Relationships matter: how workplace social capital affects absenteeism of public sector employees

ABSTRACT

Although absenteeism is a key concern in most western societies, research on reducing absenteeism in public sector organizations is scarce, particularly regarding the impact of organizational relationships. By building on the concept of workplace social capital (WSC) and using a large longitudinal cohort of Danish municipal employees, this study shows that three types of WSC (bridging, direct-leader-linking, and top-level-linking WSC) reduce absenteeism, while there is no significant effect of bonding WSC. Our empirical results further suggest that the relationships with the immediate leader and the top management (direct-leader-linking and top-level-linking WSC) are most important for employees’ absenteeism.

KEYWORDS:

• Workplace social capital
• organizational relationships
• sickness absence

Introduction

Absenteeism is a key concern in most western societies and an increasingly important theme in public management. Both in European and American contexts, research has shown that absenteeism is traditionally higher in the public sector than in the private sector (Hansen et al. 2018; Mastekaasa 2020) and researchers have argued that absenteeism will probably increase in many public sectors in the future due to challenges related to lower resources and increased public costs (Moynihan and Pandey 2008; Kroll, DeHart-Davis, and Vogel 2019; Chordiya, Sabharwal, and Battaglio 2019). This accentuates the question how managers can reduce sickness absence. While some absence is of course inevitable, prior literature indicates that some absence is avoidable at the employee’s discretion and potentially affectable by management (Rho, Yun, and Lee 2015; Jensen, Andersen, and Holten 2019). However, Shahidul, Wright, and Yukl (2014) point out that, despite the importance of this topic, research on how to reduce absenteeism in public sector organizations is scarce.

The seminal works and reviews on absenteeism research of Johns (1997), Miraglia and Johns (2021), and the research overview of Lohaus and Habermann (2019) present a variety of person-related, work-related, and organizational variables as absence determinants. They argue that the social perspective on absenteeism represents one of the greatest advances in the field and plays a critical role in shaping attendance patterns. While studies in public management research have focused on individual motivational patterns (Wright, Hassan, and Christensen 2017; Jensen, Andersen, and Holten 2019; Gross, Thaler, and Winter 2019) and formal organizational incentives or personnel policies (e.g. Paola, Scoppa, and Pupo 2014) as determinants of employee absenteeism, knowledge on the social perspective is still sparse. A few studies have focused on organizational relationships in relation to public sector employees’ absenteeism and have found either beneficial or non-significant effects (Caverley, Cunningham, and James 2007; Nielsen et al. 2006; Shahidul, Wright, and Yukl 2014; Török et al. 2018; Gross, Thaler, and Winter 2019). Yet, they largely focus on one or two specific relationships (e.g. to the team or the leader) and not the overall quality or relative importance of different types of workplace relationships. However, due to the higher levels of absenteeism in the public sector (Hansen et al. 2018; Mastekaasa 2020) and its peculiarities of service provision and inner-organizational relationships (Cunningham, James, and Dibben 2006; Caillier 2017), the need for unique evidence for the public sector is still warranted.

The aim of this study is to comprehensively examine the social perspective on absenteeism in the public sector. We therefore build on the concept of workplace social capital (WSC). WSC refers to the strength and quality of the social relationships among actors in an organization (Leana and van Buren 1999; Adler and Kwon 2002; Oksanen et al. 2010; Rugulies et al. 2016; Meng, Clausen, and Borg 2018). The concept comprises structural, relational, and cognitive elements, i.e. it pertains to the level of connectedness, mutual trust, common goals, and shared mental models across actors (Nahapiet and Ghoshal 1998; Borg, Mateu, and Clausen 2014; Kroll, DeHart-Davis, and Vogel 2019; Ko 2021).

Employees are embedded in several social relationships in an organizational context, which can be divided into different types (e.g. horizontal and vertical relations). For WSC, up to four types or dimensions are differentiated in the literature; (1) relationships between co-workers within work teams (bonding WSC), (2) relationships between work teams (bridging WSC), (3) relationship to the direct superior (direct-leader-linking WSC), and (4) relationship to the top management or the organization in general (top-level-linking WSC) (Putnam 1993a; Szreter and Woolcock 2004; Borg, Mateu, and Clausen 2014; Meng, Clausen, and Borg 2018). Overall, WSC improves collaboration and communication between colleagues and groups (Meng, Clausen, and Borg 2018). Therefore, WSC is a social resource that public managers may use actively, for the benefit of the employees and the organization (Nahapiet and Ghoshal 1998).

Until now, only few studies have analysed WSC in public sector organizations. Those studies find largely beneficial effects of social capital in organizations on employee attitudes (Gächter, Davd, and Torgler 2011; Løkke 2016; Framke et al. 2019; Kroll, DeHart-Davis, and Vogel 2019; Ko 2021) and on health and well-being (Boyas and Wind 2010; Løkke 2016; Framke et al. 2019; Ko 2021). One study assessed the link between overall WSC and absenteeism in the public sector (Török et al. 2018). However, this study focused solely on hospital employees and only examined long-term sickness absence, which is generally less affectable by management than short-term sickness absence (Mastekaasa 2020). While the studies differ in their conceptualizations of social capital at work, they all focus on examining overall WSC. Thus, even though the WSC types are theoretically well-established, the empirical evidence on the relative importance of the four types is extremely sparse (Meng, Clausen, and Borg 2018).

With our study, we contribute to the literature by deepening the social perspective on absenteeism in the public sector, and thoroughly examine how various organizational relationships (the four types of WSC) influence employees’ absenteeism. We extend prior research by differentiating between the four WSC dimensions (bonding, bridging, direct-leader-linking, and top-level-linking). This helps us to generate new evidence on the overall relevance of organizational relationships on employee absence as well as disentangle and compare these effects across different types of relationships. By testing hypotheses based on various theories within the social perspective (Miraglia and Johns 2021), we contribute to the understanding of which theoretical pathways and underlying theories may be most important for explaining how organizational relationships affect employee absences. Using a longitudinal dataset on a large sample of public sector employees, we derive robust and unique evidence for the public sector. For managers, this evidence can be used to argue for and prioritize investments in different types of WSC within public sector organizations.

Theory

Social capital was introduced in social science by Bourdieu (1986), Coleman (1988), and Putnam (1993a). In this paper, we depart from Putnam’s understanding of social capital (SC) as ‘Features of social organization such as trust, norms and networks that can improve the efficiency of society by facilitating coordinated actions’ (Putnam 1993a, 167). SC entails strong and well-functioning societal ties, and Putnam considers SC a resource for both the individual and the collective good (Putnam 1993b, 2000). In a societal perspective, SC has been discussed in a wide range of disciplines including politics (Putnam 1993a), economics (Fukuyama 1995), sociology (Bourdieu 1986; Coleman 1988), organization and management (Nahapiet and Ghoshal 1998; Adler and Kwon 2002; Boyas and Wind 2010; Kroll, DeHart-Davis, and Vogel 2019; Ko 2021), and population health and epidemiology (Kawachi et al. 1997; Oksanen et al. 2008; Rugulies et al. 2016).

When social capital is discussed within an organizational setting, it is often termed workplace social capital (WSC) (Liukkonen et al. 2004; Kouvonen et al. 2008; Oksanen et al. 2008; Väänänen et al. 2009; Oksanen et al. 2010, 2013; De Clercq et al. 2015; Rugulies et al. 2016; Kenneth and Andersen 2018; Hansen et al. 2018; Meng, Clausen, and Borg 2018; Framke et al. 2019) or organizational social capital (Kroll, DeHart-Davis, and Vogel 2019; Ko 2021). WSC has frequently been conceptualized as a multi-dimensional concept. Based on Nahapiet and Ghoshal’s work, one approach is to differentiate between structural, relational, and cognitive aspects of SC (Nahapiet and Ghoshal 1998; Kroll, DeHart-Davis, and Vogel 2019; Ko 2021). The structural aspect concerns the pattern of connections between actors, i.e. the extent of interaction and communication between employees within an organization. The relational aspect relates to the quality of social relationships and the strength of attachment within an organization, encompassing reciprocity, trust, and obligations between organizational actors. The cognitive aspect relates to collective understanding, i.e. the degree to which employees and leaders have a shared understanding of their work tasks, the teamwork, and the objectives of the organization. These three aspects form the quality of the relationships and the strength of the attachment between the members in a social system (Nahapiet and Ghoshal 1998).

Another approach is to differentiate between different types of relationships. Putnam originally introduced the distinction between two types of horizontal relationships in society, namely bonding and bridging SC (Putnam 1993a). Szreter and Woolcock (2004) added a third and vertical relationship – linking SC – to describe the links between different hierarchical layers in society. These distinctions have been translated into an organizational context, as relationships between co-workers within teams (bonding WSC) and across teams (bridging WSC) on the horizontal level, and employees’ relationships towards their immediate leader (linking WSC) on the vertical level. When developing a validated scale on WSC, Borg, Mateu, and Clausen (2014) further divided the vertical relationship into two distinct types describing the relationship to the direct leader (in this paper direct-leader-linking) and the relationship to the top-management and the overall organization (in this paper top-level-linking).

Theoretically, these two multidimensional approaches could evolve into a combined, 12-dimensional construct. In practice, however, this has not been applied. Instead, Borg, Mateu, and Clausen (2014) proposed to include indicators relating to structural, relational, and cognitive aspects within each of the four relationships in their validated measure of WSC. This seems justified as the three aspects are attested to have high interdependencies (Borg, Mateu, and Clausen 2014; Kroll, DeHart-Davis, and Vogel 2019). For our research purpose, differentiating between the different types of relationships seems the most purposeful, since very few studies have empirically investigated how the WSC types may affect employee outcomes (see Meng, Clausen, and Borg 2018 for an exception).

A definition of WSC and its dimensions is provided in Table 1. It also includes a column referring to selected studies on (W)SC as well as related and similar constructs used in prior public management research. This allows us to compare results, connect with and add insights to previous research within public management, and demonstrate WSC’s usefulness as an overarching concept.

Table 1. Workplace social capital (WSC) and related concepts.

WSC	Definition	Exemplary studies analysing WSC or related concepts*
Overall WSC	Refers to supportive and trusting relationships among peers and between employees and management within an organization. It represents the level of connectedness, mutual trust, and shared mental models and understandings among the actors within an organization. It comprises structural, relational, and cognitive aspects within all relationships within a workplace on the horizontal axis and the vertical axis.	WSC (Boyas and Wind 2010; Framke et al. 2019; Gächter, Davd, and Torgler 2011; Kroll, DeHart-Davis, and Vogel 2019; Ko 2021; Løkke 2016; Török et al. 2018) Concepts measuring strength and/or quality of workplace relationships Climate of inclusion (Brimhall, Lizano, and Barak 2014) High-quality workplace relationships (Caillier 2017) Human Relations Organizational Climate (Ancaraniet al. 2018, 2019)
Bonding WSC	Refers to relationships within a work group or a unit. These relationships are characterized by high frequency and proximity with a high degree of reciprocity. It embraces perceptions of colleague support and information sharing (structural), a sense of trust, respect and recognition (relational), a shared vision and shared mental models of the work task (cognitive) within a team.	Bonding WSC (Kenneth and Andersen 2018; Meng, Clausen, and Borg 2018) Concepts measuring strength and/or quality of relationships within teams Collegiality (Jana and Wallace 2002) Colleague interest (De Clercq et al. 2015) Colleague support/help (De Clercq et al. 2015; Boyas and Wind 2010; Moynihan and Pandey 2008; Nielsen et al. 2006; Miraglia and Johns 2021) Group culture (Moynihan and Pandey 2007) Group cohesion (Miraglia and Johns 2021¶) High-quality relationships to colleagues (Read and Laschinger 2015; Caillier 2017) Team climate inventory (Gross, Thaler, and Winter 2019; Kivimäki et al. 2001) Team-Member-Exchange (Ancarani et al. 2018; Schermuly and Meyer 2016) Trust in colleagues (Ancarani et al. 2018; Caverley, Cunningham, and James 2007)
Bridging WSC	Refers to relationships between different groups at the workplace. These relations are less dense and the relationships are weaker than under bonding WSC. It comprises a functioning communication and information flow(structural), mutual respect and shared obligations (relational), and shared objectives (cognitive) across units or teams.	Bridging WSC (Kenneth and Andersen 2018; Meng, Clausen, and Borg 2018) Concepts measuring strength and/or quality of relationships across teams Relational coordination (Bolton, Logan, and Gittell 2021¶)
Direct-leader-linking WSC	Refers to the relationship between the employee and his direct manager. The tie to the direct manager is the most frequent, proximal, and steady tie of the vertical power relationships in the organization. Direct-leader-linking WSC is characterized by interactions related to the employees’ daily practice. It covers perceived supervisor support and participatory leadership (structural), supervisor concern, mutual respect and recognition (relational), and a shared vision (cognitive).	Linking WSC (Kenneth and Andersen 2018; Meng, Clausen, and Borg 2018) Concepts measuring strength and/or quality of employee-leader relationships Leadership – including ethical, supportive, and inspirational leadership (Shahidul, Wright, and Yukl 2014; Nyberg et al. 2008; Montano et al. 2017) Leader-member-exchange (Brimhall, Lizano, and Barak 2014; Meng and Wu 2015; Nyberg et al. 2008; Montano et al. 2017) supervisor support (Nielsen et al. 2006; De Clercq et al. 2015; Boyas and Wind 2010; Caverley, Cunningham, and James 2007) Trust in leadership (Dirks and Ferrin 2002¶)
Top-level-linking WSC	Refers to the relationship between employee and top-level management and/or the organization per se. Top-level management operates on different interaction terms than the immediate managers and often has fewer direct contact with individual employees in the organization. Yet, the top-level management directs the vision and defines organizational policies and practices on a symbolic level Top-level-linking WSC includes employee involvement (structural), organizational trust (relational), and a sense of cohesion (cognitive) at top management or organizational level.	Top-level-linking WSC (Meng, Clausen, and Borg 2018) Concepts measuring strength and/or quality of employee-management/organization relationships Organizational trust/trust in organizational leadership (Dirks and Ferrin 2002¶) Perceived organizational support (Gross, Thaler, and Winter 2019; Kurtessis et al. 2017) Organizational justice (Duijts et al. 2007; Elovainio et al. 2005) Top management support (Dello Russo et al. 2013)

¶literature review; † meta-analysis.

*Note: We do not refer to studies on societal or community social capital, because these do not relate to within-organization relationships. We do not distinguish whether the studies call it WSC, OSC, or similar. We mainly selected literature reviews, meta-analyses and studies on public sector employees.

WSC and absenteeism

Absenteeism is a complex phenomenon related to several factors of an employees’ personal life and work life (Johns 1997; Mastekaasa 2020). There is general agreement that absence from work should be seen as a function of both the ability and the motivation to attend work (De Boer et al. 2002; Johns 1997; Mastekaasa 2020). To explain how WSC and its different dimensions affect absenteeism, we therefore build on the dual path model described by Miraglia and Johns (2016). This dual path model implies that absence behaviour may simultaneously be affected by health impairment and motivational paths. In a recent review, Miraglia and Johns (2021) further depict how social and relational factors at multiple levels may affect absenteeism through different and multidisciplinary theoretical pathways.

These pathways encompass theories on normative social influence (Festinger 1954; Salancik and Pfeffer 1978; Addae, Johns, and Boies 2013), social exchange (Gouldner 1960; Blau 1964; Biron and Bamberger 2012; Rho, Yun, and Lee 2015), job resources (Hobfoll and Ford 2007; Grant and Parker 2009; Halbesleben et al. 2014; Bakker and Demerouti 2017), work attitudes (Johns 1997), emotions (Grandey, Dickter, and Sin 2004), and ethics (Nicholson and Johns 1985; McKevitt et al. 1997; Baker-McClearn et al. 2010; Krane et al. 2014). Both Johns (1997) and Miraglia and Johns (2021) argue that, in reality, the different paths most often interact with each other and therefore are futile to separate empirically. In the following, we will draw on these theoretical pathways, in particular the social exchange theory, to derive how WSC affects absenteeism.

Overall WSC may affect both the ability and the motivation to attend work. According to the job demands-resource model (Bakker and Demerouti 2017), supportive relationships are a key organizational (social) resource that reduces absenteeism by offsetting the negative effects of job demands and helping to deal with various job situations, alleviating stress and exhaustion (Caillier 2017). In contrast, conflictual experiences on the job can deplete individual resources, forcing employees to use a variety of means, including taking sick leave, to restore them. Strong organizational relationships can also ease the access to other job resources, such as information and expertise, which are beneficial for buffering against job demands, stress, and exhaustion. Another fruitful perspective is the social exchange path. According to social exchange theory (Blau 1964) and the norm of reciprocity (Gouldner 1960), employees see the relationship with other organizational members as beyond the standard economic contract such that the parties operate on the basis of trust, goodwill, and the perception of mutual obligations. The exchanges denote high-quality relationships, and issues of care and consideration in the relationships are central. In this exchange relation, employees reciprocate supportive and high-quality workplace relationships with increased commitment and pro-organizational behaviour, discouraging absenteeism. In contrast, workplace conflicts are often characterized by failures of fair social exchange and violations of reciprocity norms, which in turn can determine exit behaviours and group disintegration, triggering absenteeism (Miraglia and Johns 2021). Positive relationships at work may also strengthen motivational states (e.g. work engagement) and positive attitudes towards the job and organization (i.e. satisfaction and commitment) (Johns 1997), thus preventing withdrawal behaviour (Miraglia and Johns 2016). Based on these theoretical understandings, we hypothesize that:

H1:

Overall WSC reduces public sector employees’ absenteeism.

The theoretical pathways can also be applied to explain how the four dimensions of WSC affect employee absenteeism. Within the horizontal relationships, bonding WSC provides employees with social support from the closest co-workers (Murayama, Fujiwara, and Kawachi 2012). Parallel to the theoretical mechanisms described above, such support may be a resource that decreases absenteeism. Furthermore, support from close co-workers may also buffer against distress caused by emotional labour, which is often high in public service jobs. Since absenteeism might be employed as a mood regulation mechanism to cope with low positive affect at work or to regulate negative experiences (Miraglia and Johns 2021), support from close colleagues may counteract this. Furthermore, bonding WSC reduces unclear job expectations and team conflicts, which are positively related to absenteeism (Johns 1997). Finally, bonding WSC may create a strong sense of in-group cohesion, which may activate mechanisms such as social identification and role modelling. Various social influence theories explain the power of these normative mechanisms (Festinger 1954; Salancik and Pfeffer 1978; Tajfel and Turner 1986; Hogg and Abrams 1988), as they illuminate why and how individuals from the same social unit can foster agreement on and adherence to specific attendance norms and the degree of absenteeism legitimacy (Miraglia and Johns 2021). Such normative influences may induce a moral obligation towards the team, which might urge employees to come to work to help colleagues under minor illness (especially if the employee cannot be replaced in the team) (Miraglia and Johns 2016, 2021). However, these mechanisms may also facilitate absence, depending on the salient norms and expectations in relation to absenteeism in the group (Miraglia and Johns 2021). Often normative standards tolerant of absenteeism are higher among co-workers than managers (Dello Russo et al. 2013; Miraglia and Johns 2016), which can activate a contradictory path that increases absence behaviour. This may of course attenuate the positive effect of social support within bonding WSC. However, despite the possible opposing normative paths to absenteeism, we propose that bonding WSC increases both the ability and motivation to attend work because it represents a built-up benefit for employees to remain and invest in a team and, therefore, be less inclined to engage in withdrawal behaviour.

Bridging WSC builds on horizontal relationships across teams which are more distant and have less frequent contact. Yet, in line with relational coordination theory (Bolton, Logan, and Gittell 2021), shared goals, shared knowledge, and mutual respect across teams might contribute to a general beneficial perception of the workplace, provide social resources which reduce transactional costs, buffer against workplace-induced stress, reduce the likelihood for conflicts in the workplace, and, hence, increase employee attendance motivation and health.

Direct-leader-linking WSC, is characterized by perceived supervisor support and concern, mutual respect and recognition, and participatory leadership. It thereby represents an important social resource in vertical organizational relationships. Direct-leader-linking WSC highly affects work motivation as the immediate leader is a substantial source of feedback and acknowledgement, which also strongly affects employees’ emotional state and reactions. Thereby, leaders are central in creating positive job attitudes, such as satisfaction and engagement, which also increase personal health (e.g. Shahidul, Wright, and Yukl 2014) In addition, the social exchange perspective is highly relevant for the subordinate-superior relationship in the sense that absenteeism might represent a form of worker resistance to supervisory conflict or incivility. In contrast, individuals who feel that their leader has, or will, demonstrate care and consideration will reciprocate this sentiment in the form of desired behaviours (Dirks and Ferrin 2002; Miraglia and Johns 2021). Thus, direct-leader-linking WSC is likely to positively affect both the ability and motivation to attend work.

Top-level-linking WSC is the employees’ perception of the top-level management and therefore embraces more distant relations. Still, high top-level-linking WSC contributes to a strong sense of organizational cohesion, organizational trust, and participatory leadership. Top-level-linking WSC is a central driver for the creation and acceptance of the overall work ethic defining the organizational rules and expectations regarding absenteeism (Johns 1997). Strong top-level-linking WSC induces a moral obligation to attend and stresses values of responsibility and social commitment. Therefore, top-level-linking WSC may reduce withdrawal behaviour. Moreover, the top-level management infuses meaning into the organization by directing the vision and defining organizational practices on a symbolic level, influencing employees’ work motivation and commitment (Moynihan and Pandey 2008; Meng, Clausen, and Borg 2018), which are highly important for reducing absenteeism (Johns 1997). Further, top-level-linking WSC also contains a certain level of employee involvement, which according to the Job-Demand-Control-Model reduces stress reactions and stimulates attendance to work (Karasek 1979; Miraglia and Johns 2016). In addition, absenteeism is a key commodity for employees who feel that the social exchanges are imbalanced at the organizational level, e.g. due to a lack of cohesion, trust, or involvement. In this case, employees might use absences as a means to punish the top management or the organization as a whole because absences reduce productivity and harm organizational performance. In contrast, trust in organizational leadership or high top-level-linking WSC may entail reciprocation in the form of fewer absences (Dirks and Ferrin 2002). Overall, top-level-linking WSC is probably of high importance for employees’ ability and motivation to attend work.

In sum, we expect that each of the four dimensions relates to absenteeism, as expressed in the following sub-hypotheses.

Hypotheses 1a-d: bonding, bridging, direct-leader-linking, and top-level-linking WSC reduce public sector employees’ absenteeism.

The relative importance of the different WSC types

In the broad and intertwined variety of relevant theoretical perspectives (Miraglia and Johns 2021), social exchange theory is a valuable starting point to derive propositions on the relative importance of the four WSC types for employees’ attendance motivation (Blau 1964; Biron and Bamberger 2012; Rho, Yun, and Lee 2015). The importance of absenteeism as a commodity in the exchanges will differ across the different parties (i.e. team colleagues, colleagues of other units, the direct leader, or the top management), because absenteeism will affect the parties differently. Hence, we can assume that the four different types of WSC are differently important for explaining absenteeism.

Sickness absence is usually recognized on an organizational level and affects organizational performance. It may therefore be a highly important factor for organizational-level reciprocity. Given this high (perceived) relatedness between absence and the consequences for the overall organization, it can be assumed that top-level-linking WSC has a relatively strong negative impact on employee absence (Dirks and Ferrin 2002; Dello Russo et al. 2013). In this context, employees may reciprocate perceived organizational trust and involvement with pro-organizational, absence-minimizing behaviour. Further, normative standards of absenteeism are strongly related to overall organizational culture and formed by a common organizational rhetoric that reaches beyond the boundaries of single teams and/or departments within organizations (Gaziel 2004; Jourdain and Chênevert 2015). Hence, top management has a strong influence on attendance norms as well as on work ethic, which means that in turn a high sense of cohesion – i.e. high top-level-linking WSC – will increase employees’ adherence to these norms. Therefore, the top management is probably the most important driver for the generation and adherence to the shared norms to attend work. In sum, both the social exchange perspective and the organization’s impact on the creation and acceptance of norms and work ethics speak for the highest importance of top-level-linking WSC in reducing employee absence.

Of course, absenteeism may also affect the immediate leader and the employees’ own team. With respect to the relative importance, we assume that direct-leader-linking WSC on the one hand and bonding WSC on the other hand have a relatively less strong impact on absenteeism, compared to top-level-linking WSC. The logic underneath this – again based on the social exchange theory – can be retraced to the more direct but short-term consequences of an employee’s absence for immediate leaders and close colleagues within a team. First, for the immediate leader in terms of short-term resource allocation and planning, and, second, for the close colleagues, who may face a need to follow up on ongoing responsibilities. Beside this, the amount of resources that employees may receive from the different parties differs. The direct leaders are probably the most important provider of beneficial resources, as they possess the means to affect the employees’ workload, information, and autonomy (Boyas and Wind 2010; De Clercq et al. 2015). Hence, we expect direct-leader-linking WSC to be more important for employees’ decision on absenteeism than bonding WSC.

Lastly, bridging WSC is likely to be least important for reducing sickness absenteeism. While collaboration, information flow, and mutual respect across teams are eventually important for the work situation and employee motivation (Fujita et al. 2016; Meng, Clausen, and Borg 2018; Kenneth and Andersen 2018), the relations across teams probably have a less important role than the team itself, the direct leader, or the top management. For instance, sickness absence usually has a lesser effect on other teams or departments, as the immediate consequences of absence are not directly related to the daily operational planning or goals in other teams or departments (at least not when it comes to short-term absenteeism).

Overall, we assume that top-level-linking WSC is the most important driver, followed by direct-leader-linking and bonding WSC, while bridging WSC is least important for absenteeism. We therefore expect that:

H2:

Top-level-linking WSC will have the strongest effect on public sector employees’ absenteeism, followed by direct-leader-linking, bonding, and bridging WSC.

Methods

Sample and data

This study examines public-sector employees working in a large Danish municipality via survey- and register-based data. The Danish municipalities function according to the principle of subsidiarity and are multipurpose organizations. Hence, the municipality employs a broad spectrum of occupations within social services, elderly- and child care, schools, job centres, environment, culture services and public administration. The study is part of a larger research project described by Pihl-Thingvad et al. (2020).

The survey data stems from questionnaires sent to all employees and managers in the municipality in 2014 and 2017, nested in approx. 550 organizational units (average unit size is 11 respondents, ranging from 1 to 190 respondents). The survey was distributed to a total of 9,046 individuals in 2014 (7,848 responded, which is a response rate of 87%) and to 8,643 individuals in 2017 (7,373 responded, which is a response rate of 85%). We excluded respondents who were employed less than twelve months in 2014 or 2017 or who had multiple employments in the municipality which led to more than twelve months of employment in a year. As in many Western public sectors (Mastekaasa 2020), 80% of the study population were women.

The obstacles of common source and common method bias were avoided by using a register-based measure of absenteeism. Occurrence of absenteeism in the municipal labour force was obtained from a database managed by Silkeborg Data, which provides remuneration systems to many Danish municipalities. Sickness absence from work is registered – in days – from the first day of absence until the employee’s return to work.

Measures

WSC was measured by fifteen survey items that covered bonding, bridging, direct-leader-linking, and top-level-linking WSC (see Table 2). The items were taken from the validated Social Capital Questionnaire (Borg, Mateu, and Clausen 2014). All survey questions were answered using a five-point Likert scale ranging from 1 (‘to a very low degree’) to 5 (‘to a very high degree’). As we were interested in the effects of employees’ perceptions on positive and supportive organizational relationships, we measured WSC at the individual level and created an individual additive index, based on unweighted item averaging, ranging from 1 to 5 for each of the four dimensions of WSC. Consistent with previous research, we included a second order, one-dimensional measure of WSC, which is reflected in an overall index containing all 15 survey items.

Table 2. WSC dimensions and indicators.

WSC dimension	Items
Bonding WSC	In our group, we help colleagues who have too much to do In our group, we agree on what is most important in our work tasks I trust the group’s ability to do the work well There is a sense of unity and cohesion in my group
Bridging WSC	Our group and other groups/departments recognize each other’s contribution to solving the tasks at work We agree on the main goals of our work across the departments of the organization Other groups provide us with the information we need for our work to be done well Other groups have a great understanding of the work we do in our group
Direct-leader-linking WSC	Our immediate manager considers our needs and point of view when making decisions The immediate manager helps solve specific problems we face during the workday Our immediate manager has great understanding of the work we are doing The relationship between our group and our immediate manager is characterized by mutual respect and recognition
Top-level-linking WSC	Employees are involved in decisions about changes of the workplace There is a common understanding between management and employees about how to perform the work tasks In our group we feel a strong connection to our workplace

Sickness absence. Absenteeism was measured as a continuous variable reflecting the number of sick days per year (in 2014 and 2017). It is based on an official register which records all sickness absence periods of the employees. For sensitivity analyses, we used sickness absence frequencies (i.e. how many times an employee had taken sick leave, independent of the number of days absent) and long-term absences (i.e. a dummy whether or not an employee had at least one absence longer than 30 days in a year).

Co-variates. Age, tenure and gender are three demographic characteristics which have been included in various theories and models on attendance (e.g. Steers and Rhodes 1978; Johns 1997) and for which thorough empirical evidence has been accumulated (e.g. Hackett 1990; Johns 2008; Løkke and Krøtel 2020; Mastekaasa 2020). We therefore included gender (0 = male, 1 = female), age (in years), and tenure (in years) in our models as control variables. Furthermore, prior research has pointed to the relevance of leadership for absenteeism (Johns 1997; Shahidul, Wright, and Yukl 2014; Török et al. 2018). Thus, we also controlled for whether the employee was assigned staff responsibility (0 = no staff responsibility, 1 = staff responsibility). Finally, differences due to organizational settings have been assumed and identified (Hansen, Løkke, and Sørensen 2019; Løkke and Krøtel 2020; Mastekaasa 2020); to rule out that they bias our results, we account for the organizational unit an employee works in.

Reliability and validity of WSC measure

The results of a four-factor solution derived by exploratory principal-component factor analysis with oblique rotation show that all items had sufficient factor loadings (≥.32) and no unacceptable cross-loadings (i.e. less than .15 difference from an item’s highest factor loading), as well as good internal consistency (Cronbach’s α ≥ .80 for all dimensions). Confirmatory factor analysis to test the theoretically assumed distinction between the four WSC dimensions resulted in a good model fit (CFI =.97, TLI =.96, RMSEA =.059, and SRMR =.038). Detailed results are available from the authors on request.

Data analysis

Descriptive statistics on absenteeism are displayed in Table 3. Because sickness absences are count data with great over-dispersion, yet without essential excess of zero absences, we chose negative binomial regressions for the data analysis.

Table 3. Descriptive statistics for main variables (2014 and 2017).

Variable	Year	N	Mean	SD	Min	Max
WSC
Overall WSC	2014	6,999	3.7	0.6	1	5
	2017	6,420	3.7	0.6	1	5
Bonding WSC	2014	6,961	3.9	0.7	1	5
	2017	6,395	4.0	0.7	1	5
Bridging WSC	2014	6,915	3.5	0.7	1	5
	2017	6,338	3.6	0.7	1	5
Direct-leader-linking WSC	2014	6,971	3.8	0.9	1	5
	2017	6,344	3.9	0.8	1	5
Top-level-linking WSC	2014	6,967	3.4	0.9	1	5
	2017	6,388	3.5	0.8	1	5
Control variables
Staff responsibility	2014	7,012	0.1	0.2	0	1
	2017	7,066	0.1	0.2	0	1
Female	2014	7,012	0.8	0.4	0	1
	2017	6,434	0.8	0.4	0	1
Age	2014	7,012	47.1	10.1	20	73
	2017	6,434	47.8	10.4	17	74
Seniority	2014	6,994	11.2	8.7	0	47
	2017	6,434	11.7	8.8	0	48
	Year	N	Mean	SD	Quartile 1	Quartile 2	Quartile 3	0 days absent
Sickness absence	2014	6,543	10.6	23.5	1	4	10	25%
	2017	6,217	10,0	21.7	1	4	9	22%
	2014–2017	4,286	1.033	27,67	−3	0	3	17%

Note: For sickness absence: Due to the count data nature of sickness absence, we report the quartiles instead of the minimum and maximum statistics. The last row reports changes in sickness absence from 2017 to 2017. The last column reports the share of individuals with no absence days. For the change 2014–2017 sickness absence (last cell), the number indicates the share of individuals with no change in sickness absence days 2014–2017.

To account for the fact that our data has a nested structure with three levels (time nested in individuals nested in organizational units), we estimated mixed-effects negative binomial models, which contain both fixed effects and random effects. In longitudinal data and panel data, random effects are useful for modelling intra-cluster correlation; that is, observations in the same cluster are correlated because they share common cluster-level random effects. This is relevant when analysing WSC. In addition, controlling for organizational units and applying a panel data approach levels out differences in absenteeism and WSC levels across occupations and service areas. We applied the mean-variance adaptive Gauss-Hermite quadrature as the integration method for the random effects on individual and unit levels.

We estimated four regression models for the main analysis, using two sets of main independent variables – overall WSC and the four WSC types separately – with and without controlling for gender, age, seniority, and whether an employee had staff responsibility. Coefficients were transformed into incidence rate ratios (IRR) to ease interpretation.

To test the relative importance of the different WSC types, we applied dominance analysis, which provides information on the relative importance of each predictor compared to any other predictor based on all subsets of the remaining predictors. Dominance analysis makes it possible to compare independent, potentially correlating variables in the model to each other and rank them according to their relative importance in explaining the dependent variable (Azen and Budescu 2003). For multi-level models, an extension of dominance analysis developed for linear regression has been proposed for determining relative predictor importance (Luo and Azen 2013). Specifically, we used the STATA domin with mixdom wrapper module, which relies on Snijders and Bosker’s (1994) R² metric (Luchman 2013). To use the module, we had to make slight adjustments to our model. First, the mixdom command only works for multilevel linear regression, so we log-transformed the sickness absence measure after adding one day in order to avoid losing all individuals with zero days absent. Second, mixdom only allows one level of clustering in the data; as more variation was captured on the individual than on the organizational unit level, we omitted the latter. To make sure that these adjustments did not distort our results, we compared the results of a two-level mixed regression model with log-transformed sickness absence to our main model, and the results remained stable. When we conducted the dominance analyses, the control variables were included in all subset models (Azen and Budescu 2003).

To conclude on each WSC dimension’s importance relative to the other dimensions, three forms of dominance are distinguished (Azen and Budescu 2003). General dominance statistics represent the weighted average marginal/incremental contribution to the overall fit statistic that an independent variable makes across all models. As such, general dominance statistics are also the most averaged, and hence least strong, dominance statistic. A slightly stronger statistic is conditional dominance, computed as the average incremental contributions to the overall model fit statistic within a single ‘order’ for models in which the independent variable is included – where ‘order’ refers to a distinct number of independent variables in the estimation model – in our case, four. Conditional dominance exists if one independent variable has larger conditional dominance statistics than another independent variable across all orders. The strongest form of dominance – complete dominance is derived by comparing all possible incremental contributions to model fit for two independent variables. Complete dominance occurs if one independent variable has a larger incremental contribution to model fit than another independent variable across all possible comparisons.

For sensitivity analyses, we repeated all analyses with absence frequencies as the dependent variable. To assess whether the results are comparable with results for long-term sickness absence, we generated a dummy whether an individual had at least one absence longer than 30 days in a year and estimated three-level mixed logistic regressions. All analyses were performed in Stata 16.0.

Results

Descriptive statistics are provided in Table 3 for the main variables of interest in 2014 and 2017, respectively. WSC (overall and its dimensions) is generally rather high, with bridging WSC and top-level-linking WSC on average lower than bonding and direct-leader-linking WSC. We found only very small variations in WSC averages between 2014 and 2017. Regarding our control variables, around 10% of the employees had staff responsibility, around 80% were female, average age was 47 (48) years and seniority was on average 11 (12) years in 2014 (2017). For our focal dependent variable sickness absence, the average number of days absent were 10 and 12 days in 2014 and 2017, yet with a very high variation, as indicated by a standard deviation of 24 and 28. As the sickness absence data is highly skewed, the mean is substantially affected by outliers; hence, we additionally report the quartiles: 25% (quartile 1) of the employees had no or only 1 absence day, 50% (quartile 2, representing the median) had 4 days or less, and 75% (quartile 3) of the employees had 10 or less absence days. Finally, more than every fifth employee did not have any sickness absence. The last column provides information on the changes in sickness absence days for the 5,192 employees present in both years; on average, sickness absence days increased by 2.5 days, yet again with a very high variation. 25% of the employees had reductions in sickness absence days of at least three days (quartile 1), while another 25% had increases in sickness absence days of at least four days (quartile 3). 16% of the employees had no change in sickness absence days, i.e. they had the exact same number of sickness days in 2014 and 2017.

Regressions results are displayed in Table 4. The first column represents the null model. A likelihood-ratio test shows that there is enough variability between organizational units and individuals to favour a mixed-effects negative binomial regression over a standard negative binomial regression. Column 2 and 3 report the results for the overall WSC index. In both models (without and with covariates, model 1 and 2), a one-unit increase in overall WSC yields an approximately 25% decrease in the number of days absent (IRR = 0.74 and 0.76), ceteris paribus. Thus, our results support H1 that overall WSC reduces employees’ absenteeism. All our control variables significantly related to absenteeism. Sickness absence was higher for women. It increased significantly with age but decreased significantly with seniority, ceteris paribus. Employees with staff responsibility had significantly less absence. Finally, absenteeism was significantly higher in 2017 than in 2014.

Table 4. Results of mixed negative binomial regressions for sickness absence.

Variables	Null model	Model 1	Model 2	Model 3	Model 4
Overall WSC (H1)		0.738	0.755
		(<0.001)	(<0.001)
Bonding WSC (H1a)				0.960	0.945
				(0.173)	(0.062)
Bridging WSC (H1b)				0.902	0.917
				(0.001)	(0.007)
Direct-leader-linking WSC (H1c)				0.936	0.939
				(0.010)	(0.013)
Top-level-linking WSC (H1d)				0.913	0.928
				(0.001)	(0.005)
Control variables
Woman			1.358		1.349
			(<0.001)		(<0.001)
Age			1.006		1.006
			(0.007)		(0.006)
Seniority			0.984		0.984
			(<0.001)		(<0.001)
Staff responsibility			0.350		0.351
			(<0.001)		(<0.001)
2017		1.077	1.094	1.069	1.086
		(0.005)	(0.001)	(0.012)	(0.002)
Constant	1.621	15.197	10.549	14.949	10.575
	(<0.001)	(<0.001)	(<0.001)	(<0.001)	(<0.001)
/lnalpha	0.095	1.095	1.087	1.095	1.087
	(<0.001)	(0.001)	(0.002)	(0.001)	(0.002)
Random-effects parameters
Org.unit var(_cons)	0.182	1.160	1.144	1.159	1.143
	(<0.001)	(<0.001)	(<0.001)	(<0.001)	(<0.001)
Individual (var_cons)	1.300	3.043	2.893	3.017	2.864
	(<0.001)	(<0.001)	(<0.001)	(<0.001)	(<0.001)
Observations	11,299	11,283	11,276	11,040	11,033
Number of groups	557	557	557	556	556
Number of individuals	7,080	7,074	7,073	6,986	6,985

Note. Incidence rate ratios (IRR) are reported; p-values in parentheses.

The last two columns (model 3 and 4) report the results for the four WSC dimensions. Each coefficient reports the estimated change in sickness absence by a one unit increase in each independent variable, holding all other variables constant. It shows that three WSC dimensions are significantly negatively related to sickness absence, supporting H1b-d. Only bonding WSC was not significantly related to sickness absence (no support for H1a). Looking at each dimension’s effect while holding all variables constant revealed that a change in each dimension was associated with a 6%-9% decrease in the number of days absent.

Table 5 shows the results from the dominance analysis comparing the relative importance of the four WSC types. Even though bonding WSC does not significantly affect sickness absence when all other variables are held constant, we included it as an independent variable and tested for its explanatory power because dominance analyses provide insights into the explanatory power of each variable if the variables are correlated, which is the case for the single WSC dimensions. The table displays the fit statistics, the general dominance statistics, and a vector of standardized general dominance statistics, which is the general dominance statistic vector normed or standardized to be out of 100%. The last two columns provide information on conditional and complete dominance. The WSC dimensions are ranked by dominance.

Table 5. Relative Importance of WSC Dimensions in Explaining Employees Sickness Absence.

Overall Fit Statistic: 0.0483 WSC Fit statistics: 0.0163 All Subsets’ Fit Stat. (CVs) 0.0320
WSC Dimensions	General Dominance Statistics	Standardized General Dominance Statistics	Conditionally dominates …	Completely dominates …
Top-level-linking WSC	0.0049	10.2%	Linking, Bridging, Bonding	Linking, Bridging, Bonding
Direct-leader-linking WSC	0.0045	9.4%	Bridging, Bonding	Bridging, Bonding
Bridging WSC	0.0036	7.4%	-	-
Bonding WSC	0.0033	6.8%	-	-

Note. Model controls for sex, age, seniority, and supervisory status and include year fixed effects.

The analysis shows that all independent variables explain about 4.8% of the variation in sickness absence. The R² values are comparable to those obtained in other multi-level analyses (e.g. Hargis et al. 2011; Oppel, Winter, and Schreyögg 2019). Among the WSC types, top-level-linking WSC has the highest relative importance (explaining on average 10.2% of all variance explained by the WSC types and control variables), followed by direct-leader-linking WSC (relative importance 9.4%) and bridging WSC (relative importance 7.4%), while bonding WSC has the lowest relative importance (6.8%) for predicting employee sickness absence. Of all variance explained by WSC, the vertical dimensions contribute almost 60%, while the horizontal dimensions contribute only slightly above 40%.

Considering the stronger forms of dominance, i.e. conditional and complete dominance, shows the following: Top-level-linking WSC completely dominates bonding, bridging, and direct-leader-linking WSC, underlining its highest relative importance. Direct-leader-linking WSC completely dominates bonding and bridging WSC, supporting its second strongest dominance. Bridging WSC only generally dominates bonding WSC. In sum, dominance analysis provides solid evidence that top-level-linking WSC has the highest relative importance for absenteeism, and that the two vertical (managerial) dimensions are relatively more important that the two horizontal dimensions. Least evidence is provided on the difference in relative importance between bridging and bonding WSC. Thus, H2 is only partly supported, since bridging WSC seems to be at least equally important for absenteeism as bonding WSC.

Replacing sickness absence days by sickness absence frequencies largely confirmed our findings (see Appendix A.3 for descriptive statistics, regression results and dominance analyses results for sickness absence frequency). A one-unit increase in overall WSC yielded an approximately 17% decrease in the number of times absent. For the single WSC dimensions, significant negative effects were again found for bridging, direct-leader-linking, and top-level-linking WSC. In the dominance analyses, the order of the WSC dimensions in explaining absence frequencies remained the same as for the number of absence days; there were only slight variations in the forms of dominance (less strong dominance of direct-leader-linking WSC over bridging WSC, yet stronger dominance of bridging WSC over bonding WSC). Focusing on long-term sickness absence, however, lead to different results (see Appendix A.4 for descriptive statistics, regression results and dominance analyses results for long-term sickness absence). In the regressions, only direct-leader-linking WSC significantly related to long-term sickness absence, while the other WSC dimensions did not have significant effects (ceteris paribus). The dominance analyses revealed that direct-leader-linking WSC is the most dominant dimension, followed by bonding, top-level-linking, and bridging WSC.

Discussion

The social dynamics of absenteeism: three WSC dimensions that matter

Sickness absenteeism is an outcome of perennial interest to public management research and practice because it produces high costs to organizations and ultimately harms organizational performance. Hence, contemporary public management faces the challenge of identifying levers to reduce absences among employees without significantly increasing costs to organizations. This study contributes to and extends the existing literature on sources of absenteeism in the public sector by theoretically elaborating the social dynamics of absenteeism and empirically demonstrating the pivotal role of WSC in reducing employee absences.

Our study shows that WSC has a substantial effect on employee sickness absence. It thereby corroborates Johns (1997) and Miraglia and Johns (2021) argument that the social dynamics at work need to be considered to better understand absenteeism. Theoretically, the effect of WSC on absenteeism can function through different, partially interdependent, theoretically founded paths (Miraglia and Johns 2021): WSC may affect resources (e.g. according to job demands resources theory; Bakker and Demerouti 2017), attitudes and emotions (Johns 1997), may constitute an important commodity for social exchanges in the workplace based on social exchange theory and the norm of reciprocity (Gouldner 1960; Blau 1964), and may affect adherence to attendance norms, a moral obligation to work, and a work ethic through several social influence theories (Festinger 1954; Salancik and Pfeffer 1978; Tajfel and Turner 1986; Hogg and Abrams 1988). As these pathways are interdependent, the overall analysis may only serve to corroborate their relevance in today’s workplaces and hence indicate to account for the social perspective at work.

Furthermore, we lacked knowledge about how different dimensions of WSC relate to absenteeism, to better understand the specific mechanisms through which the benefits of social capital materialize in an organization (Kroll, DeHart-Davis, and Vogel 2019). Our findings indicate that three of the four WSC types reduce absenteeism. We add new evidence that bridging WSC is an important factor for employee absenteeism; hence, our study indicates that ensuring shared goals, mutual recognition, and a functioning information flow across work units are of relevance to reduce employee absenteeism. We also provide new evidence on the beneficial effects of direct-leader-linking WSC on absenteeism, which confirms the, largely beneficial, existing evidence on related associations summarized in two meta-analyses (Dirks and Ferrin 2002; Montano et al. 2017) and single studies conducted in the public sector (e.g. Boyas and Wind 2010; Shahidul, Wright, and Yukl 2014; Ancarani et al. 2018; Chordiya, Sabharwal, and Battaglio 2019). Perceived supervisor support, mutual respect and recognition, and participatory leadership seem to affect employees’ absences, e.g. by providing crucial social resources and by achieving a mutually beneficial social exchange relation. Feeling valued by the immediate leader significantly affects employee motivation, including attendance motivation (Dirks and Ferrin 2002; Boyas and Wind 2010; Shahidul, Wright, and Yukl 2014; Ancarani et al. 2018). Similarly, positive relationships with the direct leader can increase employees’ ease at work and hence lower stress and stress-related health issues (Jana and Wallace 2002; Boyas and Wind 2010). Further, our results emphasize the importance of top-level-linking WSC. In line with social exchange and social influence theories, generating a sense of cohesion, organizational trust, and involving leadership on top management level seem to foster work engagement and well-being (Dirks and Ferrin 2002; Kurtessis et al. 2017; Meng, Clausen, and Borg 2018; Gross, Thaler, and Winter 2019), encourage pro-organizational behaviour, and the adherence to organizational adherence norms (Miraglia and Johns 2021), ultimately reducing employee absenteeism.

The insignificant coefficient of bonding WSC might be explained by its counteracting influences on absenteeism, particularly described by normative social influence theories. Bonding WSC may remove barriers and demands in work and thereby decrease stress and related absenteeism (Addae, Johns, and Boies 2013). However, social support in a group may have opposite effects and increase absenteeism if the normative expectations require the employee to stay home when sick or when more sympathetic colleagues provide replacement in case of sickness, facilitate the disclosure of illness, and remove barriers to reporting in sick (Dello Russo et al. 2013; Miraglia and Johns 2021). Therefore, the effect of bonding WSC on absenteeism may depend on the nature of the attendance culture in the group which can be both positive and negative towards absenteeism. The opposing social mechanisms will therefore equalize one another. In line with this, some scholars argue that there is an absence-prone culture in the (European) public sector (Mastekaasa 2020). Furthermore, these tendencies might be strengthened in organizations and/or countries where employee dismissal is elaborately procedural with several control mechanisms (Miraglia and Johns 2021) – which is the case in Danish municipalities. To conclude, bonding WSC probably provides employees with social resources, fosters positive work attitudes and buffers against stress and exhaustion. However, these beneficial effects might be nullified or equalized in contexts where attendance norms are rather permissive – as is supposed to be the case in the Danish public sector.

Our findings extend prior research on the role of social relations at work for employee absenteeism in the public sector. Most of the prior research has only focused on overall WSC (or related concepts) and has been conducted in the private sector (De Clercq et al. 2015; Schneider, Winter, and Schreyögg 2018) or focused more narrowly on public subsectors such as health or education (Török et al. 2018). Even though we do not make a formal comparison between public sector and private sector employees, the particular public context necessitates to not solely take over findings from other settings, but to generate unique evidence. The public setting as an employment context as well as the employees in this sector are clearly perceived – even stereotyped – as different from private sector employees (e.g. Goulet and Frank 2002; Becker and Connor 2005; Feeney and Bozeman 2009; Cooman et al. 2011; LeRoux and Feeney 2013). On the one hand, the widespread assumption – often made in scientific as well as popular literature – is that public servants are lazy, greedy, or corrupt, which is in strong contrast to recent empirical studies (e.g. Willems 2020; Boer 2020); on the other hand, public servants are attributed with different levels and types of motivation to perform the public services that they are involved in (van Loon et al. 2015). Against this background, absenteeism could either be seen as a manifestation of these stereotypical associations often made about public servants or could be a consequence of higher burn-out rates related to unfulfilled high levels of motivation. Hence, our study contributes to the evaluation of how overall absenteeism research does apply to this particular setting and this specific employee group. Nevertheless, further research could more actively integrate aspects of (self-)perceptions and motivations of public servants as additional factors explaining absenteeism, as well as differences with other sectors.

Beside this, our sensitivity analyses on long-term sickness absence indicate that the mechanisms which explain the importance of social relationships at work vary depending on whether short-term or long-term absences are in focus. In particular, the effect of WSC on attendance motivation will probably result in reductions of absenteeism of a couple of days and therefore has no effect on long-term sickness absence. Long-term absence instead might rely more on longer lasting effects due to mechanisms related to a general lack of resources, repeated negative emotions, and a depletion of work motivation rooted in the direct leader or the team, which will lead to permanent withdrawal behaviour or severe health issues that may cause long-term absences. Even though, to our knowledge, no study has yet assessed differences in the paths, prior research supports the idea that the effects of social relationships at work on long-term absenteeism differ from the effects on short-term absences (Kiss et al. 2014; Rugulies et al. 2016; Hansen et al. 2018).

Relative importance of WSC dimensions to absenteeism: setting priorities in research and practice

In addition, our findings provide new knowledge on the relative importance of the different dimensions of WSC. This is especially important knowledge for public managers since absence behaviour can emerge from several sources (Miraglia and Johns 2021), and it therefore is essential to identify which sources are the most important if managers want to reduce absenteeism in their organizations.

Specifically, we found that top-level-linking WSC has the highest relative importance of all four WSC types. Our results suggest that, for individuals to have a high attendance motivation, it is of utmost (relative) importance that the relationship to the overall organization is deemed positive; nuisances or distortions in the work unit or with the direct manager might be tolerated to some extent if the workplace in general is valued. This may also explain some of the insignificant results in previous studies (e.g. Kiss et al. 2014; Rugulies et al. 2016), which focused on aggregated measures of WSC only and thus cannot disentangle counteracting effects that might appear in situations where some types of WSC are low, whereas others are high. Several predictors with importance for absenteeism (e.g. employees’ perception of clear goals, meaning in work, overall management quality and employees’ organizational commitment (Labriola et al. 2006) are contained and fostered specifically in top-level-linking WSC. Therefore, it makes good sense that top-level-linking WSC is relatively most important in relation to absenteeism. It is, thus, important to separate the different types of WSC to gain a better understanding of the specific social dynamics of absenteeism in an organization. In addition, absenteeism indeed seems to be a very important lever for organizational-level reciprocity; perceptions of high top-level-linking WSC are reciprocated in high attendance motivation and hence low sickness absence. This may indicate that the social exchange theory – where absenteeism is seen as an exchange commodity in the relationship between employees and the overall organization – is particularly important to explain the effects of social factors on absenteeism (Dirks and Ferrin 2002; Miraglia and Johns 2021).

Although employees’ perception of high direct-leader-linking WSC is not as important as top-level-linking WSC, we found that it has higher relative importance for absenteeism than bonding and bridging WSC. This again provides evidence for the relevance of the social exchange paradigm and the idea that absenteeism is an important lever to reciprocate desired or undesired behaviours in the employee-leader relationship. The finding also indicates that the leader has a stronger effect on employee attitudes (e.g. attendance motivation) and social resources buffering against stress and exhaustion than colleagues. Combined, the results for top-level-linking and direct-leader-linking WSC are especially interesting since they suggest that vertical ties – the managerial relationships – are the most important social relationships in an organization in relation to public employees’ absenteeism.

Limitations and further research

Like any study, this study has some limitations. To start with, the WSC dimensions only explain a small part of variation in absenteeism. This was expected since not all sickness absence can or should be avoided and there are numerous determinants of absenteeism besides social dynamics (Johns 1997; Lohaus and Habermann 2019; Miraglia and Johns 2021). Furthermore, as we observe changes in variables in a longitudinal design, small coefficients are common; in our case, they nonetheless indicate that WSC is a significant and thus valuable lever for management to affect employees’ absenteeism. Second, we measured WSC at two points of time and aggregated sickness absence per year, while effects might be more pronounced on smaller time frames. Research could, for example, use qualitative diary methods to analyse more fine-grained variations in the changing perceptions of WSC and its’ impact on different patterns of absences behaviour over time. This would add to our understanding on how changes in workplace relationships evolve, how long it takes to see changes in absences, and how long these effects endure. Third, while we derived hypotheses on the potential mechanisms through which the WSC dimensions relate to sickness absence, we were unable to test the specific mediated pathways. It would be interesting to further disentangle and compare the effects of WSC on health and attendance motivation through the varying theoretical paths. However, as the paths are highly interdependent, this will probably be impossible to separate in quantitative studies. Thus, qualitative investigations of how different psychosocial mechanisms within the organizational relationships (the WSC types) affect health and absenteeism behaviour would provide us with a stronger theoretical understanding of the importance of the pathways and how they depend on different work contexts. Fourth, our sample was restricted to Danish public sector employees. In particular our finding on bonding WSC might be the result of socialization processes among public servants within organizational and legal settings specific for this public sector. Future studies could examine if some of the effects are moderated by different cultural contexts and investigate under which conditions bonding WSC supports an absence-prone environment or reduces absenteeism. Finally, this paper focused solely on social relationships as a means to reduce absenteeism, which might also embed a choice of the employee to go to work while sick (sickness presenteeism) (Miraglia and Johns 2016). Future research should investigate the role of the WSC dimensions with regards to sickness presenteeism, to ensure that presenteeism does not reach levels which harm organizational performance.

Conclusion and practical implications

In this study, we asked how managers may reduce a pressing problem in the public sector, namely employee absenteeism. To contribute to this question, we focused on a social perspective of absenteeism. We added to the public management literature by providing a nuanced understanding of how various organizational relationships (in a coherent framework of WSC) affected public employees’ absenteeism. We disentangled four types of WSC and scrutinized their effects on and relative importance to absenteeism using a large panel of public employees nested in their relevant organizational units. Our results have clear relevance to public managers since a better understanding of how employees’ perceptions of different organizational relationships influence absenteeism may help managers to plan better targeted interventions and steer absenteeism by supporting and shaping the most relevant relationships. We have shown that three types of WSC (bridging, direct-leader-linking, and top-level-linking WSC) are important in reducing public employee absenteeism. Yet, employees’ perceptions of the vertical relationships to the top-level management (top-level-linking WSC) and to the immediate leader (direct-leader-linking WSC) are more important than the horizontal relationships to other working groups (bridging WSC). Thus, public managers should invest in top-level-linking WSC by creating a strong perceived connection to the workplace among the employees, generating a common understanding between management and employees about how to perform the work tasks, and ensuring employee involvement in managerial decisions. Similarly, investing in high direct-leader-linking WSC seems beneficial for reducing employee absenteeism. Direct-leader-linking WSC may be strengthened if managers set aside time to obtain a deeper understanding of the work conducted in the units, help solve specific problems in the daily work, and openly communicate how they consider the employees’ needs and opinions when making decisions. Overall, our results underline how important it is that public managers shape organizational relationships in practice since our analyses clearly point to the conclusion that the social relationships to the management matter – a lot – in reducing absenteeism.

Author contribution statement

All authors have made substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work; and to drafting the work or revising it critically for important intellectual content. They have given final approval of the version to be published.

Ethical approval and consent

The study was approved by The Scientific Ethics Committees for the Region of Southern Denmark47 (no.20192000–40). All employees of the municipality are informed about the project and that the data collected in employee surveys can be used for statistical analyses of scientific and public interest. Therefore, it is considered as implied consent, when the employees choose to complete the various questionnaires. Participation is strictly voluntary and all data are analysed on aggregate level.

Acknowledgements

The authors thank the Esbjerg project group for their support in data mining, analyses, and feedback on prior versions of this work and Raphaele Bartels for assistance in sickness absence data preparation, data analyses, and manuscript editing.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplemental data

Supplemental data for this article can be accessed at https://doi.org/10.1080/14719037.2022.2142652

Data availability statement

Since this study contains highly sensible personal information, it is protected by the GDPR rules and follows demands on total confidentiality from the municipality. Thus, to comply with GDPR and due to the agreement with Esbjerg Municipality, data access is restricted to researchers affiliated with SDU, who have signed the cooperation agreement and code of conduct statement. Hence, public access to this dataset is not possible. The data are stored, processed and analysed according to newest safety and security standards.

Additional information

Funding

The project was funded by The Danish Working Environment Research Fund (J. nr. 20185100800). The fund did not influence the study design the future work with the data, analyses, or writing.