadings for the group-based pride variables in the study. DOI: https://doi. org/10.16993/sjwop.23.s1 Notes 1 Although the survey was in Swedish, we have included the original English-version items in this paper.