Workplace Social Capital: Redefining and Measuring the Construct

We conducted a search on MEDLINE (via Pub Med), Scopus and Google Scholar using the following keywords: ''social capital'' or ''workplace social capital'' separately or with ''definition'' or ''origins'' or ''theory'' or ''forms'' or ''dimensions'' or ''measurement''. We also included articles on SC because research on WSC is rather limited, while the two constructs are intertwined. We identified over 700 articles and book chapters. Of these, we selected 74 manuscripts based on the following criteria: 1) they concerned SC or WSC and not similar constructs (e.g., social network, social support); 2) their aim was to define (W)SC and included specific descriptions of its content and dimensions; and 3) were written English. The first and second author evaluated the inclusion of the first 50 papers independently. Results indicated an agreement of 93% that was resolved after discussion. Due to high consistency, all other manuscripts were evaluated by the first author.

The literature review resulted in 26 definitions (see APPENDIX). In most cases, (W)SC was described as including both network ties and shared values, which implies that the construct must incorporate both aspects. To synthesize a validated definition of WSC, we investigated the structure and dimensionality of the construct as depicted in the different definitions. To this end, in Table 1 we present the different forms of WSC, while in Table 2 we present the unique dimensions of WSC. Identifying the different forms is important because it allows determining the levels of measurement of the construct (i.e., the proximity and the direction of social network aspects). Identifying the different dimensions is relevant for determining the unique facets of WSC.

As shown in Table 1, the construct is mainly distinguished into: (i) bonding, bridging and linking, based on the proximity of linkages between different actors (Harpham et al., 2002; Kreuter & Lezin, 2002; Szreter & Woolcock, 2004), and into (ii) structural (e.g., network size) and cognitive (e.g., trust, reciprocity and other intangible aspects), based on its components (Grootaert & Bastelaer, 2002; Harpham et al., 2002; Uphoff & Wijayaratna, 2000). Other theoretical approaches presented in Table 1 are less common and not well-established. Specifically, in certain cases (e.g., positive/negative, instrumental/principled) the forms of SC are mixed up with its outcomes (Graeff & Svendsen, 2013), in other cases (e.g., instrumental/principled) validity is not empirically supported, while strong overlap is found between other forms (e.g., horizontal/vertical, formal/informal; Ferlander, 2007).

According to the bonding/bridging/linking distinction (see Table 1), bonding WSC refers to close ties between individuals within homogeneous groups (e.g., colleagues), bridging describes the network connections amongst individuals from different groups (e.g., colleagues from different hierarchy levels or members of the organization with other social groups or organizations), and linking refers to relationships between individuals and groups interacting across formal and institutionalized authority gradients in society (e.g., governmental organizations; Kreuter & Lezin, 2002). The conceptual similarity between bridging and linking (Adler & Kwon, 2002) explains why only some theoretical approaches incorporate linking. Importantly, most approaches distinguish only between bonding and bridging, even when they measure ties between individuals and governmental or official authorities (Chen et al., 2009; De Silva et al., 2006; Harpham et al., 2002). As concerns the structural vs. cognitive distinction (Table 1), structural WSC refers to tangible and externally observed social constructions that describe network ties and linkages that give access to resources (e.g., colleagues, professional or cultural groups; Harpham et al., 2002). Cognitive WSC refers to intangible aspects like values, attitudes and beliefs (e.g., trust, tolerance and reciprocity; Grootaert & Bastelaer, 2002; Harpham et al., 2002; Nyqvist et al., 2014). Relational WSC (Nahapiet & Ghosal, 1998) is not widely used, since it largely overlaps with cognitive and structural SC (Murayama et al., 2012).

Since the distinction between bonding and bridging is the most prominent in the literature (Table1), we determined the levels of measurement (i.e., horizontal vs. vertical) of WSC on the basis of this distinction. Accordingly, bonding WSC refers to values, beliefs, attitudes, relations and procedures that take place among employees within the same level of hierarchy (i.e., colleagues), while bridging WSC describes values, beliefs, attitudes, relations and procedures that take place among employees across different hierarchy levels (i.e., employees and management) or that extend outside of the organization (Putnam, 1995). In general, bonding SC refers to procedures and social ties of people who are alike, with inward direction, that is good for ''getting by'', whereas bridging SC refers to social ties and procedures of different social groups (or individuals that belong to them), with outward direction, that is good for ''getting ahead'' (Adler & Kwon, 2002).

Next, we determined the dimensions of the bonding and bridging forms of WSC based on the cognitive vs. structural distinction (e.g., Grootaert & Bastelaer, 2002). Table 2 suggests that the three core dimensions that appear in most approaches are social networks, trust and reciprocity. Looking at the operationalization of these dimensions (Table 2) vis à vis the description of cognitive and structural forms (Table 1), it becomes evident that social networks, as observed social constructions, refer to structural WSC, while intangible aspects such as trust and reciprocity refer to cognitive WSC. Thus, we argue that bonding and bridging WSC consist of both structural (i.e., social networks) and cognitive (i.e., trust and reciprocity) dimensions.

Specifically, structural WSC refers to the qualitative aspects of networking and can be distinguished on the basis of the direction of social action (i.e., inward/outward of the organization; Harpham et al., 2002). Hence, the network ties, the frequency of interaction and the quality of social relationships among colleagues capture structural WSC at the bonding level. Rather, structural WSC at the bridging level captures the procedures and possibilities to expand the professional network outside of the organization and not just among colleagues at different hierarchical levels. This is because the networking procedure at the bringing level primarily concerns an outward direction of social action that is mainly facilitated by the management and aims to benefit from the collaboration with other organizations (Balkundi & Kilduff, 2006).

Cognitive WSC refers to intangible components like norms, values and attitudes. Trust and reciprocity are recognized as core components of WSC in extant research (e.g., Clausen et al., 2019; Firouzbakht et al., 2018; Meng et al., 2018), constitute the most prominent dimensions of cognitive (W)SC (e.g., Kreuter & Lezin, 2002; Narayan & Cassidy, 2001), and are dimensions that may characterize all workplaces, regardless of organizational size, type of work or employees’ social characteristics. Thus, we considered these two as the core dimensions of cognitive WSC. Accordingly, trust and reciprocity among colleagues concern cognitive WSC at the bonding level, while trust and reciprocity among employees at different hierarchy levels concern cognitive WSC at the bridging level.

Based on the previous conceptual analysis, we define: WSC as a workplace resource that concerns employees’ perceptions regarding trust, reciprocity (cognitive WSC), and network interactions (structural WSC) that exist both among peers (bonding WSC) and among individuals across different hierarchical levels or organizations (bridging WSC). As shown on Fig. 1, bonding and bridging WSC refer to the levels and the direction of WSC that are assessed through cognitive (i.e., trust, reciprocity) and structural (i.e., network) WSC. Although both bonding/bridging and cognitive/structural distinctions constitute forms of WSC, the former refers to the levels of measurement and the direction of social interactions (i.e., bonding refers to social ties and procedures among peers with inward direction, while bridging to social ties and interactions among employees and management or with external organizations and includes procedures with outward direction), while the latter refers to the content of WSC that can be found at both levels of measurement and it is manifested via different dimensions (i.e., trust, reciprocity, and network).

We identified five scales that measure WSC as an ecological resource. The structure of these scales together with example items are presented in Table 3. Though relevant, these scales have certain limitations. First, although some core aspects of WSC, such as trust and reciprocity, are measured by most, these are not assessed both at the bonding and the bridging levels [e.g., the scale of Eguchi et al. (2017) captures only bonding WSC]. Second, some scales include dimensions (e.g., justice; Pejtersen et al., 2010) that are not conceived as core dimensions of WSC. Third, in some cases the emphasis is given unilaterally on common values (e.g., trust; Ommen et al., 2009; Pejtersen et al., 2010) and not on network issues that constitute a key ingredient of WSC. Fourth, most scales do not clearly distinguish between different dimensions (e.g., merge trust and reciprocity; Meng et al., 2018; Kouvonen et al., 2007; Ommen et al., 2009). Fifth, although Meng et al.’s (2018) scale captures all aspects of WSC, it can be applied only in complex, multi-team, organizations and not in small organizations or where a clear distinction between immediate and upper management is not evident. Also, the psychometric properties of most scales raise concerns. Although the tool developed by Kouvonen et al. (2007) is unidimensional, research concerning its validation has revealed a two- (Firouzbakht et al., 2018) or a three- (Middleton et al., 2018) factor structure. Similarly, several studies (e.g., Rosário et al., 2017) did not confirm the proposed three-factor structure of Pejtersen et al.’s scale (2010).

To address these limitations, we developed a measure that is grounded on our WSC definition, following a four-step procedure. First, we generated an item pool with items obtained from the scales of Kouvonen et al. (2007) and Pejtersen et al. (2010), because these capture both bonding and bridging WSC. Also, Kouvonen et al.’s scale is the most frequently used tool in WSC research. We excluded items assessing justice from Pejtersen et al.’s scale, since justice was not found to be a core dimension of WSC (Phase 1). Moreover, the scales by Eguchi et al. (2017) and Ommen et al. (2009) were excluded due to high overlap with the selected scales. Additionally, items from Meng et al.'s (2018) scale could not be incorporated since this scale was unpublished during the item development process. However, even if we could incorporate these items, we would not result in a substantially different scale because many items overlap conceptually with the selected items, some items capture aspects like conflict management and justice that are not key features of WSC, while other items measure WSC across teams of different departments within the organization, which are suitable only for assessing WSC in complex organizations. We conducted a literature review to allocate other items measuring trust, reciprocity and network in studies that examined WSC. Through this process two more items were selected: one from the scale of Oxenstierna et al. (2008) that evaluates respect between colleagues, and one from Requena's tool (2003) that evaluates social relationships and contacts between colleagues.

In the second step we evaluated the 21 selected items based on whether: (i) they were in line with our definition of WSC, and (ii) they could be applied to all organizations. Based on these criteria, 11 items were selected and adapted. Namely, the terms "employees" or "individuals" were replaced with "colleagues", in some items a clear reference to colleagues was added, questions were transformed to affirmative statements, and items of bridging WSC were formulated to refer to the immediate supervisor or management.

In the third step, we generated 24 new items to capture aspects of our definition that were not assessed by the selected items. To measure reciprocity of bonding WSC, we distinguished between instrumental and emotional support for work and non-work-related problems. Also, to measure trust and reciprocity of bridging WSC, we created bilateral items (e.g., from supervisor/management to employee and vice versa). Moreover, to evaluate the network dimension of bridging WSC, we developed items to assess opportunities that the organization provides to employees for expanding their network outside of the organization.

In the fourth step, all authors evaluated the items independently and proposed modifications to enhance clarity and face validity. The content validity of the final items was evaluated by all authors based on the definitional correspondence and definitional distinctiveness criteria in line with Colquitt et al. (2019). The final version included 37 items.

First, we performed exploratory factor analyses (EFA), to test the initial Structure of WoSCi. Namely, with these analyses we were interested in identifying the items that better capture WSC.

We assessed the adequacy of the model with Kaiser–Meyer–Olkin (KMO) and Bartlett's Test of Sphericity (ΚΜΟ = 0.95, χ² = 11,398.14, p < 0.001). The analysis resulted in five factors that explained 67.74% of the total variance. We excluded the fifth factor since it consisted of one item. Moreover, three items had loadings < 0.40 and two items loaded on two factors with loadings > 0.40, so these were also excluded.

Next, we conducted an EFA with the remaining 31 items. The EFA based on the 31-item solution resulted in the following four factors: all items measuring trust and reciprocity (i.e., cognitive WSC) at the bonding level loaded on the first factor (loadings ranged from 0.59 to 0.88, explained variance = 44.31%), all items measuring network (i.e., structural WSC) at the bonding level loaded on the second factor (loadings from 0.55 to 0.72, explained variance = 2.57%), all items measuring trust and reciprocity (i.e., cognitive WSC) at the bridging level loaded on the third factor (loadings from 0.53 to 0.94, explained variance = 11.06%), and all items measuring network (i.e., structural WSC) at the bridging level loaded on the fourth factor (loadings from 0.71 to 0.94, explained variance = 7.03%). These findings support a four-factor structure that distinguishes between structural and cognitive, as well as bonding and bridging WSC, but does not distinguish between the two dimensions (i.e., trust and reciprocity) of cognitive WSC as hypothesized (see Fig. 1).

To further reduce the length of the scale in order to address the issue of assessment burden, we excluded seven more items based on both content validity and statistical evidence (i.e., factor loadings, inter-correlations). Namely, we excluded four items due to high conceptual overlap with items with higher or similar factor loadings. Also, we excluded the only remaining reverse-worded item (see, Van Sonderen et al., 2013). Finally, we excluded two items that evaluated reciprocity less directly than the remaining items. Table 4 presents the EFA results of the final, 24-item version of the WoSCi. The EFA again resulted in four factors distinguishing between bonding cognitive WSC (merging trust and reciprocity at the bonding level), bonding structural, bridging cognitive (merging trust and reciprocity at the bridging level), and bridging structural WSC.

We investigated the factor structure of the final version of the WoSCi by performing confirmatory factor analyses (CFAs) and by accounting for the multilevel nature of the construct and the potential interdependency in the assessments of employees belonging to the same organization. CFAs are relevant because they allow testing whether the scale captures the construct in line with theory. Further, accounting for the multilevel nature of the construct is relevant because WSC, as an ecological resource, may capture both individual, as well as shared perceptions that may be aggregated at the organizational level of analysis. Given that the four-factor structure of the WoSCi that resulted from the EFAs was not aligned with the proposed theoretical framework of the construct (see Fig. 1), another aim was to test whether the hypothesized factor structure explains the data better than the factor structure that resulted from the EFA.

The results of the MLCFAs are presented in Table 5. The one-factor model (Model 1) did not show an adequate fit to the data. Models 2 and 3 had acceptable fit, although Model 3 showed a significantly better fit compared to Model 2 [Δχ² (3) = 105.69, p < 0.001]. These results suggest that the theoretically proposed, hierarchical structure (Fig. 1) better depicts the WSC construct than the four-factor structure resulted from the EFA. Importantly, the hypothesized higher-order factor structure of WSC was found to hold at the individual level of analysis after accounting for the nested structure of the data and the effect of the clustering variable (i.e., organization membership).

Since analyses supported the proposed structure of WSC, while accounting for the multilevel nature of the construct, we also investigated whether individual level data could be aggregated at the work-unit level of analysis. Thus, we estimated inter-rater agreement (IRA) and inter-rater reliability (IRR) for the total scale and for each separate forms and dimension of the WoSCi. Within-unit IRA was tested with the r_wg index and the average deviation (AD) index, while IRR (i.e., between-group variability) was tested with the intra class correlation coefficient (ICC). Values > 0.70 for r_wg, < 0.80 for AD and at least 0.12 for ICC justify aggregation (LeBreton & Senter, 2008). Results fully supported aggregation at the work-unit level of analysis (WoSCi: r_wg = 0.97; AD = 0.49; bonding WSC: r_wg = 0.95; AD = 0.48; bridging WSC: r_wg = 0.96; AD = 0.50; WoSCi sub-dimensions: 0.80 < r_wg < 0.95;0.44 < AD < 0.52). ICCs were acceptable for all items (0.21 < ICCs < 0.38).

To estimate the internal consistency of WoSCi, we estimated the Cronbach' s alpha and the Omega Index.

Convergent validity was tested by exploring the associations between WoSCi and other commonly used scales measuring WSC namely, the scales of Pejtersen et al. (2010) and Kouvonen et al. (2007). Hence:

Also, we tested the relationship of WoSCi with other, theoretically similar, yet distinct constructs. Although some scholars perceive justice as a dimension of WSC (Pejtersen et al., 2010), most studies showed that justice is a related, yet distinct construct (Eguchi et al., 2017; Kouvonen et al., 2007). Therefore, we expected:

Next, we assessed the relationships of WSC as measured with the WoSCi with different leadership styles, since leaders commonly determine organizational values. Transformational leaders can enhance key features of WSC such as trust and cooperation (Chen et al., 2016), and may motivate employees to build external networks (Brion et al., 2012). In contrast, abusive supervisors, who display hostile verbal and non-verbal behaviors toward their subordinates (Tepper, 2000) are likely to create an unsupportive environment (Shoss et al., 2013) and negative reciprocity beliefs. Hence:

Since WSC is conceptualized as a workplace resource, we expect, in the context of job demands-resources (JD-R) theory (Bakker & Demerouti, 2017), that it correlates positively with other relevant job resources (i.e., participative decision making, performance feedback, job autonomy, and colleagues' social support). Participative decision making may strengthen and improve the quality of social relationships in the workplace, especially between supervisors and subordinates (Somech, 2010). Performance feedback implies open and direct communication between different parts (Krasman, 2012), which may promote or be promoted by aspects like trust and positive interactions at work. Job autonomy was also found to associate positively with WSC (Kouvonen et al., 2007). An organization high in WSC may allocate more autonomy to employees, due to the experienced trust. Similarly, by feeling responsible for their jobs, employees with greater job autonomy are more likely to create a mutual trustful and cooperative work involvement (Wang & Cheng, 2010). Finally, social support is closely related to the reciprocity dimension of WSC. Meanwhile, trust is an important condition that facilitates the development of supportive relationships, while network dimension reflects the quality of social ties that may guarantee social support. Therefore:

We tested criterion-related validity by exploring the relationships of WSC with work engagement, burnout, and job performance. According to JD-R theory (Bakker & Demerouti, 2017), job resources are the main drivers of employee motivation since they facilitate goal attainment while promoting personal growth and development. WSC may help employees be functional in achieving work goals making them more vigorous, dedicated and absorbed (i.e., engaged) in their work (Clausen et al., 2019). Similarly, employees working in an organization high in WSC, are more likely to feel a sense of obligation that may translate in increased in-role job performance (Eisenberger et al., 2001), and to invest in positive citizenship behaviors, such as extra-role performance (Sparrowe et al., 2001). In line with JD-R theory, job resources are also the main drivers of the disengagement component of burnout since high levels of resources protect employees from becoming cynical about their work. Also, job resources help employees deal with high job demands thus, mitigating their unfavorable impact of the exhaustion component of burnout. In this context, we argue that WSC will relate negatively with burnout (Kowalski et al., 2010). Thus:

We tested the incremental validity of the WoSCi in explaining work engagement, burnout and job performance over and above other constructs that capture capital namely, personal SC (Chen et al., 2009) and psychological capital (PsyCap; Luthans et al., 2007). The common ground between the different forms of capital is that they create value (Luthans et al., 2007). However, they contribute towards that end in different ways. While in the case of WSC employees evaluate whether trust, reciprocity and social networking characterize interactions within their organization, in the case of personal SC actors evaluate the amount to which these aspects characterize their own personal interactions (Chen et al., 2009). PsyCap, just like personal SC, is a personal resource that consists of high levels of self-efficacy, hope, resilience and optimism (Luthans et al., 2007). However, although personal SC or WSC refers to trust, reciprocity and social interactions that may contribute to problem solving through answering questions like “to whom can I turn for help”, PsyCap refers to personal characteristics that promote problem-solving through personal development (Luthans et al., 2007). Hence:

Our study contributes to the development of the WSC construct by synthesizing a theory-driven definition. Although studies on WSC are constantly increasing, there is no consensus regarding its definition (Read, 2014). To address this issue, we applied a theoretically driven approach, and defined WSC based on fundamental distinctions regarding its forms and core facets. Thus, the proposed definition of WSC as an ecological workplace resource is theoretically justified and valid since it incorporates all its central components, while it also dictates how these inter-relate with each other forming the concept. In this way we expand the nomological network of WSC by clarifying what WSC is, how it is constructed and how it relates with similar, yet different constructs.

Our theoretical analysis suggested that WSC is a higher-order construct that evolves at the bonding and bridging levels each of which consist of a cognitive (which in turn is composed of trust and reciprocity lower-order factors) and a structural (i.e., network) component. Although, the results of the EFAs suggested that trust and reciprocity were merged in one factor both at the bonding and bridging levels, the MLCFAs supported our theoretical model by showing that trust and reciprocity are distinct underlying components of cognitive WSC. This is important when considering that most existing approaches (Kouvonen et al., 2007; Meng et al., 2018; Ommen et al., 2009) do not clearly distinguish between trust and reciprocity or do not evaluate them at both bonding and bridging levels. Trust and reciprocity are closely related constructs. Trust refers to expectations of honest behavior and is linked to the belief that others will react positively to an initiative, or at least they will not undermine it. Thus, it makes the various parties during a social interaction reliable (Fukuyama, 1997). Reciprocity is related to support exchange and refers to the provision of help, with the expectation that at some point it will be reciprocated (Harpham et al., 2002). Despite their similarities, trust and reciprocity differ in that reciprocity is a broader concept, which integrates expectations regarding social action and includes the criterion of practical mutual assistance (Harpham et al., 2002). An advantage of our approach is that it manages to theoretically and empirically support the unique contributions of trust and reciprocity in the WSC construct.

Implications regarding measurement are also manifold. We argued and showed that the WoSCi fully captures WSC and not just a part of it since it measures both the cognitive and the network facet of WSC, while incorporating the bonding/bridging distinction. This is relevant because it supports empirically that the cognitive and structural components of WSC may exist at different directions (i.e., inward/outward). As such, the WoSCi overcomes limitations of existing measures that do not include reciprocity (Pejtersen et al., 2010), do not distinguish between reciprocity and trust (e.g., Ommen et al., 2009), or do not evaluate these dimensions at both bonding and bridging levels (Eguchi et al, 2017; Kouvonen et al., 2007). Moreover, the WoSCi may be used to assess WSC in every work environment. Despite the complexity of its theoretical structure that often results in non-user-friendly instruments (e.g., Meng et al., 2018), the WoSCi measures WSC in a straightforward way in all organizational contexts irrespective of their size or structure. Importantly, the proposed structure of WSC, as measured with the WoSCi, was supported when accounting for the nested structure of the data, which further supports the stability of the construct, and its potential, multilevel nature.

The empirical and practical value of the WoSCi is further boosted by the findings supporting that is a reliable and valid tool. Correlational analyses supported the convergent validity of the WoSCi since it was found to relate substantially with other scales measuring the same construct as well as scales measuring related constructs. Although WoSCi correlated positively and highly with justice and leadership, the associations were not as high as to imply overlapping constructs. Especially concerning justice, results did not suggest that it is a core element of WSC, as implied by Petersen et al. (2010). Also, the positive correlations of WSC with other job resources, as well as the criterion-related validity results showing that it relates positively to work engagement and performance, and negatively to burnout, indicate that WoSCi can be perceived as a job resource (Bakker & Demerouti, 2017).

Last but not least, results provided evidence regarding WSC' s incremental validity in explaining significant outcomes, over and above other forms of capital. Namely, we showed that WSC explains variance in work engagement, burnout and extra-role performance beyond personal SC (Chen et al., 2009) and PsyCap (Luthans et al., 2007). These findings suggest that workplace capital matters and may complement individual forms of capital. However, the incremental validity of WSC was not supported for in-role job performance, indicating that PsyCap, alone or together with individual SC, are more relevant in explaining this facet of performance. This may be attributed to the nature of WSC that is more closely related to employees’ attitudes that are driven by altruism (i.e., extra-role performance), since it reflects the team spirit (i.e., trust, reciprocity) within organizations. In contrast, PsyCap reflects individual qualities that are necessary for responding to setbacks and for invest the necessary effort to achieve formal duties requirements (Luthans et al., 2007).

Our results have important implications for organizational practice. Considering the incremental validity results, organizations may implement interventions aiming at enhancing WSC. Such interventions should aim at promoting trust and reciprocity within the organization both among employees and across levels of hierarchy. Relevant action plans, that could be put in practice through workshops and other training methods, may include tasks targeting to communication, social cohesion and cooperation improvement, energetic involvement of employees in decision making process and development of shared understanding regarding work tasks (Meng et al., 2019). Also, it is important for organizations to promote networking opportunities within organization and facilitate processes that allow developing partnerships with other organizations in order to expand the network of an organization and the opportunities for employees’ professional and social development.

Our study has some limitations. First, we did not examine the test–retest reliability of the WoSCi so we do not have evidence for its stability over time. Second, criterion-related validity was also assessed using cross-sectional data. Future research should make use of longitudinal data to further test the validity of the scale. Third, all studies have been conducted in Greece, so we do not have evidence regarding the reliability and validity of the WoSCi in different cultures. Future cross-cultural studies would allow testing the factorial invariance of the scale across different contexts. Forth, we used convenience samples, which are characterized by low generalisability. However, the use of heterogeneous samples somewhat counteracts this limitation.

To conclude, in this study we defined WSC as an ecological, job resource and we developed a reliable and valid instrument to measure this construct. WSC was found to be relevant in explaining employee well-being and job performance over and above other forms of capital. Our results highlight the theoretical, empirical, and practical relevance of WSC as a type of resource that may promote flourishing in organizations.