Police work continues to be viewed as a high stress occupation with investigators often reporting higher rates of morbidity and mortality than observed in many other occupations (see, for example, Doctor, Curtis, & Isaacs, 1994). Evidence from large sample surveys indicates that police officers have high rates of marital disruption, health problems, and alcohol abuse (Blackmore, 1978). Moreover, police officers also appear to be less likely to make use of public mental health facilities, and some observers believe that this reflects the role of an occupational culture that discourages officers from admitting that they need assistance in handling the stresses of their work (Blackmore, 1978). Police officers, therefore, might doubly suffer to the extent that they are exposed to a high level of work stressors that are risk factors for health and well being while at the same time are discouraged from seeking medical or counseling assistance.
Despite the general acceptance of the "police stress hypothesis," however, some critics have questioned the claim that work stress is necessarily the cause of the mental and physical health symptoms exhibited by police (Malloy & Mays, 1984; Terry, 1981). Indeed, it is overly simplistic to assume that police work represents a homogeneous occupation with respect to the factors that most likely account for its experienced stressfulness. The thrust of our study, then, is to examine the specific factors that are hypothesized to be significant causes of stress and strain in police work. In addition, our aim is to focus especially on those factors internal to the police bureaucracy and which therefore might be amenable to control and management by the organizational hierarchy. Following Malloy and Mays (1984), we assume that these stressors vary significantly across different roles and assignments within the police organization.
THE NATURE OF POLICE STRESS
The stereotypical picture of police stress as consisting primarily of exposure to physical danger from criminals is fading, at least within the academic literature if not in popular accounts. A myriad of surveys of police stress point to the important role of what might be termed organizational or management factors, in contrast to physical or emotional threats encountered during fieldwork. This is not to say that police work does not entail physical risk. However, such risks are well known to individuals who select police work as a career, and may, in fact, even attract recruits to this occupation. A source of stress made more insidious by its chronic nature and the pervasiveness of its impact on the officer's work life and career stability arises out of the constellation of management practices and policies characterizing many police organizations. It has now been twenty years since Kroes (1976) first described the "hassles" that police officers must endure from their administrators as well from judges, lawyers, and the general public. Other researchers have since confirmed that management issues are often perceived as being the most important causes of stress (Brown & Campbell, 1990; Cooper, Davidson, & Robinson, 1982; Cooper & Grimley, 1983). Such hassles include shift schedules that disrupt sleep patterns and social life, autocratic and unsympathetic supervisors, lack of recognition for work accomplishments, and lack of autonomy in performing duties (Cooper et al., 1982). Because we believe that these types of stressors generally arise from the decisions and policies enacted by police managers, they are also the ones that ultimately are the most controllable by making appropriate interventions in the police management structure. For this reason we focused our efforts in this study on understanding how managerial actions might directly affect the stress that police officers experience, and how such managerial actions might also indirectly determine officers' well being through their buffering effects on stressors arising from sources external to the department.
Police organizations are characterized by a para-military structure and climate which encourages an authoritarian approach to leadership. The assumption that such structures must be inherently para-military has been challenged, however (Angell, 1971; Cordner, 1978; Jermier & Berkes, 1979; Sandler & Mintz, 1974). There is still surprisingly little empirical evidence concerning how different leadership styles relate to either police effectiveness or to the mental and physical health of officers, although some data suggest that police officers might prefer a more participative and supportive approach. A survey of 365 law enforcement officers in the state of Arizona, USA, showed that police described the management style of their department as being "benevolent-authoritative" (Bruns & Shuman, 1988). On the other hand, these officers indicated that their preferred style was "participative." An assessment of the implementation of community policing in Madison, Wisconsin, moreover, indicated that officers began to see the management style of their department as becoming more participative over the first two years of adoption (Wycoff & Skogan, 1994). Perceptions that management was more participative were also positively correlated with job satisfaction, perceived significance of their work, and work autonomy. Thus, there is some emerging evidence that alternative styles of management in police organizations are worthy of consideration. Although we do not see the existing evidence as being sufficient to make a compelling argument for the adoption of participative management styles in all police organizations, we do believe that the research argues for the need to make closer examinations of leadership variables in police work. Moreover, unlike many of the externally-based stresses that police officers face during the course of their work, management styles and leadership behaviors are ultimately under the control of the police organization itself. Thus they constitute viable targets for interventions aiming to improve the effectiveness and well being of police officers. Our study focuses, then, on the critical role that leadership behaviors play in the experience of stress and strain in police work. Below we present a causal model of how police leadership behaviors influence health and well being through their influence on the key intervening constructs of control and social support.
Our theoretical model aims to illuminate the role of leadership behaviors in the experience of stress and well being by linking them to two other theoretical constructs -- personal control and social support -- that are prominent in the work and stress literature. Personal control refers to the individual's belief that he or she can influence important elements of the work environment such as work pacing, work methods, social interactions, and rules and policies (Ganster & Fusilier, 1989). Much evidence now exists suggesting that such control perceptions are directly related to enhanced mental and physical well being and sometimes also play a buffering role by negating the negative impact of other job demands (Fox, Dwyer, & Ganster, 1993; Ganster & Fusilier, 1989; Karasek, 1979). Moreover, recent evidence also indicates that certain organizational-level policies can improve employee health and well being by enhancing personal control (Thomas & Ganster, 1995). Although some aspects of personal control are determined by broader organizational policies, an individual's immediate supervisor is hypothesized to have a significant impact on this construct. Participation in decision making is one way that supervisors can enhance the personal control that officers experience over how they do their work. In addition, clarifying role expectations and setting specific goals can enhance the officer's perception of control over performing his or duties as well as giving officers a greater sense of control over their performance evaluation and career progress.
Social support refers to instrumental assistance and emotional supportiveness as perceived by the police officer. There is now a large body of literature that attests to the benefits of social support in alleviating experienced stress and improving general well being (see Ganster & Victor, 1988, for a review of the general health literature, and Ganster, Fusilier, & Mayes, 1986, for a review of the work literature). Much of the evidence from the occupational sphere, moreover, points to the important role played by social support from the supervisor in particular (Ganster et al., 1986; Thomas & Ganster, 1995). Thus, we include social support in our model because it exerts a significant effect on experienced stressfulness and is also hypothesized to be related to specific managerial behaviors.
Figure 1 displays our model of the police stress process that focuses on managerial behaviors as the key organizational variables. The model includes five leadership variables that we hypothesize will be related to police officer health and well being through their influence on perceptions of personal control and social support. Consideration refers to the leader's display of behaviors that acknowledge the feelings and emotions of the subordinate and which show a concern for his or her welfare. Our model hypothesizes that this leadership style will be associated with greater perceptions of social support by the police officer. Participation in decision making occurs when the leader solicits the input of subordinates in the decisions that affect the unit. We expect this behavior to be a significant predictor of the personal control perceived by subordinates over their work situation. Role clarification refers to communications from the leader that help the officer understand which behaviors the leader expects to be displayed. These communications also help clarify the basis for evaluation of the subordinate's job performance. It is predicted that role clarification will increase subordinates' sense of personal control by making clearer to them the contingent relationships between their own job behaviors and the consequences of those behaviors as mediated by the superior. Leader goal setting refers to efforts by the supervisor to set specific goals for each subordinate. We hypothesize that such goal setting behaviors will help enhance the control that officers have on their job. Finally, performance emphasis refers to pressure on subordinates from the leader to achieve high levels of job performance. This leader behavior is akin to the "task orientation" traditionally referred to in the broader leadership literature. We hypothesize that when leaders' styles are characterized by high levels of such performance pressure that officers are more likely to experience stress-related symptoms.
Thus, our model hypothesizes that leader behaviors can have a significant effect on the experienced stress symptoms of police officers. The effects of the specific leader behaviors discussed above are produced primarily by the effects that they have on the mediating variables of social support and personal control. Two other intermediate factors are also examined in this study. First, shift schedules that disrupt sleep patterns and the social lives of police officers are known to be significant sources of stress. In addition, just having sufficient advance notice of one's shift schedule can help alleviate, through the opportunity for planning, the disruptions caused by otherwise inconvenient shifts. In our model both of these variables are hypothesized to be predictive of mental and physical well being. Moreover, we expect the leader behaviors themselves to be causal factors in these shift variables, for leaders often have much discretion in planning the schedules of their units. Thus, we explore the relationships between these leader behaviors and the prevalence of noxious shift schedules and an absence of prior notification.
Finally, there are some stresses, such as those encountered through contact with criminals and the community, that are not generally under the direct influence of the officer's supervisor. However, leader behaviors might still be important in determining what effects these external stresses have on the well being of officers. Our main hypothesis in this regard is that the leader behaviors will help determine experienced social support and personal control, and these variables will serve to buffer the negative effects of outside sources of stress. The buffering effects of social support and personal control are well documented in the literature (Fox, Dwyer, & Ganster, 1993; Ganster et al., 1986).
In sum, in this study our aim is to explore the role that key leader behaviors play in determining the mental and physical well being of police officers. We hypothesize that these leadership styles can have direct effects on mental and physical health as well as indirectly affecting these outcomes through their effects on the levels of social support and personal control that police officers experience on their job. An ancillary aim of our research is to investigate the applicability of this model, which is based primarily on research conducted in the United States and Western Europe, to the national police force of Slovenia.
Data were obtained from police officers working in Slovenia. Respondents were queried at twelve police stations. In all, 192 police officers were included in this study. Approximately 58% of the sample reported their current job position as "police officer" which includes patrol, traffic, and border police jobs. The remaining job positions were reported as follows: detective (14%); dispatcher (2.6%); community police officer (15.2%); chief of police unit (.5%); deputy chief of police unit (6.3%); and chief of police department (2.1%). The average length of time incumbents had worked for the police force was 8.29 years, with a mean of 3.98 years working in their current position. The majority of the sample was male (96%) with a mean age of 28 years and a range 19 to 52 years.
Leader behaviors. Measures of leader behavior were obtained from subordinates using the five subscales of the Managerial Practices Survey developed by Yukl and Nemeroff (1979). Each of the five subscales consisted of six items scaled on a five-point Likert type scale. The five scales included: Leader Consideration (alpha=.89; e.g., "My supervisor makes subordinates feel at ease when talking with them"); Leader Role Clarity (alpha=.85; e.g., "My supervisor explains each subordinate's duties and job responsibilities"); Leader Decision Participation (alpha=.89, e.g., "My supervisor allows subordinates to have substantial influence in the making of decisions"); Leader Goal Setting (alpha=.90, e.g., "My supervisor meets with individual subordinates to jointly establish goals and objectives for each important aspect of the subordinates job"); and Leader Performance Emphasis (alpha=.87, e.g., "My supervisor closely checks on the performance of subordinates to see if it is adequate").
Control. Job control was assessed using a 22-item scale developed by Ganster (1985). This measure reflects the degree of perceived job control an employee has over tasks, decisions, resources and the physical environment. This scale used a 5-point Likert-type scale anchored with (1) "very little" to (5) "very much" (alpha=.90).
Social Support. Social support received from supervisors was measured using a 4 item scale asking subordinates to indicate the extent to which they felt their supervisors: (1) were willing to listen to personal problems; (2) went out of their way to make work life easier; (3) were easy to talk to; (4) could be relied upon when things were tough at work. Responses were anchored on a five-point Likert-type scale (alpha=.84).
Environmental Stress. This 3-item scale was developed for this study and measured commonly reported sources of stress for police officers including fear of revenge from criminals, inappropriate behavior from the public, as well as sense of personal endangerment. Responses ranged from (1) never to (7) frequently (alpha=.87).
Depression. Levels of incumbent depression were measured using the NIMH Center for Epidemiological Studies Depression Scale (CES-D) (Radloff, 1977). The scale comprises 20 items which ask the incumbent to indicate their feelings about the future, themselves, their mood, and physical symptoms (e.g., sleep patterns, appetite). Scores range from 0 (rarely) to 3 (most or all of the time), with higher scores indicating higher levels of depression. The coefficient alpha for the current study was .93.
Somatic complaints. Physical symptomatology was measured using a 17-item expanded version of Caplan et al.'s (1975) somatic complaints index (alpha=.86). Each item asks respondents to indicate the frequency with which they have experienced a variety of symptoms (e.g., headache, stomach pain) in the past month. Higher scores reflect higher levels of psychosomatic symptomatology.
Suicide. Suicidal ideation was assessed using a single item scale developed by Beehr and his colleagues (1995). Respondents were asked, "During the past six months how often have you thought about committing suicide?" Responses ranged from 1 (never) to 7 (always).
Medical Problems. Two types of health-related disorders (back problems, heart disease) were gathered through a check-list format. Respondents checked "yes" or "no" whether they "had been told by a doctor or treated for any of the following conditions in the past year.
All measures were translated from English into Slovenian by a professional translator. These measures were then back translated into English by one of the project officers to ensure the original translation into Slovenian had been correct and the meanings of the items had not been altered.
All data were collected in group administrations supervised by a research assistant during working hours at the police stations. Confidentiality and anonymity were guaranteed and all participation was voluntary.
RESULTS AND DISCUSSION
Table 1 lists the descriptive statistics and correlations among all study variables. We begin assessing the validity of our model by examining the correlations of the leadership behavior variables with the intervening variables of social support and control. As can be seen, each of the leadership style variables is significantly correlated with social support, with correlations ranging from .30 for leader performance emphasis to .63 for leader consideration. Not surprisingly, consideration shows the largest correlation with social support. When social support is regressed on this set of leadership variables, they jointly account for 40 % of the variance (R2=.40, F(5, 180) = 24.18, p<.01). Because of the high intercorrelations among the leadership variables, however, not all of the leader variables have significant regression coefficients.
Each of the leadership variables is also significantly correlated with personal control perceptions, with correlations ranging from .24 for performance emphasis to .41 for leader consideration. As with social support, consideration is the largest correlate of control perceptions. When control is regressed on the set of leadership behaviors they account for 18 % of the variance (R2=.18, F(5, 177) = 7.68, p<.01). Again, however, because of the multi- collinearity within the set of leadership variables, some of the leader variables do not have significant regression coefficients.
We also examined the relationships between police leader behaviors and the shiftwork variables. First, none of the leader behaviors was associated with being assigned to different shift schedules, implying that such shift assignment decisions are unrelated to the leadership styles we studied. However, leader consideration and participation in decision making were significantly correlated with the amount of advance notice that officers had regarding their shift schedules. External police stress was not correlated with any of the leader behaviors, confirming our expectations that this source of stress was truly outside the domain of the police command structure.
When we examined the health outcomes, we found that depression is negatively correlated with personal control. Social support is negatively associated with depression and somatic complaints. External police stress is associated with elevated levels of all the health outcomes except suicide thoughts, but including back problems and high blood pressure. Finally, the leader behaviors are negatively associated with depression and somatic complaints but not with the back problem, suicide thoughts, and high blood pressure variables.
According to our model, we do not expect leader behaviors to be directly related to environmental sources of stress, as these refer to events generally outside the control of the command structure itself. However, we hypothesized that leader behaviors could make a difference in how officers were able to cope with these environmental stresses through their effects on control and social support. To test this hypothesis, we calculated interaction effects between environmental stress and control and social support to see if these intervening variables moderated the relationship between environmental stress and the health outcomes. Unfortunately, none of the moderator tests indicated that control and social support were significant buffers of the effects of environmental stress. Likewise, neither control nor social support buffered the relationship between shift schedules and the health outcomes.
The general pattern of results indicates that leader behaviors can serve to increase the personal control and social support that police officers experience at work. Control and social support, in turn, are significant predictors of mental and physical health outcomes. Thus, it appears that police organizations might be able to improve the well being of their members by teaching police supervisors to adopt more effective leadership behaviors. These behaviors include being more considerate of police officers' personal feelings, including them in joint decision making, clarifying their roles, and setting specific performance goals.
We also found that environmental sources of stress were significant predictors of health outcomes and were independent of the leadership behaviors, as expected. Although a large literature suggests that social support and control often buffer the effects of work stress, our findings failed to show that they buffer the effects of police stress from the external environment. Thus, although leader behaviors show many other beneficial effects in the context of police well being, our data do not suggest that we can rely on them to help police officers effectively deal with threats from criminals and the community. Perhaps more specific coping behaviors are needed to deal with these stresses. Given the reluctance of police officers to admit to feelings of stress arising from their work, especially to representatives of the command structure, supervisors are not likely to be useful sources of coping resources. In this area counseling services that provide confidentiality to the officer may represent the best avenue for helping them cope with stresses outside the control of the police organizational structure. Our findings suggest that investigating such sources of coping resources can yield many benefits, as environmental stress appears to be a potent correlate of both mental and physical health outcomes that are independent of their experienced control and social support.
TABLE 1: Means, Standard Deviations, and Correlations among Study Variables
Variable MEAN Standard 1 2 3 4 5 6 7 8 9 10 11 12 13 Deviation 1. Control 2.51 .54 - 2. Leader Consideration 3.18 .99 .42** - 3. Leader Decision Making 2.99 .99 .35** .73** - 4. Leader Goal Setting 3.31 .84 .28** .65** .70** - 5. Leader Performance Emphasis 4.01 .66 .25** .55** .59** .67** - 6. Leader Role Clarity 3.65 .65 .24** .55** .68** .74** .73** - 7. Advanced Schedule Notice 3.40 1.67 .24** .27** .16* .12 -.04 .02 - 8. External Stress 3.26 .89 -.02 .04 -.10 -.03 .07 -.05 .09 - 9. Social Support 2.52 .78 .36 .63** .45** .42** .30** .33** .24** .02 - 10. Depression .65 .43 -.21** -.28** -.23** -.23** -.17* -.18** -.12 .25** -.25** - 11. Somatic Complaints 1.73 .66 -.14 -.16** -.16* -.15* -.15* -.21** -.13 .35** -.21** .61** - 12. Back Problems 1.22 .41 -.06 -.07 -.13 -.06 .09 -.01 -.11 .17** -.12 .24** .31** - 13. High Blood Pressure 1.11 .30 -.07 .02 -.02 -.01 .07 .05 -.14 .20** -.04 .09 .26** .20** -
Figure 1. Model of Police Stress and Leader Behavior