In recent years, the concept of job control or decision latitude has increasingly been incorporated into many studies in the medical literature looking at a wide range of aspects of physical health. This is largely due to the success of Karasek’s model in offering a simple framework enabling key work variables central to his theory of job strain to be measured using brief scales. Thus, it has been suggested that high demands and low control (i.e. job strain) are related to musculoskeletal disorders (e.g. neck pain) in sales people (Skov et al., 1996) and adverse outcome of pregnancy in clerical and commercial workers (Brandt & Nielsen, 1992). Job strain has also been shown to be associated with non-medical drug use (Storr et al., 1999). There is less evidence for associations in the fewstudies that have looked at cancer risk (Achat et al., 2000; Courtney et al., 1996; Van Loon et al., 2000). However, the bulk of the literature focuses on cardiovascular disease and the associated risk factors.
Cardiovascular Disease
Schnall et al. (1994) conducted an extensive review of the literature in relation to heart disease. They considered 36 studies published between 1981 and 1993 and concluded that most found a significant relationship between job strain and cardiovascular or all-cause mortality, or between job strain and risk factors for cardiovascular disease (CVD).
Epidemiological studies examining morbidity and mortality have used two main methods of classifying employees according to the job strain dimensions. A number of studies have tried to achieve a relatively objective measurement of the work stressors associated with particular jobs by using a methodology that classifies individuals on the job strain dimensions according to their job title (based on average ratings of job incumbents). Thus, for example, waiters might all be classified as having low-control, high-demand jobs. Using this occupation-level analysis, Alfredsson et al. (1982) found that hectic work combined with low control was associated with higher incidence of heart disease. Other studies have assessed job stressors using the more subjective method of asking individuals to rate their levels of demand and control. For example, Johnson et al. (1989) found the greatest risk was in high-demand, low-control isolated jobs.
Increasingly, studies have also focused on the relationship between job strain and the risk factors that are implicated in CVD such as high blood pressure (Brisson et al., 1999; Fletcher & Jones, 1993; Fox et al., 1993) or measures of adrenalin and cortisol (Fox et al., 1993; Pollard et al., 1996). For example, Fox et al. in their study of nurses found that the combination of high demands and low control predicted both blood pressure and cortisol levels. However, Fletcher & Jones (1993), in a sample from heterogeneous occupations, found no relationships between control and blood pressure, and where demands showed relationships, these were in the opposite direction to that predicted (i.e. those with lower demands had higher blood pressure). Such mixed findings seem typical of this area. For example, Schnall et al. (1994) in their review, found that in eight studies of casual blood pressure (measured in a clinic) only one found a significant association, but five out of nine found associations for ambulatory blood pressure (which is a more reliable measure). In an attempt to shed light on the possible mechanisms whereby job strain may impact on health, there are a small number of experimental studies that have manipulated levels of job strain in the laboratory. These have examined the relationship of experimental tasks to short-term physiological indicators that are implicated in CVD development. These include heart rate levels and cortisol (Perrewe & Ganster, 1989; Steptoe et al., 1993). For example, Steptoe et al. found that middle-aged men showed greater changes in blood pressure when they could not control the pace at which they performed laboratory tasks involving problem solving and mirror drawing. However, pacing had little effect on cortisol, suggesting work pace has a specific effect on cardiovascular functioning.
A review by Van der Doef & Maes (1998) of the impact of job strain on health concluded that, across different populations, measurement methods and job designs, there is substantial support for the hypothesis that high-demand, low-control jobs lead to increased CVD. However, the focus of this review is on the combined effects of demands and control rather than their independent impacts. The earlier review of Schnall et al. agreed with the conclusion of Van der Doef & Maes, but where possible they also considered the separate effects. They concluded that while 17 out of 25 studies found significant associations between job decision latitude and outcome, only 8 out of 23 studies showed significant relationships between demand and outcome. A few studies have also found an effect for demands opposite to that predicted (Alterman et al., 1994; Hlatky et al., 1995; Steenland et al., 1997).
Overall, therefore, deriving a clear message from this literature is difficult. However, where demand and control are separated, the evidence seems to point to the importance of job control more strongly than demands. While evidence here is mounting, further work is needed (including more laboratory studies) to find out what specific aspects of control may be important. For example, is it control over pace of work that is important (i.e. control that enables the employee to modify demand) or does more general involvement in decisions about work have an impact? Furthermore, research evidence needs to clarify the nature of the relationship and the effects (if any) of individual differences.
Psychological Well-Being
Many researchers have studied the effect of job control on psychological well-being, not only because psychological distress is important in its own right, but because it is assumed to be the vehicle wherebywork stressors ultimately may lead to illness (both mental and physical). As a result there is a plethora of research indicating that low job control is associated with poor psychological well-being. This is typically measured in terms of scores on the General Health Questionnaire (Goldberg, 1978) or on more specific measures of anxiety, depression or job satisfaction. Although there are many exceptions, studies also generally support the relationship between a combination of lowcontrol and high demand with poor psychological well-being, with additive effects found more frequently than moderated effects (Van der Doef&Maes, 1999). However, the majority of studies are subject to serious methodological limitations. Typically, studies are cross-sectional and based on self-reports of both control and psychological well-being. Such methods have a number of well-established difficulties, not least the fact that they are open to the alternative explanation that being anxious or depressed may cause people to describe their jobs more negatively.
The fact that people who report lowlevels of demand and control also report high levels of distress at work is likely to be important for employers. However, the assumption that these associations are causal and further that they represent the first step towards damaged physical health and serious psychiatric illness is, as yet, not well tested. In particular, while many studies use validated self-report measures of psychiatric symptoms, there is surprisingly little research looking at the associations between job control and independently verified psychiatric illness.However, a fewstudies have looked at psychiatric illness using diagnostic interview schedules. These are often designed to be administered by non-clinicians and are used to classify individuals according to well-established psychiatric criteria (e.g. DSM-III). They are typically thorough and are likely to be both more objective and more valid than assessments based on brief self-report measures. A number of such studies have found associations between low control and psychiatric symptoms. For example, Muntaner et al. (1991) and Mausner-Dorsch & Eaton (2000) found that occupations associated with lower levels of control had higher levels of depression. Furthermore, Cropley et al. (1999), in a study of teachers, found that job strain was associated with neurotic disorder.
A sizeable literature also investigates the impact of autonomy on self-reported psychological well-being using the JCM (research which also typically suffers from the limitations discussed above). The most popular outcome considered, in this context, is job satisfaction. A meta-analysis of 28 studies of the relationship between job characteristics and job satisfaction found support for this relationship and found that of all the core job characteristics, autonomy had the strongest relationship with satisfaction (Loher et al., 1985). Other outcomes considered have included anxiety, depression and general mental health. An early meta-analysis by Spector (1986) looked at perceived control (most commonly based on the measures taken from the JDS) in relation to 19 outcome variables, including some healthrelated outcomes. This supported the relationship between autonomy and emotional distress (as well as absenteeism and physical symptoms). More recently Saavedra and Kwun (2000) have used the model to predict other affective states, including positive affect, and found that autonomy is particularly associated with enthusiasm, the implication being that not only will increasing autonomy relieve job dissatisfaction but it may also serve to “energise, reinforce and maintain work behaviour” .
Health Behaviour
While the effect of job control on psychological well-being is considered to be one possible mechanism whereby job control may impact on health, an alternative mechanism is that job control may influence health by its effect on health behaviour. Thus, it would be hypothesised that having a low-control, high-demand job may lead to people perhaps smoking and drinking alcohol more, eating less healthily and exercising less. This hypothesis was confirmed by Weidner et al. (1997), who found that general health damaging behaviour (smoking, drinking alcohol, drinking coffee and failing to exercise) increased under conditions of low control, if demand was high. However, it seems that there may be different patterns of relationships depending on the health behaviour under consideration.
Smoking has been subject to the greatest amount of research but findings are nonetheless mixed. A number of studies have found that, after controlling for socioeconomic variables, men in high-strain jobs tend to smoke more (Green & Johnson, 1990; Hellerstedt & Jeffrey, 1997), yet others have failed to find this association (Alterman et al., 1994; Reed et al., 1989). Johannson et al. (1991) found that only demands and not control were associated with increased smoking. Hellerstedt&Jeffrey also found this to be the case in the women in their sample. In contrast, Alterman et al. only found that low levels of decision latitude had an impact on smoking. Similarly mixed findings exist in relation to alcohol use (Landsbergis et al., 1998).
A few studies have also looked at the effects of demands and control on exercise and diet. Johannson et al. (1991) found both demands and control were predictive of exercise, whereas Hellerstedt&Jeffrey (1997) found that decision latitude but not demandwas related to exercise. However, job demands were related to increased fat intake in men and higher body mass index in women.
One prospective study by Landsbergis et al. (1998) looked at change in job characteristics and change in health behaviour in a sample of male employees in a variety of jobs. They found that an increase in decision latitude was associated with a decrease in smoking in men over a period of three years. However, change in job characteristics was not associated with any change in weight or alcohol consumption.
Overall, as can be seen, the pattern of results remains somewhat inconsistent. It is certainly not clear that high levels of demand and control are linked to uniformly worse health behaviours, and any impact of these variables on health behaviours may be modest (Landsbergis et al., 1998). Nevertheless, Landsbergis et al. suggest that increasing job decision latitude may help reduce smoking. More research is needed into the exact mechanisms whereby job strain impacts on health behaviour. However, one longitudinal study suggests that the mechanisms may not be straightforward (Payne et al., in press). This study found that people high in job strain tended to exercise less than those in low-strain jobs (though they did not intend to do any less) and that high job demands and low control were related to low self-efficacy for exercise (a predictor of actual exercise). However, once people had formed an intention to exercise, demands rather than control disrupted these intentions.
Impacts of Job Control on Home Life and Well-Being of Other Family Members
A considerable body of research suggests that the effects of work (including the types of activities people engage in at work and their resulting affective states) spill over into the home environment. For example, Rousseau (1978) asked individuals to rate aspects of work and non-work using the JDS and found positive relationships between home and work ratings. She also found relationships betweenwork and non-work satisfaction. Karasek et al. (1987) also found that lowjob controlwas related to lower levels of social participation. This was particularly the case for women, for whom low levels of job control were associated with lower levels of participation in political and sporting activities. However, it is by no means clear that these spillover effects are caused by job characteristics rather than by other influences such as personal preferences for both low control and low activity. Furthermore, a more limited body of research supports an opposing hypothesis that employees may compensate for work stressors in their home environment and leisure pursuits (for a review see Kinman&Jones, 2001). Thiswould suggest, for example, that someone with lowcontrol at work might be expected to exert greater control in the home environment.
In addition to the spillover and compensation processes affecting the employees themselves in their home environment, it has further been suggested that work stressors may affect marital partners, a process known as crossover. For example, a number of studies have found one person’s work stressors to be associated with anxiety and depression in their spouse (e.g. Jones & Fletcher, 1993; Westman, 2001). This literature has seldom explicitly addressed the effects of job control. However, a longitudinal study by Stets (1995) specifically examines the effects of work autonomy in husbands and wives and finds that lack of autonomy not only leads to depression in the job holder, but can also lead people to compensate for the lack of control by controlling their spouse. This in turn is related to increased levels of depression in the spouse, suggesting a mechanism whereby job control at work may be implicated in the crossover of strain to partners.
Summary of Effects of Job Control
Overall, there is now mounting evidence for the importance of job control as a variable implicated in both increased coronary heart disease and reduced psychological well-being and it may even also impact on the well-being of marital partners. Nevertheless, there are many inconsistent studies that have failed to find effects and the mechanisms remain unclear. More particularly, however, there are conflicting findings in relation to the nature of the relationship. These are discussed in the following section.
Tags: aspects of physical health, Cardiovascular disease, concept of job control, decision latitude, Epidemiological studies examining morbidity and mortality, Health Behaviour, Impacts of Job Control on Home Life, increased coronary heart disease, low control jobs, neurotic disorder, Psychological Well-Being, short term physiological indicators, theory of job strain