Span of control and the significance for public sector managers’ job demands: A multilevel study

Abstract

The aim of this article is to investigate how span of control influences seven job demands common amongst operational public sector managers. Self-administrated surveys were collected from 434 managers organized in 37 municipal operations, i.e. management teams in a set of public organizations in Sweden, and complemented with register data. Multilevel analyses were performed and it was shown unfavorable to have a large number of subordinates for the majority of job demands investigated. Furthermore, managers who were members of management teams with a higher average span of control experience some job demands to an even higher extent. Thus, the results of this study demonstrate that span of control is a key upstream component of managers’ job demands and emphasize the value of a reasonable number of subordinates.

Keywords

Management organization public sector span of control working conditions

Introduction

In order to facilitate healthy work for managers, understanding of relevant structural conditions that shape working conditions is essential (Härenstam, 2008, 2009). The number of employees that the manager is responsible for is one such important structural condition and in order to create good prerequisites for management a functional organization with a reasonable number of subordinates per manager is of importance (Andersson-Felé, 2008). However, since the 1990s financial difficulties and fewer resources (Westerberg and Armelius, 2000) have increased the demands on the public sector in Sweden to be more cost-effective and still be able to offer services of high quality (Montin, 2004). New Public Management (NPM), i.e. introducing economic and managerial ideas from private companies into public organizations (Hood, 1991), has often been launched in a time of such financial crisis accompanied by downsizing (Bach and Della Rocca, 2000). The organizational changes in the public sector in Sweden have generally been made in line with NPM (Montin, 2004), which has major consequences for those employed in the public sector (Chandler et al., 2002; Farrell and Morris, 2003). In Sweden, NPM has for instance resulted in reduced resources and downsizing leading to, for example, wider spans of control (SOCs), a larger the number of subordinates per manager and increased job demands for many public sector managers (Harder et al., 2000; Ohlsson and Rombach, 1998; Thylefors, 2007). Furthermore, the SOCs are often wider and the job demands higher compared to corresponding services in the private sector (Höckertin, 2007). Hence, it is of great significance to establish how this wider SOC affects public sector managers’ job demands thereby addressing a gap in the work psychology literature regarding the origins of psychosocial working conditions as pointed out, for example, by Dollard and Bakker (2010).

Although some studies highlight the importance of managers’ SOC, i.e. the number of subordinates per manager (as defined by e.g. Gulick et al., 1937; Lucas et al., 2008; Mintzberg, 1993; Schyns et al., 2010), the research on SOC is relatively sparse and often lacks a solid empirical foundation. Furthermore, little attention has been paid to the implications of SOCs for managers and their working conditions. A central issue in early SOC research has instead been the optimal SOC for e.g. performance showing an optimal range from three to 30 subordinates per manager (e.g. Bell, 1967; Gracunius, 1937; Gulick et al., 1937; Urwick, 1956). However, several studies have shown that the optimal SOC varies between organizations, at different levels in organizations (Meier and Bohte, 2000; Mintzberg, 1993; Simon, 1946), with the most important organizational goal at hand (Theobald and Nicholson-Crotty, 2005) and structural conditions such as diversity of function amongst subordinates, employee- and environment stability as well as geographical space and size of the organization (Gulick et al., 1937; Meier and Bohte, 2003). More recent research has however changed focus to investigate the influence of SOC on different outcomes. SOC affects the interaction between managers and subordinates (Brazier, 2005; Lucas et al., 2008) and large SOCs can degrade the quality of exchange between the two levels (Brazier, 2005; Green et al., 1996). Group processes (Gittell, 2001), managers’ ability to empower (Lucas et al., 2008) and support their subordinates (Harder et al., 2000; Hjalmarsson et al., 2004; Lucas et al., 2008) are negatively affected by a wide SOC as well as the effects of positive leadership styles (Dorian et al., 2004; McCutcheon, 2004). Negative effects amongst employees with managers with wider SOCs include increased employee turnover and intention to quit (Burke, 1996; McCutcheon, 2004), reduced employee engagement and commitment (Burke, 1996; Cathcart et al., 2004; Green et al., 1996), less monitoring of health and safety issues (Harder et al., 2000) and lowered safety performance (Hechanova-Alampay and Beehr, 2001) as well as lowered organizational performance (Bohte and Meier, 2001; Meier and Bohte, 2000). Thus, the more recent studies have mainly pointed to the influence of managers’ SOC on the quality of subordinates’ working conditions. One could however expect that managers’ SOC also affects the managers’ own working conditions mainly in the form of the wide array of job demands present in their work. Understanding where the job demands experienced by managers originate from can be of upmost importance since those demands for instance may lead to health problems, burnout (Demerouti et al., 2001) as well as lowered work engagement (Hu et al., 2011; Llorens et al., 2006). Such effects are causing negative organizational outcomes, for example turnover intentions and lack of organizational commitment (Hu et al., 2011).

Theoretical background

This study is based on and expands the Job Demands–Resources (JD–R) model (Figure 1) with a focus on managers’ SOC and job demands. The JD–R model is both comprehensive and applicable in a wide variety of occupational settings characterized by various demands and resources (Bakker and Demerouti, 2007; Bakker et al., 2003, 2010). In this model Bakker et al. (2004: 86) have defined job demands as ‘those physical, psychological, social, or organizational aspects of the job that require sustained physical and/or psychological (cognitive and emotional) effort and are therefore associated with certain physiological and/or psychological costs’, e.g. physical workload, role overload, emotionally demanding recipient contacts and time pressure (Bakker and Demerouti, 2007; De Jonge et al., 1999; Demerouti et al., 2001; Söderfeldt et al., 1996). As demands represent the possible stressor side of working conditions (Demerouti et al., 2001), the JD–R model refers to resources at work as ‘those physical, psychological, social, or organizational aspects of the job that are (1) functional in achieving work goals; (2) reduce job demands and the associated physiological and psychological costs; or (3) stimulate personal growth and development’ (Bakker et al., 2004: 86), e.g. feedback, control, supervisory support, rewards, role clarity and participation (Bakker et al., 2003; Demerouti et al., 2001). The two types of working conditions forming the JD–R model are believed to cause two separate psychological processes. High demands cause health problems and burnout through a health-impairment process (Demerouti et al., 2001) whilst resources facilitate the achievement of objectives and thus increase engagement and commitment through a motivational process (Bakker et al., 2003). The health-impairment and motivational processes have gained considerable support (e.g. Hu et al., 2011; Korunka et al., 2009; Llorens et al., 2006; Schaufeli and Bakker, 2004) where the health- impairment process has been shown to be more persuasive (Schaufeli and Bakker, 2004). More recent studies have also found moderating effects between these two processes where resources for instance have been shown to buffer high demands and thus reduce health problems (Bakker et al., 2005; Llorens et al., 2006) while high demands in combination with high levels of resources result in higher levels of work engagement (Bakker and Demerouti, 2007). However, the moderating effects have usually received mixed statistical support (e.g. Bakker et al., 2004, 2005; Hu et al., 2011; Xanthopoulou et al., 2007).

Figure 1.

The Job Demands–Resources model – modified from Demerouti and Bakker (2011). The focus of the present study is on upstream structural conditions for job demands (highlighted in the grey boxes). The job demands studied are work overload, conflicting logics, excessive role demands, buffer problems, container function, group problems and individual employee problems.

When focusing on managers, especially public sector managers, it is necessary to investigate such multitude of aspects of job demands and job resources as the JD–R model implies (Van den Broeck et al., 2008). Several researchers further highlight that a wider range of job demands must be taken into account (De Jonge et al., 1999; Peeters and Le Blanc, 2001; Schnall et al., 1994; Schwartz et al., 1996; Söderfeldt et al., 1996). Although organizations in the public sector generally have been found to have many similarities with the private sector, they also face some unique prerequisites (Boyne, 2002; Rainey and Bozeman, 2000; Rosenberg Hansen and Villadsen, 2010). They have, for example, a multitude of stakeholders as the organizations are governed by a political level (Westerberg and Armelius, 2000) and a special relation to their recipients (Hasenfeld, 1983). This implies a wider, and perhaps different, range of job demands, something that has been argued to be negative for managers’ health and well-being in the public sector (Skagert et al., 2008). However, in practice, managerial work in diverse parts of the public sector, e.g. human services and technical services, differs, where human service managers often tend to have worse structural conditions (Forsberg Kankkunen, 2006) as well as less favorable working conditions (Berntson et al., 2012). Consequently, wider SOCs may constitute one important explanation as to why managers in human services tend to experience less favorable working conditions than managers in technical services. The following manager-specific job demands, which have been emphasized as especially important for Swedish public sector managers (e.g. Dellve and Wikström, 2006; Dellve et al., 2008; Skagert et al., 2004, 2008), are addressed in this study.

Work overload

The public manager’s work is usually hectic (Abma and Noordegraaf, 2003) with an excessive workload (Lindholm et al., 2004; Skagert et al., 2004). The workload among managers in the public sector, particularly first-line managers with many subordinates, seems to have increased (Harder et al., 2000) and public services engaged in care and education, i.e. human services, where female managers often are overrepresented, tend to have a even higher workload. This could in part be a consequence of female managers having significantly larger SOCs than men in general (Westerberg and Armelius, 2000).

Conflicting logics

Public managers are often exposed to conflicting and increasing expectations (Forsberg Kankkunen, 2006; Skagert et al., 2004; Thylefors, 2007) and thus experience difficulties balancing different logics or tasks present in their work such as administrative, strategic and personnel-related tasks, i.e. conflicting logics (Wiktröm and Dellve, 2009). The increased administrative burden among public sector managers (Dellve and Wikström, 2006; Skagert et al., 2008) has been proposed as one possible explanation for conflicting logics in their work (Skagert et al., 2004; Wikström and Dellve, 2009) and a wider SOC may be one of the reasons why the administrative burden has increased in the first place (Forsberg Kankkunen, 2006). A wider SOC also increases the amount of personnel-related tasks, further adding to possible conflicting logics in managers’ work (Andersson-Felé, 2008; Höckertin, 2007).

Excessive role demands

The managerial role includes a broad variety of tasks and responsibilities with a high degree of complexity (Boyne, 2002; Mumford et al., 2000; Rosenberg Hansen and Villadsen, 2010) and fragmentation between different tasks (Arman et al., 2009). Moreover, the risk of ambiguity concerning the managerial role and its responsibilities may be greater in politically controlled organizations (Pandey and Wright, 2006; Skagert et al., 2004) and changes in line with NPM seem to have made the managerial role in the public sector even more complex (Christensen, 2001). In addition, the large SOCs often present in the public sector have shown to cause the natural tasks and demands in managers’ work to become excessive (Andersson-Felé, 2008).

Buffer problems

Being a manager includes dealing with relationships with employees, superiors and clients (Moynian and Pandey, 2007). Managing the tension between the various stakeholders’ needs and demands comprises a main challenge for public sector managers (e.g. Boyne, 2002; Hjalmarson et al., 2004; Pandey and Wright, 2006; Peeters and Le Blanc, 2001) and this balancing act may give rise to buffer problems (Skagert et al., 2008). Gaining influence to balance the requirements, obtain resources for the operations while creating and maintaining legitimacy from superiors and subordinates (Dellve and Wikström, 2006) has proven to be a difficult task for public sector managers (Dellve and Wikström, 2006; Skagert et al., 2004). Therefore, operational managers at lower levels in the organization may act as ‘shock absorbers’ or buffers in strategies to reduce the load on their own staff from demands from above. These strategies may be beneficial for subordinates’ psychosocial work environment but not for the managers themselves (Skagert et al., 2008) and it may be reasonable to assume that a wider SOC might increase problems related to the buffer role.

Group problems and individual employee problems

Not being able to trust and having to deal with less motivated subordinates have been suggested to give rise to stress among public sector managers (Skagert et al., 2008). Communication, which tends to be facilitated in smaller work groups, leads to trust between employees and management and generates happier and more committed employees (Burke, 1996). However, managing a wider SOC usually leads to more distance between the two levels and thus fewer opportunities to communicate. This lack of communication is not only negative in itself but can further inhibit communication within the group, weakening group processes, causing, for instance, lowered problem solving and thus poorer group performance (Gittell, 2001). In addition, managers with narrower SOCs interact more with individual group members (e.g. Gittell, 2001; Harder et al., 2000; Lucas et al., 2008). Hence, wider SOCs usually lead to subordinates feeling unseen and managers insufficient (Ohlsson and Rombach, 1998).

Taken together, the available data indicate, but do not prove, that SOC is of structural importance for managers’ job demands. Our main aim is therefore to establish how SOC, as one of several possible structural origins of working conditions, affects operational public sector managers’ job demands, combining self-reported job demands and register data on SOC. Additionally, based on previous results showing that managers in technical services commonly have smaller SOCs and more favorable working conditions than managers in human services (care of the disabled and elderly, pre-school/compulsory school and upper secondary school), this study further examines whether there are diverse effects of SOC on managers’ job demands in these different types of municipal services. Thus, the following research questions were empirically investigated: (1) How does span of control influence the levels of job demands that operational public sector managers experience at work? (2) Are there diverse effects of span of control in different types of municipal services?

Methods

Study setting, sample and procedure

The present study was performed as a part of a larger project exploring organizational prerequisites for managers in the Swedish public sector. This study is based on cross-sectional questionnaire data and register data collected in spring 2009. The questionnaire was approved by the regional ethical committee of Gothenburg, Sweden (Dnr 048-09).

A two-stage sampling procedure was used based on previous local government research highlighting the significance of size and the political and demographical composition of municipalities in order to explain social science phenomena (Szücs and Strömberg, 2006). Furthermore, since the Swedish public sector is gender segregated, with women mostly working in human service organizations, a stratified sample of organizations in both male- and female-dominated services was obtained. First, seven municipalities in the western region of Sweden were chosen. The municipalities are divided into departments which consist of several different non-private municipal operations engaged in different types of services. Human services, care of the disabled and elderly, pre-school/compulsory school, upper secondary school and technical services were included. These municipal operations are further divided into units. The municipal operations are governed by a middle manager and the units belonging to the operation are governed by first-line managers. The managers in the respective municipal operations constitute the management teams studied. Altogether, 28 departments and 37 municipal operations, i.e. management teams within these departments, were chosen and the questionnaire was administrated to all 766 managers within these departments via contact persons in the organizations. The questionnaires were returned to the researchers by mail, response rate 72% (n = 555). Five main categories of managers answered the questionnaire, including heads of department, middle managers, first-line managers, team leaders and heads of section (e.g. heads of HR and IT). For the purpose of the present study the operational managers, i.e. first-line and middle managers, of the sample were selected (n = 442). This choice was made since heads of departments and sections often are responsible for more than one municipal operation making it difficult to decide their organizational belonging, and in the present study managers who work under the same conditions in the same context were sought.

We define a manager as a person with a formal position including personnel, economic and operational responsibility. Of the selected operational managers, 441 had subordinates and thus qualified for further analysis whereof 405 managers had complete information records. After removing seven cases with too few responses (50% or fewer), data were imputed using the EM algorithm in PASW Statistics 18. The imputation procedures were carried out following the recommendations by Schafer and Graham (2002) and Little and Rubin (1987). Imputation was made for the job demands separately and missing data concerning gender, managerial position, managerial experience and SOC were obtained from records. All in all, 0.3% of the values in the job demands were imputed. Univariate and multivariate outlier analyses were performed for dependent and independent variables respectively and three cases were detected as univariate outliers and two cases as multivariate outliers. After close inspection these extreme observations were not considered unrepresentative of the population and were therefore kept (Hair et al., 1995). Thus, the effective final sample consisted of 434 managers.

Of the 434 operational managers included in the analysis, 16.6% were middle managers (n = 72) consisting of 50% women and 50% men. Thus, the largest part of the sample was comprised of first-line managers, 83.4% (n = 362) of whom 264 (72.9%) were women. The number of managers in the 37 municipal operations, i.e. management teams, varied from three to 37 managers (mean = 11.7). In the technical services the management teams were dominated by male managers while the management teams in the human services were mainly dominated by women. The management teams in upper secondary schools were however mixed to a higher extent than the others. The management teams have regular team meetings and the majority of them meet every two weeks. The meetings are characterized mainly by dialogue and revolve around common concerns primarily regarding operational- and financial issues.

Measures

Span of control

The structural condition SOC was assessed by means of a continuous variable. The numbers were validated using registers from their respective human resources department. Since an ICC(1,1) value over 0.05 can be considered relevant for aggregation (Hox, 2002) the individual-level SOC ICC(1,1) = 0.36 was aggregated in order to represent a shared group construct for each municipal operation, i.e. management team. The significant variation at the organizational level indicates that the SOC mean for each management team can be considered a level two variable (Enders and Tofighi, 2007; Paccagnella, 2006). This variable will be referred to as the average SOC in the article. To account for the variation in SOC for each manager, i.e. individual SOC, the raw score was also included in the analysis.

A total of 32 items, combined into seven indexes, reflecting job demands common to managers in the public sector were included.¹ The job demand measures were based on the Gothenburg Manager Stress Instrument (Eklöf et al., 2010) that was developed from several qualitative research studies in Swedish human service organizations (Dellve and Wikström, 2006; Dellve et al., 2008; Skagert et al., 2004, 2008). Exploratory and confirmatory factor analysis was performed regarding the seven dimensions indicating discriminant as well as content validity (Eklöf et al., 2010).

Job demands

Work overload was measured with a four-item index (alpha = .85) reflecting a demand of having too much to do and difficulties keeping up (e.g. ‘That you actually do not have time to do everything you feel that you should do at work’). Conflicting logics were assessed with a five-item index (alpha = .85) reflecting conflict between different types of tasks included in managerial work (e.g. ‘That there are conflicts between administrative work, service development and employee contact’). Excessive role demands were measured with a six-item index (alpha = .85) concerning demands that are naturally occurring in the managerial role but the overall experience was that the role demands were excessive (e.g. ‘That the responsibility for subordinates is excessive’). Buffer problems were measured with a four-item scale (alpha = .75) involving the difficulties that can arise when managers have to balance demands and expectations upward and downward in the organization (e.g. ‘That you must be a buffer between higher levels in the organization and your employees’). Container function was assessed by a two-item scale (alpha = .82) mirroring managers as receivers or containers of subordinates’ frustration in pressured situations (e.g. ‘That pressured employees burden you with their problems’). Group problems were assessed with a seven-item scale (alpha = .82) regarding problems associated with the subordinate work groups’ performance and function (e.g. ‘That cooperation is problematic or there are conflicts between employees’). Individual employee problems were measured with a six-item index (alpha = .73) reflecting problems with individual subordinates’ performance and motivation (e.g. ‘That you have to help employees plan and organize their work’). Response alternatives ranged from 1 (never/almost never) to 5 (always/almost always) for all job demand items.

Control variables

Three items were included in the analysis as possible confounders, in order to control for compositional effect. Gender was chosen because of its relation to working conditions and its variability between municipal operations. Several studies have shown that women have fewer beneficial working conditions than men. Female managers constituted the reference category since a majority of the respondents were women. Managerial position was assessed with two categories of operational managers, first-line and middle managers, and included first-line managers as a reference since they constituted the majority of respondents. Managerial experience was measured as the number of years as a manager.

Analytical strategy

When participants are nested in groups the key assumption of independence in observations of many standard statistical procedures, e.g. OLS regression, is violated (Glick, 1985; Goldstein et al., 2002; Hox, 2002; Snijders and Bosker, 1999; Tabachnick and Fidell, 2007). When observations are dependent, i.e. hierarchically ordered or nested, multilevel modeling (MLM) should be used. Managers are nested within operations and therefore influenced by the organizational context (Hox, 2002; Snijders and Bosker, 1999). Ignoring this fact can lead to incorrect inferences since it causes an underestimation of the standard errors of regression coefficients and thus infers that relationships between variables exist when they in fact do not (Rasbash et al., 2009). MLM further has the ability to capture much of the nested complexity in organizational life (Klein and Kozlowski, 2000) and brings contextual factors into consideration and can thus more accurately assess what is due to the individual and the context respectively (Härenstam, 2009).

The 37 municipal operations, i.e. management teams, were set as level two. This was the organizational level where the main part of the level two variance was found and thus the main source of second-level variability. The 434 managers constituting these management teams formed the first level. The similarity between the micro units belonging to the same macro unit was calculated by using the ‘intraclass correlation coefficient’ ICC(1,1), i.e. the amount of dependence between the observations (Hox, 2002; Schrout and Fleiss, 1979) or the amount of variance attributable to the group or macro level, which in this case is made up by management teams (Snijders and Bosker, 1999). Management team will hereafter be used to denote the group level throughout the text.

Data analysis

Prior to analysis the variables were examined for accuracy of data entry, missing values as well as assumptions for multivariate analysis and all the job demands were standardized in order to facilitate comparison between them. The assumption of normal distribution of first- and second-level residuals was valid and checked by inspection of the normal plots.

The multilevel analyses were performed in sequential steps, with the main variables of interest added in the first steps in line with recommendations made by Spector and Brannick (2011). First, a random intercept model was estimated. In the following models, predictor variables at both the individual and management-team level were added and grand mean centered in accordance with Wu and Wooldridge (2005). This approach takes into account that estimates of organization-level effects should control for confounding due to individual-level effects (Wu and Wooldridge, 2005). In addition, centered predictor variables are comparable across different models (Hox, 2002), which helps to reduce collinearity between levels (Wu and Wooldridge, 2005) as well as establish a zero point on the scales that otherwise lack such a value (Enders and Tofighi, 2007; Hox, 2002; Wu and Wooldridge, 2005). The multilevel analysis was implemented with the MLwiN 2.1 software using the restricted maximum likelihood (RIGLS) for empty models and iterative generalized least squares (IGLS) estimation method in the following steps, described below:

Step 1: Empty model analysis (Model 1). In accordance with Hox (2002), an empty two-level model was initially fitted for each of the job demands; estimating the organizational and individual variation respectively. These results were used as the basis for the ICC(1,1) calculations as well as for significance testing of the level two variance. This was done by comparing model fit through Δloglikelihood for the one-level and two-level model (Rasbash et al., 2009) (Table 1). Furthermore, the empty model served as a benchmark to which the other models could be compared (Hox, 2002).

Step 2: Adding the predictors (Models 2–3). In order to explain the differences that were found between the management teams, workplace predictors were included in the model as the next step. SOC as an aggregated level two variable was added followed by the grand-mean centered raw scores for SOC (Tables 2 –8).

Step 3: Adding control variables (Models 4–6). Next, confounders at the individual level controlling for compositional effects were entered stepwise in order to explore their separate effects and secure the relationship between SOC and the job demands (Tables 2 –8).

Step 4: Investigating possible interactions between SOC and type of service. To allow for the effect of SOC to vary across types of services this variable and interaction terms were added to the model. The interaction terms were, in accordance with Rasbash et al. (2009), created by multiplying SOC with each type of service and a separate linear regression was thus fitted to each of them in order to test whether SOC had differing impacts on job demands in different types of municipal services.

Table 1.

Do operational public sector managers in the same municipal operation, i.e. management team, experience similar job demands? Empty models showing the amount of organizational variance ICC(1,1) for the seven job demands included in the study (Model 1).

Job demand	Work overload		Conflicting logics		Excessive role demands		Buffer problems		Container function		Group problems		Individual employee problems
Empty models	b	SE	b	SE	b	SE	b	SE	b	SE	b	SE	b	SE
Intercept	0.008	0.057	−0.007	0.054	−0.008	0.074	−0.020	0.073	−0.034	0.076	−0.001	0.062	−0.010	0.058
Variance components
Individual (1)	0.971	0.068	0.980	0.069	0.902	0.064	0.903	0.064	0.892	0.063	0.951	0.067	0.967	0.068
Organizational (2)	0.031	0.026	0.020	0.023	0.107	0.045	0.104	0.044	0.120	0.048	0.051	0.032	0.032	0.027
ICC(1,1)	3.09%		2. 00%		10.60%		10.33%		11.86%		5.09%		3.20%
N	434		434		434		434		434		434		434
Loglikelihood	1229.103		1229.517		1215.959		1215.743		1213.625		1225.600		1227.751
ΔLoglikelihood one level/two level model	1.536 (p = .2152)		1.122 (p = .2895)		14.68* (p = .0001)		14.896* (p = .0001)		17.014* (p = .0000)		5.039* (p = .0248)		2.888 (p = .0892)

Notes: The estimated intercepts in the empty models shows the average job demand levels across all management teams and managers. The ICC(1,1) shows how much variance of the job demands is at the management-team level. The Δloglikelihood one level/two level model compares the model fit for the one level and two level model to test significant variability between management teams (level two).

p < .05 indicating significant variability between management teams.

Table 2.

Span of control and work overload. Multilevel models for the job demand work overload with span of control (SOC) as a main predictor and compositional variables included.

Work overload	Model 2SOC (2)		Model 3SOC (1)		Models 4–6Compositional effect (1)
Variable (1 = level one; 2 = level two)	b	SE	b	SE	b	SE
SOC (2)	0.010	0.006	0.000	0.007	−0.002	0.008
SOC (1)			0.009*	0.004	0.011*	0.004
Gender (1)
Woman					Ref.
Man					−0.168	0.108
Managerial position (1)
First-line manager					Ref.
Middle manager					0.217	0.141
Managerial experience (1)					−0.021*	0.005
Intercept	0.009	0.055	0.010	0.055	0.030	0.072
Variance components
Individual (1)	0.969	0.068	0.956	0.067	0.903	0.064
Organizational (2)	0.022	0.024	0.023	0.024	0.043	0.029
N	434		434		434
Loglikelihood	1226.532		1220.841		1202.885

Notes: The numbers in the parentheses indicate if the variable is at the first or second level.

SOC (2) is the average span of control at the management-team level and SOC (1) the managers’ individual span of control.

p < .05.

Table 3.

Span of control and conflicting logics. Multilevel models for the job demand conflicting logics with span of control (SOC) as a main predictor and compositional variables included.

Conflicting logics	Model 2SOC (2)		Model 3SOC (1)		Models 4–6Compositional effects (1)
Variable (1 = level one; 2 = level two)	b	SE	b	SE	b	SE
SOC (2)	0.019*	0.005	0.001	0.007	−0.002	0.007
SOC (1)			0.017*	0.004	0.015*	0.004
Gender (1)
Woman					Ref.
Man					−0.141	0.104
Managerial position (1)
First-line manager					Ref.
Middle manager					−0.173	0.137
Managerial experience (1)					−0.019*	0.005
Intercept	−0.001	0.048	−0.001	0.048	0.072	0.059
Variance components
Individual (1)	0.969	0.068	0.922	0.065	0.880	0.062
Organizational (2)	0.002	0.018	0.004	0.017	0.002	0.016
N	434		434		434
Loglikelihood	1230.637		1229.503		1195.318

Notes: The numbers in the parentheses indicate if the variable is at the first or second level. SOC (2) is the average span of control at the management-team level and SOC (1) the managers’ individual span of control.

p < .05.

Table 4.

Span of control and excessive role demands. Multilevel models for the job demand excessive role demands with span of control (SOC) as a main predictor and compositional variables included.

Excessive role demands	Model 2SOC (2)		Model 3SOC (1)		Models 4–6Compositional effects (1)
Variable (1 = level one; 2 = level two)	b	SE	b	SE	b	SE
SOC (2)	0.029*	0.006	0.016*	0.007	0.016*	0.007
SOC (1)			0.013*	0.004	0.011*	0.004
Gender (1)
Woman					Ref.
Man					0.044	0.106
Managerial position (1)
First-line manager					Ref.
Middle manager					−0.155	0.138
Managerial experience (1)					−0.007	0.005
Intercept	−0.001	0.055	−0.001	0.055	0.010	0.068
Variance components
Individual (1)	0.905	0.064	0.879	0.062	0.869	0.061
Organizational (2)	0.027	0.024	0.029	0.024	0.032	0.025
N	434		434		434
Loglikelihood	1230.637		1215.943		1187.439

p < .05.

Table 5.

Span of control and buffer problems. Multilevel models for the job demand buffer problems with span of control (SOC) as a main predictor and compositional variables included.

Buffer problems	Model 2SOC (2)		Model 3SOC (1)		Models 4–6Compositional effects (1)
Variable (1 = level one; 2 = level two)	b	SE	b	SE	b	SE
SOC (2)	0.017*	0.007	0.008	0.008	0.003	0.008
SOC (1)			0.008*	0.004	0.005	0.004
Gender (1)
Woman					Ref.
Man					−0.247*	0.107
Managerial position (1)
First-line manager					Ref.
Middle manager					−0.397*	0.138
Managerial experience (1)					−0.002	0.005
Intercept	−0.011	0.067	−0.011	0.067	0.133	0.075
Variance components
Individual (1)	0.904	0.064	0.892	0.063	0.865	0.061
Organizational (2)	0.073	0.037	0.074	0.037	0.063	0.033
N	434		434		434
Loglikelihood	1210.688		1205.661		1189.943

p < .05.

Table 6.

Span of control and container function. Multilevel models for the job demand container function with span of control (SOC) as a main predictor and compositional variables included.

Container function	Model 2 SOC (2)		Model 3 SOC (1)		Model 4–6 Compositional effects (1)
Variable (1 = level one; 2= level two)	b	SE	b	SE	b	SE
SOC (2)	0.034*	0.005	0.023*	0.007	0.017*	0.006
SOC (1)			0.011*	0.004	0.011*	0.004
Gender (1)
Woman					Ref.
Man					−0.318*	0.103
Managerial position (1)
First-line manager					Ref.
Middle manager					−0.010	0.136
Managerial experience (1)					−0.006	0.005
Intercept	−0.003	0.048	−0.004	0.048	0.100	0.058
Variance components
Individual (1)	0.899	0.063	0.878	0.062	0.864	0.059
Organizational (2)	0.007	0.018	0.009	0.018	0.000	0.000
N	434	434	434
Loglikelihood	1188.801	1179.368	1167.976

p < .05.

Table 7.

Span of control and group problems. Multilevel models for the job demand group problems with span of control (SOC) as a main predictor and compositional variables included.

Group problems	Model 2SOC (2)		Model 3SOC (1)		Models 4–6Compositional effects (1)
Variable (1 = level one; 2 = level two)	b	SE	b	SE	b	SE
SOC (2)	0.012*	0.006	0.000	0.007	−0.003	0.008
SOC (1)			0.012*	0.004	0.009*	0.004
Gender (1)
Woman					Ref.
Man					−0.178	0.108
Managerial position (1)
First-line manager					Ref.
Middle manager					−0.256	0.141
Managerial experience (1)					−0.004	0.005
Intercept	0.002	0.058	0.002	0.058	0.100	0.070
Variance components
Individual (1)	0.951	0.067	0.929	0.065	0.912	0.064
Organizational (2)	0.034	0.027	0.036	0.027	0.035	0.026
N	434		434		434
Loglikelihood	1222.139		1213.002		1204.676

p < .05.

Table 8.

Span of control and individual employee problems. Multilevel models for the job demand individual employee problems with span of control (SOC) as a main predictor and compositional variables included.

Individual employee problems	Model 2SOC (2)		Model 3SOC (1)		Models 4–6Compositional effects (1)
Variable (1 = level one; 2 = level two)	b	SE	b	SE	b	SE
SOC (2)	−0.002	0.006	−0.009	0.007	−0.013	0.007
SOC (1)			0.007	0.004	0.007	0.004
Gender (1)
Woman					Ref.
Man					−0.200	0.110
Managerial position (1)
First-line manager					Ref.
Middle manager					−0.008	0.143
Managerial experience (1)					−0.009	0.005
Intercept	−0.01	0.057	−0.01	0.057	0.054	0.068
Variance components
Individual (1)	0.967	0.068	0.960	0.068	0.948	0.067
Organizational (2)	0.028	0.025	0.029	0.025	0.024	0.024
N	434		434		434
Loglikelihood	1230.64		1227.737		1220.721

p < .05.

Results

Span of control in the studied organizations

Mean SOC was 27.4 (SD = 14.99) and ranged from two to 105 and two to 50 subordinates amongst first-line managers and middle managers respectively. On average, male managers had 23 subordinates while female managers had 29, which constitutes a significant difference (p < .001). There was also a significant difference (p < .001) in SOC between the different types of services included, with the most evident difference found between the human services (mean SOC = 30) and the technical services (mean SOC = 16). The average SOC for the 37 management teams ranged from five to 46 subordinates.

Do operational public sector managers in the same municipal operation, i.e. management team, experience similar job demands?

According to the intraclass correlation coefficients ICC(1,1), roughly 2–12% of the variance in the job demands studied can be attributed to the management-team level (Table 1). This variance can thus be explained by organizational features common for all the managers in the management team. The remaining part of the variance may thus be explained by aspects more tightly associated with the individual managers such as individual span of control.

How does span of control influence the levels of common job demands that operational public sector managers experience at work?

The results of the multilevel analysis are displayed in Tables 1 –8, including the fixed effects estimates of the predictors, and the amount of remaining unexplained variance at levels one and two (random effects) as well as model fit (loglikelihood) for each outcome, and are presented in greater detail in the text that follows.

Work overload

Work overload did not have a significant variation at the management-team level (Table 1). However, individual-level SOC showed a significant effect on work overload also after controlling for compositional effects through gender, managerial position and managerial experience (Table 2).

Conflicting logics

Conflicting logics did not have a significant variation at the management-team level (Table 1). When first entered into the model, the average SOC showed a significant effect. However, after controlling for individual-level SOC, this effect disappeared. The individual-level SOC showed a significant effect on conflicting logics after controlling for compositional effects through gender, managerial position and managerial experience (Table 3).

Excessive role demands

Approximately 11% of the variation in excessive role demands could be attributed to management-team belonging (Table 1). After controlling for compositional effects through gender, managerial position and managerial experience, the effect of SOC was consistent (Table 4). Thus, individual-level SOC as well as the average SOC in the management team contributed to managers’ experience of excessive role demands.

Buffer problems

Approximately 10% of the variation in job demand regarding buffer problems could be attributed to management-team belonging (Table 1). When first entered into the model, the average SOC showed a significant effect on buffer problems. However, after controlling for individual-level SOC, which showed a significant effect, this effect disappeared (Table 5). But after controlling for compositional effects through gender, managerial position and managerial experience the effect of SOC diminished and became non-significant. Thus, buffer problems in managers’ work seem not to be explicable by SOC.

Container function

Approximately 12% of the variation in this job demand could be attributed to management-team belonging (Table 1). After controlling for compositional effects through gender, managerial position and managerial experience, the effect of SOC was consistent (Table 6). Thus, individual-level SOC as well as the average SOC in the management team contributed to managers’ experience of exerting the role of container for subordinates’ frustration.

Group problems

Approximately 5% of the variation in the job demand group problems could be attributed to management-team belonging (Table 1). When first entered into the model, the average SOC showed a significant effect. However, after controlling for individual-level SOC, this effect disappeared. Individual-level SOC showed a significant effect on group problems also after controlling for compositional effects through gender, managerial position and managerial experience (Table 7).

Individual employee problems

Job demands concerning problems with individual subordinates’ performance and motivation did not show a significant management-team variation (Table 1). Further, SOC was not a significant predictor of managers’ problems with individual subordinates, either on individual or on management-team level (Table 8).

Are there diverse effects of span of control in different types of municipal services?

Furthermore a second aim was explored regarding different effects or consequences of SOC in different types of municipal services. To test for an influence of type of service, we entered type of service as well as an interaction term between SOC and type of service in the models concerning each job demand. The introduction of the interaction term did not significantly improve the fit in any of the models. Thus, it may be concluded that SOC has the same effect on managers’ job demand levels, regardless of the type of service in which they work.

Discussion

The main results of this study demonstrate, in line with our expectations, that SOC is of outmost importance for managers’ experience of a wide array of job demands common among operational managers in the public sector irrespective of the type of service performed. Thus, this study expands the JD–R model by establishing SOC as one important upstream structural condition casually prior to job demands. This finding is a significant contribution because it provides some insight into the reason for the experience of high levels of job demands, which are known predictors of, for example, health problems (Demerouti et al., 2001; Schaufeli and Bakker, 2004), lowered work engagement (Hu et al., 2011; Llorens et al., 2006) as well as turnover (Hu et al., 2011; Schaufeli and Bakker, 2004).

SOC significantly influences five of the seven job demands studied: work overload, conflicting logics, excessive role demands, group problems and container function. In these cases a wider SOC means a higher degree of demands in managers’ work despite the managers’ gender, position and experience. Hence, in the majority of job demands investigated regarding workload, the managerial role as well as relations with subordinates, it is unfavorable to have a large number of subordinates. Even though the regression coefficients may not seem very high (individual-level coefficients ranging from 0.009 to 0.015 in the final models), SOC ranges from two to 105 subordinates in the sample, therefore the predicted difference between the manager with the smallest and the manager with the widest SOC is e.g. (105–2) 103 × 0.015 = 1.55 standard deviations in the job demand (conflicting logics).

Furthermore, an additional important effect of the average SOC in the management team was found for two types of job demands: excessive role demands and container function. Thus the average SOC explained some of the shared variance found in these two job demands in the management teams. This implies that just being a member of a management team where the average SOC is higher increases the levels of job demands for the individual managers. The fact that managers’ colleagues have many subordinates may lead to less time to interact within the team and work alongside each other, which has been shown to be of great importance for e.g. the possibility of giving and receiving feedback (Bandura, 1999) and informal social support (Eraut, 2004). Hence, in management teams where the average SOC is higher the time to support and help fellow colleagues to handle the difficulties included in the managerial role may be limited and thus affect the levels of job demands experienced. An alternative explanation could be that management teams with higher average SOC share other similar organizational conditions. Irrespective of the explanation, this result should be taken into account when building public sector organizations and conducting future organizational research.

Work overload and excessive role demands

Several studies have shown that an excessive workload is a common problem for public sector managers (Lindholm et al., 2004; Skagert et al., 2004; Thylefors, 2007) and the present study indicates that managers’ SOC provides a partial explanation of work overload. Since several studies have shown that a high quantitative workload, in line with the assumptions underpinning the JD–R model, are associated with impaired health, i.e. emotional exhaustion (e.g. Peeters and Le Blanc, 2001), burnout (e.g. Hakanen et al., 2008; Schaufeli and Bakker, 2004) and psychosomatic health problems (e.g. Schaufeli and Bakker, 2004), it is important to diminish an excessive workload. Based on our data we suggest that a reduced SOC is one important way to improve the working conditions and thus health of managers.

In addition to quantitative work overload, managers may experience a lot of complexity as part of their work (e.g. Boyne, 2002; Mumford et al., 2000; Thylefors, 2007) and additionally so in public sector domains where NPM has had a major impact (e.g. Christensen, 2001), which, similar to other Nordic countries, has been the case in Sweden (e.g. Montin, 2000). The results of this study, demonstrating that an additional number of subordinates add to this complexity and causes the natural tasks and demands in managers’ work to become excessive, are not unexpected and supported by previous findings (Andersson-Felé, 2008). However, the finding that just being a member of a management team where the managers’ average SOC is higher increases the levels of excessive role demands experienced by the individual managers may be seen as more remarkable. Since both an optimal workload and balance between various roles are main criteria characterizing good work organizations (Lindström, 1994) it might be of great importance not only to look at the individual managers’ SOC, but also the average SOC in the management team in order to create a sustainable work situation for managers.

Conflicting logics

The present study shows that a wider SOC increases conflicting logics in managers’ work, thus making it more difficult to balance the different tasks included in the managerial role, e.g. administrative, strategic and personnel-related tasks. Previously, an increasing amount of administrative tasks along with a reduction in administrative support have been found to cause conflicting logics in the organization and thereby in managers’ work (Skagert et al., 2004; Wikström and Dellve, 2009). A wider SOC may be one reason why the amount of administrative (Forsberg Kankkunen, 2006) as well as personnel-related tasks has increased thus adding to possible conflicting logics in managers’ work (Andersson-Felé, 2008; Höckertin, 2007).

Buffer problems and container function

Contrary to our expectations, the present study did not provide any support for SOC’s influence on buffer problems. Instead, this job demand was heavily associated with the first-line managerial position, which could be a result of the somewhat diverse functions that managers at different levels in the organizational hierarchy hold (Arman et al., 2009; Mintzberg, 1993). However, exerting a container function, which can be considered a subset of buffer problems, was greatly influenced by the managers’ individual SOC. Interestingly, this was also the job demand most strongly associated with the average SOC of the management team. Thus, in order to keep the container function of the managerial job at a reasonable level the individual SOC as well as the average SOC in the management team should be targeted. Furthermore, female managers experience both these functions to a higher degree than male managers. Problems related to the buffer and container function may simply involve expectations of how male and female managers should act (Heilman, 2001).

Group problems and individual employee problems

Increasing problems with work groups’ performance and function as a result of wider SOCs as shown in this study are in accordance with previous research that has found that large SOCs can lower levels of group performance (Gittell, 2001). Group problems could also be a consequence of managers’ difficulties to involve employees in decisions and improvements under wide SOCs (Wikström and Dellve, 2009). Earlier studies have argued that SOC particularly may affect the interaction between managers and their subordinates (Brazier, 2005; Hultberg, 2007; Lucas et al., 2008). Managers with wide SOCs, for instance, interact less with individual group members (Gittell, 2001; Harder et al., 2000; Lucas et al., 2008) and have difficulties exerting good managerial work (Höckertin, 2007). It is therefore not inconceivable that the managers with many subordinates have less time for individual subordinates and thus less opportunity to be involved in their work and listen to criticism, while the groups may be harder to escape, for example because of scheduled work meetings. This may explain why SOC showed to be significant for problems with groups of employees but not individual employees.

Practical implications

Reduced resources (Westerberg and Armelius, 2000) and downsizing in line with NPM lead to work intensification and excessive SOCs amongst Swedish public sector managers (Harder et al., 2000; Ohlsson and Rombach, 1998; Thylefors, 2007). In accordance with previous studies (e.g. Andersson-Felé, 2008; Harder et al., 2000; Höckertin, 2007) the present study confirms that wide SOCs are common in municipal operations in Sweden and that female managers have significantly larger SOCs than men in general (Westerberg and Armelius, 2000). In line with these results, this study moreover validates that managers in human services on average have a wider SOC than managers in technical services (Forsberg Kankkunen et al., 2010). However, our study clearly shows that SOC is of equal importance for the job demands that public sector managers experience at work independently of type of service performed, and it is therefore important to question whether these differences in SOC can be justified.

Since job demands, in line with the JD–R model, play a central role in the health-impairment process that might lead to various health problems, but also to potential negative organizational outcomes such as turnover intention, reducing those job demands seems to be warranted. Thus, the results of this study point to an important structural condition to target. It can be considered as a valuable alternative or complement to adding more job resources in order to buffer the effect of high job demands especially since the buffer hypothesis still receives mixed statistical support (e.g. Bakker et al., 2004, 2005; Hu et al., 2011; Xanthopoulou et al., 2007). In addition, the empirical evidence regarding how adding resources as a buffer could be done successfully in practice has gained limited support (e.g. De Jonge and Dormann, 2006; Kompier and Kristensen, 2001), which makes other alternatives of lowering job demands particularly important. From a preventative point of view, decreasing job demands is to be preferred above increasing job resources (Schaufeli and Bakker, 2004) and the results from this study thus give organizations more alternatives to chose from when taking preventative actions for managers’ sustainability. Hence, most of the job demands investigated here could be lowered through a reduction in the individual SOC, but rather unexpectedly we found that lowering the average SOC in the management team also could be of importance. This is not only of utmost importance for managers to do their job and stay healthy and committed, but essential to the organization as a whole. Thus, without improving the working conditions for managers, efforts to recruit the best leaders and educate them may be wasted.

Methodological considerations

First, while it is possible to find support for certain associations in a cross-sectional sample, the direction of these associations are difficult to establish and it is not possible to rule out any spurious relationships, which is necessary in order to draw causal inferences. Therefore, the results need to be interpreted with caution, although it makes sense to see excessive role demands, for example, as a result of SOC rather than the other way around.

Second, the job demands were self-reported in the present study, which in the view of well-being can be considered more important than the results of a more objective assessment (Frese and Zaph, 1988; Lindström, 1994), since objective job demands do not automatically trigger stress reactions (Frese and Zaph, 1988). Even so, it would be interesting to investigate the associations longitudinally and with objective measures as references in future research. It could also be noted that common method variance due to self-reported data (Campbell and Fiske, 1959) probably was a minor problem in this study. The independent variables were more on the objective side of self-reports, since they can be considered ‘low in dependency on cognitive and emotional processing’ (Frese and Zaph, 1988). Furthermore, the focal predictors in terms of SOC were validated through register data, controlling for common method bias (Podsakoff et al., 2003).

Third, our sample is restricted both by the number of groups and the number of members in each group, which may be a disadvantage in MLM. Snijders and Bosker (1999) argue that group sizes may be as small as 1 as long as other groups are larger. Although, a larger number of group members might have been able to identify potential between- or cross-level interactions there is usually sufficient power if the number of groups exceed 20 and the sample size at the first level is not too small. The number of groups is usually more important than the number of cases per group (Tabachnick and Fidell, 2007).

Fourth, when controlling for compositional effects other independent variables could also have constituted appropriate choices. Socioeconomic status or educational level could for instance have been included in the analysis, but were not chosen since the sample is relatively homogeneous.

Directions for future research

In future studies it would be highly valuable to explore how SOC and relevant job demands found in this study fit into a larger set of working conditions and outcomes in accordance to the JD–R model in order to test the full potential of the model in relation to SOC. Further, it would be of great interest to deepen the understanding concerning the average SOC in management teams. More conclusive answers require longitudinal studies that combine job demands and job resources and relevant outcomes, e.g. health and commitment as well as turnover intentions. In addition, since the results from the present study are in a Swedish context it would be of great interest to make comparisons with other countries as well as similar operations run in the private sector to determine whether the conclusions may be more general.

Conclusions

Our study demonstrates that SOC is a key upstream component of operational public sector managers’ job demands and thus underlines the value of finding a reasonable number of subordinates per manager. It is disadvantageous to have a larger number of subordinates regarding the majority of job demands common among operational public sector managers. Excessive role demands and acting as a container for subordinates’ frustration were most heavily associated with SOC. The multilevel approach allowed us to reveal that these job demands besides being associated with the managers’ individual SOC were also influenced by the average SOC in the management team. Next to these two job demands, conflicting logics were most strongly associated with SOC, followed by work overload and group problems. Thus, a reasonable SOC may not only prevent unnecessary suffering for the managers as well as the subordinates, but most importantly could also improve the management team and the organization as a whole. Reducing SOC could thus be an important alternative or complement to adding more job resources. We also emphasize that SOC is of equal importance for managers’ levels of job demands in human services as well as technical services, suggesting that SOC is of general importance in the public sector in Sweden. The results imply that structural organizational interventions with goals to increase sustainability in terms of working conditions, health, performance and intention to stay as a manager should include efforts to reduce the number of employees that each manager is responsible for.

Footnotes

Funding

This research was supported by grants (dnr 2008-01951) from the Swedish Governmental Agency for Innovation Systems (VINNOVA).

Notes

Author biographies

Linda Wallin is a PhD student of work science in the Department of Sociology and Work Science at the University of Gothenburg. Her thesis focuses on public sector managers’ psychosocial working conditions and their association with different aspects of health and turnover.

Anders Pousette is an assistant professor at the Department of Psychology and a researcher at the Sahlgrenska Academy, Gothenburg University. His research interest includes safety and security in organizations, organizational climate and culture, psychosocial working conditions and stress, group processes and leadership.

Lotta Dellve is a professor in caring science and leadership at the University of Borås and KTH – Royal Institute of Technology, Stockholm. Her main research focus is sustainable psychosocial work environment and leadership in health care organizations.

References

Abma

Noordegraaf

(2003) Public managers amidst ambiguity: Towards a typology of evaluative practices in public management. Evaluation 9(3): 285–306.

Andersson-Felé

(2008) Leda lagom många – Om struktur kontrollspann och organisationsideal [Leading just the right number – on structure span of control and organizational ideals]. Doctoral dissertation, University of Gothenburg Sweden (ISBN-978–91–628–7352–3 ISSN 1653–8358).

Arman

Dellve

Wiktröm

Törnström

(2009) What health care managers do: Applying Mintzberg’s structured observation method. Journal of Nursing Management 17(6): 718–729.

Bach

Della Rocca

(2000) The management strategies of public service employers in Europe. Industrial Relations Journal 31(2): 82–96.

Bakker

Demerouti

(2007) The job demands–resources model: State of the art. Journal of Managerial Psychology 22(3): 309–328.

Bakker

Demerouti

de Boer

Scahufeli

(2003) Job demands and job resources as predictors of absence duration and frequency. Journal of Vocational Behavior 62(2): 341–356.

Bakker

Demerouti

Euwema

(2005) Job resources buffer the impact of job demands on burnout. Journal of Occupational Health Psychology 10(2):170–180.

Bakker

Demerouti

Verbeke

(2004) Using the Job Demands–Resources model to predict burnout and performance. Human Resource Management 43(1): 83–104.

Bakker

Van Veldhoeven

Xanthopoulou

(2010) Beyond the demand–control model. Journal of Personnel Psychology 9(1): 3–16.

10.

Bandura

(1999) Self-efficacy: Toward a unifying theory of behavioral change. In: Baumeister

(ed.) The Self in Social Psychology. New York: Psychology Press, pp. 285–298.

11.

Bell

(1967) Determinants of span of control. American Journal of Sociology 73(3): 90–101.

12.

Berntson

Wallin

Härenstam

(2012) Typical situations for managers in the Swedish public sector: Cluster analysis of working conditions using the Job Demands–Resources model. International Public Management Journal 15(1): 100–130.

13.

Bohte

Meier

(2001) Structure and the performance of public organizations: Task difficulty and span of control. Public Organization Journal: A Global Journal 1(3): 341–354.

14.

Boyne

(2002) Public and private management: What’s the difference? Journal of Management Studies 39(1): 97–122.

15.

Braizer

(2005) Influence of contextual factors on health care leadership. Leadership and Organizational Development Journal 26(2): 128–140.

16.

Burke

(1996) Unit size work experiences and satisfactions: An explanatory study. Psychological Reports 78: 763–767.

17.

Campbell

Fiske

(1959) Convergent and discriminant validation by the multitrait-multimethod matrix. Psychological Bulletin 56(2): 81–105.

18.

Cathcart

Jeska

Karnas

. (2004) Span of control matters. Journal of Nursing Administration 34(9): 395–399.

19.

Chandler

Barry

Clark

(2002) Stressing academe: The wear and tear of the New Public Management. Human Relations 55(9): 1051–1066.

20.

Christensen

(2001) Administrative reform: Changing leadership roles? Governance: An International Journal of Policy and Administration 14(4): 457–480.

21.

De Jonge

Dormann

(2006) Stressors, resources, and strain at work: A longitudinal test of the triple-match principle. Journal of Applied Psychology 91(5): 1359–1374.

22.

De Jonge

Mulder

MJGP

Nijhuis

FJN

(1999) The incorporation of different demand concepts in the job demand–control model: Effects on health care professionals. Social Science and Medicine 48(9): 1149–1160.

23.

Dellve

Wikström

(2006) Hållbart ledarskap i sjukvården – utveckling av ledarskap och stödstrukturer ur individ- och organisationsperspektiv [Sustainable leadership in health care – developing leadership and support structures from an individual- and organizational perspective]. Gothenburg: University of Gothenburg. Available at: www.vgregion.se/upload/PSA/H%C3%A5llbart.pdf.

24.

Dellve

Skagert

Eklöf

(2008) The impact of systematic occupational health and safety Management for occupational disorders and long-term work attendance. Social Science and Medicine 67(6): 965–970.

25.

Demerouti

Bakker

(2011) The Job Demands–Resources model: Challenges for future research. Journal of Industrial Psychology 37(2): 1–9.

26.

Demerouti

Bakker

Nachreiner

Schaufeli

(2001) The job demands–resources model of burnout. Journal of Applied Psychology 86(3): 499–512.

27.

Dollard

Bakker

(2010) Psychosocial safety climate as a precursor to conducive work environments, psychological health problems, and employee engagement. Journal of Occupational and Organizational Psychology 83(3): 579–599.

28.

Dorian

McCutcheon

Evans

. (2004) Impact of the Manager’s Span of Control on Leadership and Performance. Ottawa: Canadian Health Services Research Foundation.

29.

Eklöf

Pousette

Dellve

. (2010) Gothenburg Manager Stress Inventory (GMSI). Institute of Stress Medicine Region Västra Götaland, Sweden. Available at: www.vgregion.se/upload/Regionkanslierna/ISM%20Institutet%20f%C3%B6r%20stressmedicin/Publikationer/ISM-rapporter/ISM-rapport%209%20.pdf.

30.

Enders

Tofighi

(2007) Centering predictor variables in cross-sectional multilevel models: A new look at an old issue. Psychological Methods 12(2): 121–138.

31.

Eraut

(2004) Informal learning in the workplace. Studies in Continuing Education 26(2): 247–273.

32.

Farrell

Morris

(2003) The ‘neo-bureaucratic’ state: Professionals, managers and professional managers in schools, general practices and social work. Organization 10(1): 129–156.

33.

Forsberg Kankkunen

(2006) Könade verksamheter? [Gendered operations?] (Report No. 2006:43). Stockholm: Arbetslivsinstitutet. Available at: nile.lub.lu.se/arbarch/arb/2006/arb2006_43.pdf.

34.

Forsberg Kankkunen

Ylander

Höckertin

(2010) Managing the psychosocial environment: Logics of leadership in different organizational contexts. In: Marklund

Härenstam

(eds) The Dynamics of Organizations and Healthy Work. Work Life in Transition, 05/2010. Växjö: Linnéuniversitetet.

35.

Frese

Zaph

(1988) Methodological issues in the study of work stress: Objective vs subjective measurements of work stress and the question of longitudinal studies. In: Cooper

Payne

(eds) Causes, Coping and Consequences of Stress at Work. Oxford: John Wiley and Sons.

36.

Gittell

(2001) Supervisory span relational coordination and flight departure performance: A reassessment of post bureaucracy theory. Organization Science 12(4): 468–483.

37.

Glick

(1985) Conceptualizing and measuring organizational and psychological climate: Pitfalls in multilevel research. Academy of Management Review 10(3): 601–616.

38.

Goldstein

Browne

Rasbash

(2002) Tutorial in biostatistics: Multilevel modeling of medical data. Statistics in Medicine 21: 3291–3315.

39.

Gracunius

(1937) Relationship in organization. In: Gulick

Urwick

(eds) Papers on the Science of Administration. New York: Institute of Public Administration, Columbia University, pp. 183–187.

40.

Green

Anderson

Shivers

(1996) Demographic and organizational influence on leader–member exchange and related work attitudes. Organizational Behaviour and Human Decision Processes 66(2): 203–214.

41.

Gulick

Urwick

Mooney

. (1937) Papers on the Science of Administration. New York: Institute of Public Administration, Columbia University.

42.

Hair

Jr Anderson

Tatham

Black

(1995) Multivariate Data Analysis with Readings. Upper Saddle River, NJ: Prentice Hall.

43.

Hakanen

Schaufeli

Ahola

(2008) The Job Demands-Resources model: A three-year cross-lagged study of burnout, commitment and work engagement. Work and Stress 22(3): 224–241.

44.

Harder

Svärd

Wigforss

Hedén

(2000) Stress och belastning i vård och omsorg [Stress and strain in health care organizations] (No. 2000:8). Stockholm: Swedish Work Environment Authority.

45.

Härenstam

(2008) Organizational approach to studies of job demands control and health. Scandinavian Journal of Work Environment and Health 6: 144–149.

46.

Härenstam

(2009) Exploring gender working and living conditions and health: Suggestions for contextual and comprehensive approaches. Scandinavian Journal of Work Environment and Health 35(2): 127–133.

47.

Hasenfeld

(1983) Human Service Organizations. Englewood Cliffs, NJ: Prentice Hall.

48.

Hechanova-Alampay

Beehr

(2001) Enpowerment span of control and safety performance in work teams after workforce reduction. Journal of Occupational Health Psychology 6(4): 275–282.

49.

Heilman

(2001) Description and prescription: How gender stereotypes prevent women’s ascent up the organizational ladder. Journal of Social Issues 57(4): 657–674.

50.

Hjalmarsson

Norman

Trydegård

(2004) Om man ska vara stöttepelare åt andra måste man stå stadigt själv [You have to stand firm to be able to support others]. Stockholm: Stiftelsen Stockholms läns äldrecentrum. Available at: www.aldrecentrum.se/Global/Rapporter/2004/2004_2_stottepelare.pdf.

51.

Höckertin

(2007) Organisational characteristics and psychosocial working conditions in different forms of ownership. Doctoral dissertation, Umeå University Department of Public Sociology, Umeå, Sweden.

52.

Hood

(1991) A public management for all seasons? Public Administration 69(1): 3–19.

53.

Hox

(2002) Multilevel Analysis: Techniques and Applications. Mahwah, NJ: Lawrence Erlbaum.

54.

Schaufeli

Taris

(2011) The Job Demands–Resources model: An analysis of additive and joint effects of demands and resources. Journal of Vocational Behavior 79(1): 181–190.

55.

Hultberg

(2007) Ett hälsosammare arbetsliv. Vägledning för goda psykosociala arbetsförhållanden [A healthier worklife: Guidance for god psychosocial working conditions]. Socialmedicinsk Tidskrift 2: 114–122.

56.

Klein

Kozlowski

(2000) From micro to meso: Critical steps in conceptualizing and conducting multilevel research. Organizational Research Methods 3(3): 211–236.

57.

Kompier

MAJ

Kristensen

(2001) Organizational work stress in interventions in a theoretical, methodological and practical context. In: Dunham

(ed.) Stress in the Workplace: Past, Present and Future. Philadelphia, PA: Whurr Publishers, pp. 164–190.

58.

Korunka

Kubicek

Schaufeli

(2009) Work engagement and burnout: Testing the robustness of the Job Demands–Resources model. The Journal of Positive Psychology 4(3): 243–255.

59.

Lindholm

Dejin-Karlsson

Westin

. (2004) Physicians as clinical directors: Working conditions psychosocial resources and self-rated health. Occupational Medicine 54(3): 182–189.

60.

Lindström

(1994) Psychosocial criteria for good work organizations. Scandinavian Journal of Work Environment and Health 20: 123–133.

61.

Little

RJA

Rubin

(1987) Statistical Analysis with Missing Data. New York: Wiley.

62.

Llorens

Bakker

Schaufeli

Salanova

(2006) Testing the robustness of the job demands-resources model. International Journal of Stress Management 13(3): 378–391.

63.

Lucas

Laschinger

Wong

(2008) The impact of emotional intelligent leadership on staff nurse empowerment: The moderating effect of span of control. Journal of Nursing Management 16(8): 964–973.

64.

McCutcheon

(2004) Relationships between leadership style span of control and outcomes. Doctoral dissertation, University of Toronto, Graduate Department of Nursing Science, Toronto.

65.

Meier

Bohte

(2000) Ode to Luther Gulick: Span of control and organizational performance. Administration and Society 32(2): 115–137.

66.

Meier

Bohte

(2003) Span of control and public organizations: Implementing Luther Gulick’s research design. Public Administration Review 63(1): 61–78.

67.

Mintzberg

(1993) Structure in Fives: Designing Effective Organizations. London: Prentice Hall.

68.

Montin

(2000) Between fragmentation and coordination. Public Management: An International Journal of Research and Theory 2(1): 1–24.

69.

Montin

(2004) Moderna kommuner [Modern Municipalities]. Malmö: Liber.

70.

Moynian

Pandey

(2007) The ties that bind: Social networks person-organization value fit and turnover intention. Journal of Public Administration Research and Theory 18(2): 205–227.

71.

Mumford

Zaccaro

Harding

. (2000) Leadership skills for a changing world: Solving complex problems. Leadership Quarterly 11(1): 11–35.

72.

Ohlsson

Rombach

(1998) Res pyramiderna. Om frihetsskapande hierarkier och tillplattningens slaveri [Raise the pyramids. Freedom-creating hierarchies and the flattening slavery]. Stockholm: Svenska förlaget.

73.

Paccagnella

(2006) Centering or not centering in multilevel models? The role of the group mean and the assessment of group effects. Evaluation Review 30(1): 66–85.

74.

Pandey

Wright

(2006) Connecting the dots in public management: Political environment organizational goal ambiguity and the public manager’s role ambiguity. Journal of Public Administration Research and Theory 16(4): 511–532.

75.

Peeters

Le Blanc

(2001) Towards a match between job demands and sources of social support: A study among oncology care providers. European Journal of Work and Organizational Psychology 10(1): 53–72.

76.

Podsakoff

MacKenzie

Lee

J-Y

Podsakoff

(2003) Common method biases in behavioural research: A critical review of the literature and recommended remedies. Journal of Applied Psychology 88(5): 879–903.

77.

Rainey

Bozeman

(2000) Comparing public and private organizations: Empirical research and the power of the a priori. Journal of Public Administration Research and Theory 10(2): 447–469.

78.

Rasbash

Steele

Browne

Goldstein

(2009) A User’s Guide to MLwiN. Version 2.10. Bristol: Centre for Multilevel Modelling, University of Bristol.

79.

Rosenberg Hansen

Villadsen

(2010) Comparing public and private managers’ leadership styles: Understanding the role of job context. International Public Management Journal 13(3): 247–274.

80.

Schafer

Graham

(2002) Missing data: Our view of the state of the art. Psychological Methods 7(2): 147–177.

81.

Schaufeli

Bakker

(2004) Job demand, job resources and their relationship with burnout and engagement: A multi sample study. Journal of Organizational Behaviour 25(3): 293–315.

82.

Schnall

Landsbergis

Baker

(1994) Job strain and cardiovascular disease. Annual Review of Public Health 15: 381–411.

83.

Schrout

Fleiss

(1979) Intraclass correlations: Uses in assessing rater reliability. Psychological Bulletin 86(2): 420–428.

84.

Schwartz

Pickering

Landsbergis

(1996) Work-related stress and blood pressure: Current theoretical models and considerations from a behavioral medicine perspective. Journal of Occupational Health Psychology 1(3): 287–310.

85.

Schyns

Maslyn

Weibler

(2010) Understanding the relationship between span of control and subordinate consensus in leader–member exchange. European Journal of Work and Organizational Psychology 19(1): 1–19.

86.

Simon

(1946) The proverbs of administration. Public Administration Review 4(1): 16–30.

87.

Skagert

Dellve

Eklöf

. (2004) Om ledarskap och stress i politiskt styrd verksamhet [Leadership and stress in politically governed operations]. Gothenburg: Institute for Stress Medicine.

88.

Skagert

Dellve

Eklöf

. (2008) Leaders’ strategies for dealing with own and their subordinates stress in public human service organizations. Applied Ergonomics 39(6): 803–811.

89.

Snijders

Bosker

(1999) Multilevel Analysis: An Introduction to Basic and Advanced Multilevel Modeling. London: Sage.

90.

Söderfeldt

Muntaner

. (1996) Psychosocial work environment in human service organizations: A conceptual analysis and development of the demand-control model. Social Science and Medicine 42(9): 1217–1226.

91.

Spector

Brannick

(2011) Methodological urban legends: The misuse of statistical control variables. Organizational Research Methods 14(2): 287–305.

92.

Szücs

Strömberg

(2006) Local Elites Political Capital and Democratic Development: Governing Leaders in Seven European Countries. Wiesbaden: Erlag für Sozialwissenschaft.

93.

Tabachnick

Fidell

(2007) Using Multivariate Statistics. Boston: Pearson Education.

94.

Theobald

Nicholson-Crotty

(2005) The many faces of span of control: Organizational structure across multiple goals. Administration and Society 36(6): 648–660.

95.

Thylefors

(2007) Ledarskap i Human service- organisationer [Leadership in human service organizations]. Stockholm: Natur och Kultur.

96.

Urwick

(1956) The managers’ span of control. Harvard Business Review May–June: 39–47.

97.

Van den Broeck

Vansteenkiste

De Witte

Lens

(2008) Explaining the relationships between job characteristics burnout and engagement: The role of basic psychological need satisfaction. Work and Stress 22(3): 277–294.

98.

Westerberg

Armelius

(2000) Municipal middle managers: Psychosocial work environment in a gender-based division of labor. Scandinavian Journal of Management 16(2): 189–208.

99.

Wikström

Dellve

(2009) Contemporary leadership in healthcare organizations. Journal of Health Organization and Management 23(4): 411–428.

100.

Wooldridge

(2005) The impact of centering first-level predictors on individual and contextual effects in multilevel data analysis. Nursing Research 54(3): 212–216.

101.

Xanthopoulou

Bakker

Dollard

. (2007) When do job demands particularly predict burnout? The moderating role of job resources. Journal of Managerial Psychology 22(8): 766–786.