Smoking and sedentary behavior as related to work organization

Smoking and sedentary behavior as related to work organization

Sot. Sci. Med. Vol. 32, No. 7, pp. 837-846, Printed in Great Britain SMOKING 0277-9536191 $3.00 + 0.00 Pergamon Press plc 1991 AND SEDENTARY BEHAV...

1MB Sizes 1 Downloads 31 Views

Sot. Sci. Med. Vol. 32, No. 7, pp. 837-846, Printed in Great Britain

SMOKING

0277-9536191 $3.00 + 0.00 Pergamon Press plc

1991

AND SEDENTARY BEHAVIOR TO WORK ORGANIZATION

AS RELATED

GUNN JOHANSSON,’JEFFREYV. JOHNSON~and ELLEN M. HALLS ‘Department of Psychology, University of Stockholm, S-106 91 Stockholm, Sweden and 2Division of Occupational Health, Department of Environmental Health Sciences, The Johns Hopkins School of Hygiene and Public Health, 615 North Wolfe Street, Baltimore, MD 21205, U.S.A. Abstract-There is little research which has investigated whether working life may a&t health behaviors. However, there is data suggesting that smoking as well as leisure activities are affected during times of stress. Both theoretical work and research suggests that work may socialixe people such that the use of leisure time for active pursuits, including exercise, may be contingent upon jobs which promote interaction, learning, and activity on the job. In investigating whether the psychosocial structure of work might affect smoking and sedentary behavior, a subsample (n = 7.201) of a representative sample of the Swedish population aged 1645 years was selected for study. Reports on job characteristics and health behaviors were obtained in personal or telephone interviews and a logistic regression analysis was performed. In general, job demands like shift work, piece work, hazardous exposure, and physical load tended to be associated with smoking and sedentary behavior, whereas job resources, including persona1 autonomy, were predictive of regular exercise, but unrelated to smoking behavior. Correlational patterns varied somewhat between sexes. The implications of these findings with respect to work organization, considerations in epidemiological research, and the conduct of health promotion programs are discussed. Key words-work

organization,

smoking, sedentary behavior, stress, job socialization

The purpose of this study was to explore the possible influences of work organization, work-related stress and work-related adult socialization on two health behaviors--cigarette smoking and level of physical activity during leisure time. To assess work organization characteristics we examine matters such as: control over work pace,

physical and psychological demands, job complexity, possibilities for social interaction on the job, duration of work cycles, etc. These job dimensions cut across occupations and allow a detailed analysis beyond that of occupational and white/blue collar categories (cf. Refs [l, 2)). Despite an increasing interest in health behaviors and a concurrent increase in occupationally situated wellness programs (e.g. Refs [3,4]), job characteristics are rarely taken into account when health promotion programs in the workplace are evaluated (e.g. Refs [S, 61). If the character of work is considered, it is usually treated in terms of white-collar vs blue-collar work or occupational classifications. In other cases, job characteristics such as extensive travel and short lunch breaks have been acknowledged as practical obstacles to participation in health promotion programs (61, but not as factors influencing the development of health-related behaviors. Thus, most of the relevant literature to date does not even permit comparisons by occupational categories, let alone by job characteristics.

psychobiological reactions, i.e. short-term stress reactions and their accumulated long-term effects in terms of health and well being. The other line of research is concerned with impacts of certain job characteristics on behavior patterns and personality. It is assumed that such impacts are mediated through that part of the adult socialization which takes place in working life. Relevant aspects of the two research areas are summarized below, and a graphic outline of potential factors involved is presented in Fig. 1. It must be emphasized that the figure only serves the purpose of describing hypothetical pathways leading from work organization to unhealthy behavior and that the present study does not allow a test of any one of the suggested pathways. Needless to say, smoking as well as sedentary behavior are regulated through a multitude of factors. Work organization is the focus of this study, because it has received little previous attention as a possible agent in the regulation of health-related behaviors.

THEORETICAL BACKGROUND

The theoretical foundation of this study is to be found in research on two different types of consequences of work organization (cf. Refs [7,8]). One focusses on the impacts of work organization on

Fig. 1. Potential factors mediating impacts of adverse job design characteristics on two health-related behaviors. 837

838

GUNN JOHANSWNet al

Work organization, stress, and health-related behaviors

Psychobiological reactions aroused through the process of occupational stress (Fig. 1) have been thoroughly investigated and reviewed (e.g. Refs [9111).A general assumption has been that work-related psychobiological stress reactions will add to the wear and tear of the organism and increase the risk of disease, in particular cardiovascular disease. An important part of occupational stress research are the epidemiological studies which have analyzed somatic health in relation to the specific psychosocial character of work [12-151. Other studies have analyzed work organization and health by survey technique or by including on-the-job measurement of psychobiological reactions. These analyses have generally treated healthrelated behaviors, such as smoking and sedentary behavior, as independent risk factors unrelated to work. There is, however, several reasons to assume that the stress process will affect both sedentary behavior and smoking behavior in ways which may either strengthen or weaken the associations observed between adverse psychosocial work conditions and disease. Occupational stress and sedentary behavior

High physical and mental job demands can lead to sedentary behavior since the demands will result in fatigue and a need for long recovery times after work (cf. Fig. 1). The ‘unwinding’ process, as measured by self-ratings and adrenaline excretion, has been shown to be prolonged during a period of excessive overtime [ 16, 171and after a day of intense and repetitive dataentry work [lS]. Self-reports indicate that recovery is also delayed in association with fragmented, machinepaced, repetitive, short-cycle industrial tasks [19,20]. Shiftwork and other irregular working hours are obvious factors affecting recovery time and leisure activity. In a study of retail-trade workers [21], 56% of those working irregular hours but only 19% of workers on regular day-time schedules reported that fatigue and practical difficulties was an obstacle to involvement in sports and exercise. Workers both on continuous and two-shift systems tend to participate less than daytime workers in political, trade union, social, and cultural activities [22,23]. Occupational stress and smoking

Several models have attempted to explain the initiation, maintenance, cessation and resumption of smoking behavior. Recent theories emphasize the complex interplay between psychological and biological factors (for a review, see e.g. Ref. (241). One of the most detailed discussions was presented by Leventhal and Cleary [25], who suggest a multiple-regulation model. They emphasize the role of emotional regulation rather than a purely pharmacological regulation of smoking behavior. According to this view, smokers regulate their nicotine intake not because of the drop in nicotine level per se, but rather because of the unpleasant emotional state associated with the sensation of craving. It is reasonable to assume that the need to avoid the unpleasant feeling should increase under feelings of distress. Therefore, occupational stress may also add to an urge for nicotine (cf. Fig. 1). Some researchers have suggested that smoking

behavior may be regulated by the smoker’s need to regulate arousal. Myrsten, Andersson, Frankenhaeuser and Elgerot [26] noted differences between those who relied on smoking in boring (low arousal) situations, and those who used it in stressful (high arousal) situations. They found that performance and well being were enhanced in boring situations, for those inclined to need nicotine for such purposes, and the same ‘needs fulfillment’ was observed for the group who smoke’d to calm down in high arousal situations. Therefore, it is not surprising that stress has been found to contribute to the intensification of smoking [27] and that smokers can withstand longer and more intensified stress when smoking [28]. One of the few studies which addressed the relationship between smoking behavior (smoking cessation) and psychosocial job characteristics [29] was based on the assumption that smoking serves to release tension. This assumption was confirmed insofar as quitters reported less quantitative work load and less deadline pressures than smokers. Quitters also reported less responsibility for persons and things than smokers. Thus, smoking may help the individual under stress to face demands and strain, but there is also evidence that individuals differ with respect to increasing or decreasing substance consumption in response to stressful conditions 1301. The relationship between smoking intensity and some, but not all, job stressors was supported in a study of 560 male members of kibbutzim [3 I]. When conditions were more favorable (e.g. involving high work-place social support) there was also an increased probability for smoking cessation; persons reporting low support remained smokers significantly more than those with high support. Similar findings were reported in a recent study of health behaviors in 800 employees of a large chemical plant [32,33]. Smokers of high strain jobs tended to smoke more heavily than those in low strain positions. Again, a positive association was observed between co-worker support and quitting among those with higher education, but a negative association was found for those with less education. One study, reported by House [34], has failed to find any significant correlation between objective job stress and smoking behavior. Thus, the literature to date is suggestive, but not conclusive concerning the possibility that occupational stress may influence cigarette smoking behavior. It should be noted that these reports are based on male smokers. For this group, there is evidence that an increase in the pressure of work may lead to an increase in cigarette consumption, but personality also appears to be a factor in this relationship. When analyzing smoking behavior in relation to conditions, a distinction should be made between factors which cause non-smokers to begin smoking and factors that influence the amount of nicotine consumed or the cessation of smoking. In western industrialized countries, smoking behavior is often established before the individual enters the labor market. Therefore, it may be generally assumed that occupational factors exert little impact on smoking per se. Seen in that perspective, it is more likely that work conditions will influence the amount smoked and/or the maintenance of the smoking habit

Health related behaviors and work organization rather than whether the individual smokes or not. The possibility that occupational conditions might cause non-smokers to begin, or to maintain smoking, is beyond the scope of the present paper, but it is possible that persons who entered the work force 20-30 years ago may have been influenced in their attitudes towards and decision concerning smoking by occupational conditions and/or co-worker pressure. Certainly, restrictions on smoking in many work places may function to discourage workers from initiating as well as from maintaining this behavior. ADULT

SOCIALIZATION

Work organization and active/passive life-styles

There is increasing evidence that the relation between work and non-work activities is characterized by an impact of work on leisure rather than an impact in the opposite direction [35]. According to empirical evidence there is a ‘spill-over effect’ of job experience into leisure rather than a ‘compensation effect’ by which a dull job is counterbalanced by rich and active leisure-time activities [36]. This approach is represented by the lower branch of relations in Fig. 1. It has been stimulated by research showing that individuals with jobs which are demanding, and which also seriously limit autonomy and social interaction at work, take less part in organized and goal-oriented activities outside work which require planning and cooperation with others (for review, see [22]). Further evidence supporting the role of work organization in adult socialization has been found in experiments with autonomous work groups, where jobs were redesigned to require more skill and to provide more variety (e.g. Refs [37-391). Within a few years, some of these experiments resulted in increased worker interest in both internal activities (i.e. work place problems including health and safety concerns) and external activities (increased involvement in community and trade union affairs). As a consequence, increasing attention has been paid to work organization as a factor in life-span psychological development [35]. To a considerable extent, personal characteristics lead the individual to select certain jobs. Therefore, it has long been assumed that any relations between job characteristics and behavior outside work could be wholely explained with reference to personality characteristics [40]. Thus, stable personality characteristics would be responsible for the choice of a lowautonomy job as well as a passive life-style. However, in a dynamic labor market, performing a stimulating job which promotes personal development and skills may make the individual prepared to seek a new job with a different job demand profile. This, in turn, will influence personal and occupational development. In this interactionist perspective on personality [41] it is equally apparent that monotonous, fragmented tasks, with little opportunity for development of skills, may induce a passive attitude, a low probability for the individual to seek new jobs, and maintain a continued passivating job socialization process. Separating the effects of job selection and of job socialization is a major difficulty. However, longitudinal data suggest that the socialization factor oper-

839

ates in relationship to work organization and activity level (Fig. I). Kohn and Schooler [42] reported results from a IO-year follow-up survey of a representative sample of U.S. males in civilian employment. They related job complexity (requirements for thought and independent judgment) to the intellectuality of leisure activities and found several job conditions, especially job complexity, to affect leisure time intellectuality. Although they ascribe a considerable part of the correlation to individual characteristics and disposition, they conclude that job characteristics do play a role [431. Job socialization and health-related behaviors

There are at least two kinds of evidence to support the existence of a link between job socialization and health-related behaviors. One type of evidence has been collected in tests of the ‘job strain model’ and the other type in research on ‘learned helplessness’. The job strain model. Adult socialization on the job is a central mechanism in the job strain model suggested by Karasek [44-461. The model identifies two dimensions of work organization, the stress/ demand dimension and the autonomy dimension. The two dimensions combine into four major categories of work organization. Karasek predicts that exposure to high job demands may lead to either of two outcomes depending on the level of decision latitude associated with the job: high job demands and little autonomy results in accumulation of unresolved strain with impaired health as a possible long-term consequence. More importantly in this context, the model also predicts that an active lifestyle is observed most often when high job demands (demands for speed, attention, problem solving, etc.) are combined with individual control and autonomy (influence over the pace of work, work routines, decision at work unit level, etc.). The association between passive work and a passive life-style has received considerable support [46,47] which is most convincing in a longitudinal study based on the Swedish Level of Living Survey [44,45]. It was found that those workers whose jobs changed between 1968 and 1974 towards richer job content and more autonomy had also increased their participation outside work in voluntary associations, in study, and in trade union and political activities. Learned heIpIessness. A socialization process in the work setting can be described in terms of a learning perspective, with reference to the concept of learned helplessness [48]. In a work situation with minimal possibilities for individuals or groups to control or alter aversive factors such as monotonous work routines, accident risks, environmental hazards, etc., the individual will experience the quality of the work environment as relatively independent of his own and his peers’ behavior. These conditions might lead to resignation and ‘helplessness’, and, in turn, passivity. Such a mechanism is likely to influence non-work behavior not only via a decrease in behavioral activity but also by a decrease in self-efficacy [49, 501. Poor general self-efficacy will most certainly affect people’s perception of their own capability to influence their own health status (Fig. 1). The research reviewed is in line with German research initiated by Hacker [51] and Volpert [52].

GUNNJOHANSSON et al.

840

This work concludes that problem solving and opportunities for learning on the job are crucial elements in adult intellectual development. It is therefore reasonable to assume that the character of work might be one of several factors influencing people’s healthrelated behavior as well as their perceived capability and readiness to modify such behaviors in response to health promotion programs. Study hypotheses

Job demands Psychological demands. This is a Guttman type scale, with the following items scored 0 = no, 1 = yes, and summated. l

The above discussion suggests that work conditions may influence smoking and exercise behavior. Although major mechanisms behind such possible associations are likely to be different for the two behaviors, Fig. 1 suggests that some mechanisms may be involved in both behaviors. Therefore, smoking and sedentary behavior will both be regarded as examples of adverse health behavior, and the following two hypotheses will be addressed in the present investigation: (1) Work characteristics that may serve to increase stress reactions, i.e. job demands, will increase the likelihood of engaging in adverse health behaviors. (2) Work characteristics that provide the possibilities for control, learning, and social involvement, i.e. work resources, will decrease the likelihood of engaging in adverse health behaviors. MATERIALS

structing subscales within these conceptual categories. The number of factors was restricted using the Kaiser criteria of an eigenvalue greater than or equal to 1 [53].

AND METHODS

Study sample

The data used was the Swedish Central Bureau of Statistics (SCB) Survey of Living Conditions (‘ULF’ in Swedish). This on-going survey was mandated in the early 1970’s by the Swedish Parliament as the major social accounting system to investigate the distribution of health status, income, education and aspects of the social and work environment. Over the past 15 years, considerable effort has been devoted to constructing reliable and valid indicators. During the 1970’s the Living Conditions Survey was conducted annually, and in the 1980’s biannually. A systematic random sample of individuals born on the 15th of each month was obtained from the Swedish National Registry of Births. Spouses were also interviewed. The present study used the 1977 sample, which included information on both work environment characteristics and health behavior. In 1977 the sample consisted of 14.500 and the response rate was 81%. The survey data were collected in a one hour personal or telephone interview performed by a trained SCB interviewer. The present investigation used a sub-sample of 7.201 subjects consisting of employed persons from the ages of 16 to 65; 3656 (50.8%) were males and 3545 (49.2%) were females. Males had a mean age of 39 and females a mean age of 38. Measurement of job characteristics

Two basic categories of work environment characteristics were investigated: Job demands and work resources. Principal component factor analysis with orthogonal rotation was used as the basis for con-

l

Is your work hectic? Is your work psychologically

strenuous?

Scores range from 0 to 2, with a mean of 0.91 and a standard deviation of 0.79. The Guttman coefficient of reproducibility is 0.92 and the coefficient of scalability is 0.79. The scale reliability coefficient is 0.60. Physical load. The following items were scored 0 = no, 1 = yes, and summated. l

l l

l l

Does your work require bent, twisted or otherwise unsuitable working postures? Does your work require heavy lifting? Does your work require you to perspire daily from physical exertion? Do you get dirty in your work? Do you think your work is safe from the point of view of accident risk?

Scale scores range from 0 to 5 with a mean of 1.57 and a standard deviation of 1.57. The Cronbach’s alpha reliability coefficient is 0.73. Hazardous exposure. Subjects were asked if they were exposed to and bothered by the following working conditions on their jobs: l l l l l l l

Noise Heavy shaking or vibrations Cold Drafts Inadequate ventilation Bad lighting Gas, mist or smoke

Items were scored as yes = 1 and no = 0 and summated. Scale scores range from 0 to 7 with a mean of 1.65 and a standard deviation of 1.75. The Cronbach’s alpha reliability coefficient is 0.71. Monotonous work. This is a Guttman type scale with the following items scored 0 = no, 1 = yes, and summated. l l

Is your work monotonous? Does your work involve many repeated and one sided movements?

The scale scores range from 0 to 2 with a mean of 0.62 and a standard deviation of 0.75. The Guttman coefficient of reproducibility is 0.95 and the coefficient of scalability is 0.84. The scale reliability coefficient is 0.78. Shift work. This is a single item indicator which was not included in the factor analysis. Subjects were asked the following question; l

Which part of the 24 hr day do you work, or which type of shift work do you have? 1 = daytime 2 = evening or night time 3 = 2-shift operation 4 = continuous 3-shift operation

Health related behaviors and work organization 5 = intermittent 3-shift operation 6 = working hours, irregularly distributed over the 24 hr day and over the week, according to a special working plan or scheme. If the subject responded with items 1 or 2 they were given a score of 0, indicating non-shift work. If they responded with items 3-6, they were given a score of 1, indicating shift work. 856 individuals or 12% of the working population were categorized as shift workers. Piece work. This is a single item indicator, not included in the factor analysis. Subjects responded yes (1) or no (0) to the following question: l

Is any of your work income based on a piece rate payment system?

1177subjects or 16.4% of the sample indicated that they worked some proportion of their time under a piece rate payment system. Work resources Personal freedom. The following items were scored according to whether the subject had never (0), sometimes (1) or often (2) any of the following: l l l l l

l

Influence over how time is used in work The planning of work breaks The planning of vacations Flexible working hours Freedom to receive a phone call during working hours Freedom to receive a private visitor at work

Scores range from 0 to 12 with a mean of 6.05 and a standard deviation of 2.73. The Cronbach’s alpha reliability coefficient is 0.59. Learning opportunities. The following items were scored according to whether the subject had never (0), sometimes (1) or often (2) any of the following: l

l

l l l

Possibilities for on-going education as part of the job The experience of personal fullfilment on the job The possibility of learning new things Varied task content Varied work procedures

Scores range from 0 to 10 with a mean of 5.89 and a standard deviation of 2.99. The Cronbach’s alpha reliability coefficient is 0.61. Work process control. The following items were scored according to whether the subject had never (0), sometimes (1) or often (2) any of the following: l l l l

Influence over the planning Influence over the setting of Influence over the selection Influence over the selection

of work the work pace of supervisor of co-workers

Scores range from 0 to 8 with a mean of 3.31 and a standard deviation of 1.69. The Cronbach’s alpha reliability coefficient is 0.59. Social interaction. The following dichotomous items were scored yes = 1 and no = 0. Respondents were asked whether they: l

Could talk to co-workers during breaks

SSM 32.7-G

l l

l

l

841

Could leave their job to talk with co-workers Could interact with co-workers as part of their work Met with co-workers, outside of the work place Had met with a co-worker during the last 6 months

Scores range from 0 to 5 with a mean of 3.34 and a standard deviation of 1.49. The Cronbach’s alpha reliability coefficient is 0.75. Measures

of health behavior

In order to facilitate statistical analyses according to parallel structures for both health behavior variables, both were dichotomized as described below. Cigarette smoking. Respondents were asked whether they smoked daily, whether they smoked cigarettes, a pipe, or cigars and approximately how much their daily consumption was. Individuals were categorized as being a smoker if they smoked more than one cigarette on a daily basis. In the study sample 38.8% or 2776 subjects were defined as cigarette smokers and given a score of 1. The remaining 61.2% of the sample were defined as not being cigarette smokers and were given a score of 0. Sedentary behavior. Respondents were asked the following question concerning physical exercise during non-working time: I would like to know how much physical exercise you get during your leisure time. Which of the following alternatives are the closest to your own situation? Hardly any physical exercise at all Exercise now and then. (For example, an occasional walk, going skiing now and then, swimming every so often, going out to pick mushrooms, and the like.) Regular exercise. (For example swimming, jogging, skiing, mountain climbing, brisk walking, cycling, or similar activities at least once a week.) As the primary concern of the present investigation was on regularly performed fitness promoting physical activity, individuals who had responded to the question by indicating either number 1 or 2 above were considered to be sedentary, and were given a score of 1. Individuals who indicated that they performed regular exercise were given a score of 0 on the sedentary behavior variable. Sixty percent of the sample was categorized as sedentary using this criterion. Control variables

Both age and educational level were used as control variables in the analysis as they were correlated with both the work characteristics and the health behaviors. Age was measured in years. Education was measured by an eight level indicator of completed formal schooling. The analysis was performed separately for males and females. Statistical

methods

The data were analyzed using a logistic regression program developed by Harrell[54]. The health behaviors were coded as binary (O-l) dependent variables,

GUNN JOHAN~~ON et

842

and a model which included age, education, and a work characteristic was fitted using maximum likelihood methods computed by the Newton-Raphson method. Standardized odds ratios (SOR) were calculated using the logistic coefficient (B) and the population standard deviation (SD) of that characteristic and applying the following formula: SOR = e@x SD) The SOR measures the increase or decrease in the odds of being a smoker, or being sedentary, for each standard deviation change in the work characteristic [55]. If the association between the job characteristic and the health behavior was negative, a minus sign was used rather than presenting the SOR reciprocal, in order to increase the comparability of the relative magnitude of the relationships. The total population standard deviation was used in calculating the SOR rather than sex specific standard deviations, in order to make the metric of measurement comparable between males and females. An estimation of an odds ratio (OR), which compares the top to the bottom decile on the distribution of the work characteristic, was calculated following the procedure used by Karasek et al. (see Ref [55]). Finally, a multi-variable logistic regression analysis was performed which included age, education, and all of the work characteristics found to be statistically significant in the bivariate analysis for each behavior and sex specific case. RESULTS

As Table 1 indicates, in the age and education adjusted bivariate analyses, physical load, hazardous exposure, shift work and piece work were significantly associated with an increased likelihood of being a smoker for male workers. The availability of resources, however, seemed to have little effect on smoking prevalence among men. When the work characteristics, which were statistically significant in the bivariate analysis, are conTable 1. Bivariate logistic regression analysis of work charactertstics and smoking prevalence for males’

Work characteristic

Logistic coefficient

P

0.0523 0.0675 0.0367 0.0365 0.2536 0.2705

0.234 0.002 0.047 0.454 0.010 0.001

-0.0010 -0.0128 0.0265 0.0355

0.941 0.302 0.195 0.21 I

SOR’

Decile to decile OR’

Demati:

Psychological demands Physical load Hazardous exposure Monotonous work Shift work Piece work

1.04 I.11 I .07 I .03 1.29’ 1.31’

1.14 1.41 1.25 I.10 -’ -J

Resources:

Personal freedom Learning opportunities Work process control Social interaction

1.00 - 1.04 I .05 I .05

1.00 - I.13 1.16 1.19

‘Adjusting for the covariates age and educational level; n = 3627, with 29 observations deleted due to missing values in each of the separate analyses, except for that of piece work where n = 3619, with 37 observations deleted due to missing values. *Standardized Odds Ratio, using the total population standard deviation for each work characteristic. ‘Contrasts the highest with the lowest decile on the distribution of the exposure characteristic. ‘For dichotomous variables the SOR reverts to tbe OR which contrasts the presence versus the absence of the exposure characteristic. ‘Not calculable in the case of a dichotomous variable.

al.

Table 2. Multi-variable logistic regression analysis of smoking prevalence for males, n = 3615’

Independent variables

Logistic coefficient

Age Education Physical load Hazardous exposure Shift work Piece work

-0.0243 -0.2418 0.0517 0.0064 0.2359 0.2134

P

0.000 O.ooO 0.063 0.784 0.018 0.014

SOR’ -1.35 -1.19

1.08 I.01 1.27’ 1.24’

Decile to docile OR’ -2.66 -1.78 1.31 1.04 -* -’

‘37 observations deleted due to missing values. ‘Standardized Odds Ratio, using the total population standard deviation for each variable. ‘Contrasts the highest with the lowest decile on the variable distribution ‘For dichotomous variables the SOR reverts to the OR which contrasts the presence versus the absence of the exposure characteristic. ‘Not calculable in the case of a dichotomous variable.

sidered simultaneously in a multi-variable model (see Table 2), shift work and piece work continue to be statistically significant (P < 0.02), physical load is of borderline significance (P = 0.063), and hazardous exposure is no longer independently associated with smoking prevalence for men. The strength of the associations, however, is only moderately reduced for shift and piece work, and physical load. The pattern of associations was somewhat different for female workers, as is shown in Table 3. In the age and education adjusted bivariate analysis, for women, all of the job demands were significantly associated with the odds of being a smoker. For both physical load and hazardous exposure the strength of the association is substantially greater for women than for men. In terms of work resources, women with a greater degree of social interaction at work are more likely to be smokers than those with low levels of social interaction. In the multi-variable model (see Table 4), social interaction continues to have an independent association with increased likelihood of being a smoker among women. Physical load also has a statistically significant (P c 0.05) association with smoking prevalence. There is also an indication that monotonous work has an independent association with smoking prevalence as it is very close to being statistically significant (P < 0.0522). Although psychological Table 3. Bivariate logistic regression analysis of work characteristics and smoking prevalence for females’

Work characteristic

Logistic coefficient

P

0. I530 0.1110 0.0825 0.1417 0.2419 0.2537

0.001 0.000 0.002 0.004 0.040 0.032

-0.0022 -0.0129 0.0000 0.0703

0.975 0.328 0.999 0.002

SOR’

Decile to decile OR’

Demands:

Psychologtcal demands Physical load Hazardous exposure Monotonous work Shift work Piece work

1.13 1.19 1.16 I.11 1.27’ 1.29’

1.50 1.77 I .63 I.41 -’ -s

Resources:

Personal freedom Learning opportunities Work process control Social interaction

-1.01 -1.04 1.00 I.11

-1.02 -1.14 1.00 I .41

‘Adjusting for the covariates age and educational level; n = 3533 with I2 observations deleted due to missing values in each of the separate analyses, except for that of piecework whefe n = 3527 with 18 observations deleted due to missing values. 2-‘See Table 1.

Health related behaviors and work organization Table 4. Multi-variable logistic regression analysis of smoking prevalence for Swedish females, n = 3527’

Logistic Work characteristic

coefficient

Age

-0.0454 -0.3483 0.0873 0.0734 0.0167 0.0952 0.0855 0. I252 0.0638

Education Psychological demands Physical load Hazardous exposure Monotonous work Shift work Piece work Social interaction

P 0.000 0.000 0.080 0.012 0.568 0.052 0.482 0.310 0.005

SOR2 -1.74 -1.28 I .07 1.12 I .03 I .08 1.09’ 1.13’ 1.10

Decile to decile OR’ -6.20 -2.29 I .25 1.46 1.10 1.28 -3 -5 I .37

843

Table 6. Multi-variable logistic regression analysis of sedentary behavior prevalence for Swedish males, n = 3636’ De&

Logistic Independent variable

coefficient

Age

0.0087 -0.1308 0.0551 0.1090 0.1166 0.0105 -0.0417 -0.0406 - 0.0957

Education Hazardous exposure Monotonous work Piece work Personal freedom Learning opportunities Work processcontrol Social interaction

P 0.002 0.013 0.005 0.154 0.184 0.485 0.042 0.094 0.001

to decile SOR’ OR’ I.11 -1.10 1.10 1.08 1.12’ I .03 -1.13 -1.07 -1.15

1.40 -1.37 1.37 1.31 -J 1.09 -1.51 -1.25 -1.60

’ I8 observations deleted due to missing values. ‘-‘See Table 2.

‘20 observations deleted due to missing values. ‘-‘See Table 2.

demands continues to be positively associated with smoking (SOR = 1.07) it is of borderline statistical significance (P < 0.10). In the multi-variable analysis

that the significant predictors of sedentary behavior, hazardous exposure, learning opportunities, and social interaction, have SORs greater than or equal to that of age and education. The results of the analysis of the bivariate relationship between work characteristics and sedentary behavior among women workers are shown in Table 7. Females with psychological demands, physical load, monotonous work, shift work and piece work are more likely to be sedentary during their leisure time. Learning opportunities and work process control are associated with a decreased odds of sedentary behavior. In contrast to the effect of social interaction on physical activity level found in men, for women there is no statistically significant relationship. When those work characteristics which were statistically significant in the bivariate analysis are considered simultaneously (see Table 8), the only characteristics significantly associated with sedentary behavior for women were psychological demands (P < 0.05) and learning opportunities (P < 0.01). Although monotonous work, shift work, and work process control continue to show effects in the hypothesized direction, their P values were greater than 0.10. Physical load does not appear to be independently associated with sedentary behavior for women. Again, we note that both psychological demands and learning opportunities are more strongly associated with sedentary behavior than is age or education.

hazardous exposure, shift and piece work, do not demonstrate significant independent associations with smoking prevalence. The findings for the bivariate associations between work characteristics and the prevalence of sedentary behavior among male workers are shown in Table 5. Men with monotonous work, hazardous exposure and piece work have a greater odds of being sedentary during leisure time. Work resources are all negatively associated with the likelihood of sedentary behavior, indicated by negative SORs and decile to decile ORs. Social interaction at work has the strongest association with decreased odds of being sedentary (SOR = - 1.19) followed by learning opportunities (SOR = - 1.15). The associations between social interaction and learning opportunities and sedentary behavior prevalence is only slightly diminished in the multi-variable analysis (see Table 6). Hazardous exposure also continues to have an independent, statistically significant association with sedentary behavior among men (P < 0.01). Although monotonous work and piece work continue to show associations, they are no longer statistically significant in the multi-variable model. Work process control is associated at only a borderline level of significance (P < 0.10). We note Table 5. Bivariate logistic regression analysis of work characteristics and the prevalence of sedentary behavior during non-work time for males’

Work characteristic Demands: Psychological demands Physical load Hazardous exposure Monotonous work Shift work Piece work Resources: Personal freedom Learning opportunities Work processcontrol Social interaction

Logistic coefficient 0.0177 0.033 I 0.0697 0. IO26 0.1004 0.2121 -0.0256 -0.0464 -0.0719 -0.1147

P 0.677 0.126 0.000 0.033 0.309 0.01 I 0.043 0.000 0.000 0.000

SOR’ 1.01 1.05 I.13 I .08 1.11’ 1.24’ -1.07 -1.15 -1.13 -1.19

Decile to decile OR’ 1.03 1.17 I .50 I .29 -5 -’ -1.25 -1.58 -1.49 -1.75

‘Adjusting for the covariates age and educational level; n = 3644 with I2 observations deleted due to missing values in each of the separate analyses, except for that of piece work where n = 3536 with 20 observations deleted due to missing values. ‘“See Table I.

Table 7. Bivariate logistic regression analysis of work characteristics and sedentary behavior prevalence during non-work time for females’

Work characteristic Demands: Psychologicaldemands Physical load Hazardous exposure Monotonous work Shift work Piece work Resources: Personal freedom Learning opportunities Work processcontrol Social interaction

Decile to decile

Logistic coefficient

P

SOR’

OR’

0. I298 0.0581 0.0399 0.1042 0.2482 0.1931

0.005 0.025 0.127 0.03 I 0.040 0.113

I.11 1.10 1.07 I .08 1.28’ 1.21’

I .41 I .37 I .25 I .29 -s -I

0.315 0.000 0.003 0.352

- 1.04 -1.15 -1.12 -1.03

-0.0135 -0.0476 -0.0685 -0.0201

-1.13 -1.58 -1.46 -1.10

‘Adjusting for the covariates age and educational level; n =

3538 with 7 observations deleted due to missing values in each of the separate analyses except for that of piece work where n = 3531 with 14 observations deleted due to missing values. *-*See Table I.

GUNN JOHANSSON et al.

844

Table 8. Multi-variable logistic regression analysis of sedentary babavior orevalence for Swedish females, n = 3538’

Independent variables

Logistic coefficient

Aee E&cation Psychological demands Physical load Monotonous work Shift work Learning opportunities Work process control

-0.0049 -0.1293 0.1235 0.0153 0.1283 0.1716 - 0.0680 - 0.0392

SOR2

P

0.099 0.003 0.011 0.590 0.108 0.169 0.002 0.106

-1.06 - 1.09 1.10 I .02 1.10 1.31’ -1.23 - I .07

Decile to decile OR’ -1.22 -1.36 1.38 I .08 1.38

_.

-1.95 -1.24

‘7 observations deleted due to missing values 2-5See Table 2.

SUMMARY AND DISCUSSION

The results of this investigation lend support to the hypothesis that structural aspects of the work environment are related to health-related behaviors of workers. Practically all relations which were observed between work conditions and health behaviors were in the expected direction: job demands usually associated with unhealthy behavior and job resources with healthy behavior. We note, however, that the patterns of association were different for smoking and sedentary behavior as well as for men and women. Therefore, and in view of the low to moderate size of the associations, the results do not allow definite statements about the impact of the work environment relative to other, non-occupational factors in the regulation of health behaviors. Smoking

Daily smoking behavior tended to be positively associated with job demands but unrelated to work resources. Thus, for both sexes physical load, hazardous exposure, shift work, and piece work were associated with smoking behavior. Among women, psychological demands and monotonous work were also significantly associated with smoking. Only one of four work resource factors, the opportunity for social interaction, was significantly correlated with smoking, and only among women. But instead of decreasing the likelihood of adverse health behaviors, as stated in our hypothesis, opportunities for social interaction slightly increased the risk for smoking behavior among women. This finding is not easily explained on the basis of our initial theoretical considerations. It may be that in female-dominated work (nursing, secretarial work and teaching) smoking occurs during break periods, when co-workers gather to have coffee and socialize. Earlier studies relating work organization to healthrelated behaviors have been concerned with smoking only. The present investigation corroborates their findings of a positive relationship between occupational stress and smoking [ 13,31,32]. However, of the few studies that exist, most have either been restricted to particular occupations, to one sex, or to relatively small samples. The primary strength of the present investigation is that it is based on a random sample of an entire working population; it includes both sexes and nearly all occupations.

Sedentary

behavior

Sedentary behavior was positively associated with hazardous exposure and monotonous work among men as well as women. For women, both psychological demands and physical load also increased the likelihood of being sedentary, and for men piece work was so associated. Shift work, however, was entirely unrelated to sedentary behavior, a finding which contradicts the assumption that slow recovery from shift work would be an obstacle to regular physical exercise. Although job resources like personal freedom, learning opportunities, work process control and social interaction were largely unrelated to smoking, they were significantly and negatively associated with sedentary behavior. Thus, in the male group all job resource factors decreased the likelihood of being sedentary. In the female sample this was true only of learning opportunities and work process control. The results are in agreement with a recent suggestion by Peterson and Stunkard [56] of personal and collective control as crucial concepts for a theory of health promotion. Patterns

of associations

Although none of the individual work characteristics were consistently associated with both health-related behaviors among both sexes, we see that shift work, hazardous work, physical load and monotonous work are positively associated with one or both of the health behaviors in both sexes in the bivariate analysis. These demands are similar in that they (1) involve physical strain and (2) are relatively impervious to personality and individual proclivities and (3) are less prone to subjective interpretation. One reason might be that the wording of questionnaire items such as ‘do you sweat’, ‘do you do shift work’, etc., is relatively unambiguous, and therefore less likely to involve subjective appraisal. In contrast, the psychological demands indicator does involve subjective interpretation. The worker is required to set his/her own criteria for evaluation, and to assess whether or not his/her overall occupational situation is ‘hectic’, ‘strenuous’, etc. There are several differences in the patterns of association observed for males and females. This may be due to differences in the character of work performed by males and females. If the range of job demand and job resource measurements is different for males and females (e.g. males facing generally higher physical loads and higher levels of control, cf. Ref. [57]) non-linear regression might result in different correlational patterns. Such differences in patterns may also be due to a sex difference in the appraisal of work situations, and/or in the willingness to report distress. Possibly women and men experience, and/or respond and/or report differently with respect to their work conditions. There may even be sex differences with respect to the ‘triggers’ or ‘pathways’ which link psychological demands to health behaviors. Due to the cross sectional nature of this investigation we do not claim that the observed associations are evidence of a causal relationship. In order to document such causality, it would be necessary to conduct longitudinal research and to document work conditions and health behaviors over time, preferably over the life course. More objective and detailed measures of both the work conditions and health

Health related behaviors and work organization behaviors would be necessary. While it is unlikely that the reporting of either activity in this study is inaccurate, the information is relatively unspecific. It is also unlikely that smoking or sedentary behavior has contaminated the reporting of information about occupational conditions. However, although our data were adjusted for educational level, we cannot eliminate the possibility that smokers and sedentary individuals may have selected themselves into adverse work situations, by virtue of some other common factor or factors, such as personality and/or class background. The theoretical framework of this study (Fig. 1) assumes several possible pathways which may lead from psychosocial work conditions to health behaviors. The character of the present data limits the possibilities to explore these pathways, but the study offers questions and observations worth pursuing in future research. It is obvious, for example, that the two healthrelated behaviors recorded in this study differ with regard to the decisions and behaviors involved. Smoking occurs throughout the day, in the workplace, and at home and can be used in a fashion which augments occupational functioning. Not only can nicotine be used to modify arousal levels, in response to stressful conditions, it can be used as a vehicle for ‘controlling time’ (e.g. if a worker is smoking, it is clear that he/she is not ‘just sitting around’). Exercise, on the other hand, is an intermittent behavior which occurs outside of the workplace. Although smoking cessation requires some planning, it does not take extra time and the planning is temporary and minor compared to the planning involved in regular exercise: planning of time, special clothing, commitment to be at a certain place and so forth. Neither exercise nor sedentary behavior act to augment occupational functioning in a short-term perspective. As mentioned above, the strength of this study is in breadth rather than depth, and it is beyond its scope to further analyze the mechanisms by which occupational factors might affect health behaviors. Future research needs to consider smoking intensity, the context in which smoking occurs, the number of years of smoking, and quitting/relapse in smokers. The amount and type of activity which are performed among those who exercise, and further information on ‘lapsed’ or seasonal exercise would also be of value. Finally, another factor worth considering, is the meaning and importance of a specific health behavior to specific cultures and values developed within an occupation or an organization (weight reduction among professional dancers, smoking among teachers vs taxi drivers, etc.). To return to an earlier observation, concerning the origins of health behaviors, while our data do not support a strong, causal linkage between all major aspects of work organization and health behaviors, they do suggest that it may be useful to speculate more deeply concerning such behaviors and to more critically inquire into various sources of reinforcement for and discouragement of health behaviors. In short, despite the limitations of the present investigation we suggest that the topic of work organization and health behaviors deserves further attention, more refined theory, and careful empirical research.

845

Acknowledgements-This work was supported by a grant from the John D. and Catherine T. MacArthur Foundation (Karolinska node of the network on Determinants and Consequences of Health-Damaging Behavior). It was facilitated by support to Gunn Johansson from the Swedish Council for Research in the Humanities and Social Sciences, and to Jeffrey Johnson from the Swedish Work Environment Fund (grant No. 86-119) and The National Institute of Health (grant No. NHLBl R29 HL39633).

REFERENCES 1. McQueen D. V. and Siegrist J. Social factors in the etiology of chronic disease: An overview. Sot. Sci. Med. 16, 353-367, 1982. 2. Karasek R. A., Schwartz J. and Theorell T. Job charac-

3. 4. 5.

6.

7.

8.

9.

teristics, occupation and coronary heart disease. Final report to the National Institute for Occupational Safety and Health, U.S. Department of Health and Welfare, Contract No. R-Ol-0HO0906. Department of Industrial Engineering and Operations Research, Columbia University, 1982. Cataldo M. F. and Coates T. J. (Eds) Health and industry. A Behavioral Medicine Perspective. Wiley, New York, 1986. Feldman R. H. L. (Ed.) Health Promotion in Worksettings. Special Issue: Hifh Educ. Res. 2, l-71, 1987. Wilbur C. S., Hartwell T. D. and Piserchia P. V. The Johnson & Johnson Live for Life program: Its organization and evaluation plan. In Health and Industry. A Behavioral Medicine Perspecfive (Edited by Cataldo M. F. and Coates T. J.), pp. 338-350. Wiley, New York, 1986. Naditch M. P. Staywell: Evolution of a behavioral medicine program in industry. In Health and Industry. A Behavioral Medicine Perspective (Edited by Cataldo M. F. and Coates T. J.), pp. 323-337. Wiley, New York, 1986. Johansson G. and Garde11 B. Work-health relations as mediated through stress reactions and job socialization. In Topics in Health Psychology (Edited by Maes S. et al.), pp. 271-285. Wiley, New York, 1988. Johnson J. V. Control, collectivity and the psychosocial work environment. In Job Stress and Work Control (Edited by Sauter S., Hurrell J. and Cooper C.), pp. 55-74. Wiley, London, 1989. Levi L., Frankenhaeuser M. and Garde11 B. Work stress related to social structures and processes. In Stress and Human Health. Analysis and Implications of Research

(Edited by Elliott G. R. and Eisdorfer C.), pp. 119-146. Springer, New York, 1982. 10. Kasl S. V. and Cobb S. Psychological and social stresses in the work place. In O&palional Health (Edited by Levy B. S. and Wenman D. H.1. I. DD. _. 251-263. Little Brown, Boston, 1985. 11. Celentano D. and Johnson J. V. Stress in health care workers. Occup. Med.: State of the Art Rev. 2, 593-608, 1987. 12. Alfredsson L., Karasek R. and Theorell T. Myocardial

infarction risk and psychosocial work environment characteristics: An analysis of the male Swedish work force. Sot. Sci. Med. 16, 463-467, 1982. 13. Caplan R., Cobb S., French J. R. P. Jr, Van Harrison R. and Pinneau S. R. Jr Job Demands and Worker Health. Washington, D.C., National Institute for Occupational Safety and Health, HEW Publication No. (NIOSH) 75-160, 1975. 14. Johnson J. V. and Hall E. M. Job strain, work place social support and cardiovascular disease: A cross sectional study of a random sample of the Swedish working population. Am. J. publ. Hlth 78, 1336-1342, 1988. 15. Johnson J. V., Hall E. M. and Theorell T. The combined effects of job strain and social isolation on cardiovascular

GUNN JOHJLNS.WN et al.

846

disease morbidity and mortality in a random sample of the Swedish male working population. Stand. J. Work, Environ. Hlth 15, 271-279,

Int. J. Hlth Serv. 12, 527-558,

1989.

16. Rissler A. and Elgerot A. Stressreaktioner vid iivertidsarbete. (Stress reactions related to overtime at work.) Department of Psychology, University of Stockholm, Rapporter, nr 23, 1978. (In Swedish.) 17. Frankenhauser M. Psychoneuroendocrine approaches to the study of emotion as related to stress and coping. In Nebraska Symposium on Motivation 1978 (Edited by Howe H. E. and Dienstbier R. A.), pp. 123-161. University of Nebraska Press, Lincoln, 1979. 18. Johansson G. and Aronsson G. Stress reactions in computerized administrative work. J. Occup. Behav. 5, 159-181, 1984. 19. Garde11 B. Psychosocial aspects of industrial production methods. In Society, Stress and Disease, Vol IV: Working Life (Edited by Levi L.), pp. 65-75. Oxford University Press, Oxford, 1981. 20. Johansson G., Aronsson G. and Lindstrljm B. 0. Social psychological and neuroendocrine stress reactions in highly mechanized work. Ergonomics 21.583-599, 1978. 21. Garde11 B., Nilsson C., Bostrand L., Gehlin J. and Magnusson M. Sociala effekter av arbetstidsfiirhallanden inom detaljhandeln (Social consequences of work hours in the retail trade.) Research Unit for the Social Psychology of Work, Department of Psychology, University of Stockholm, Report No. 4, 1976. (In Swedish.) 22. Garde11 B. Scandinavian research on stress in working life. Int. J. Hlth Serv. 12, 31-41, 1982. 23. Cox T. Repetitive work. In Current Concerns in Occupational Stress (Edited by Cooper C. L. and Payne R.). Wiley, Chichester, 1980. 24. Feuerstein M., Labbe E. E. and Kuczmierczyk A. R. Health

Psychology.

A

Psychobiological

Perspective,

pp. 381-424. Plenum Press, New York, 1986. 25. Leventhal H. and Cleary P. D. The smoking problem: A review of the research and theory in behavioral risk modification. Psychol. Bull. 88, 370-405, 1980. 26. Myrsten A.-L., Andersson K., Frankenhauser M. and Elgerot A. Immediate effects of cigarette smoking as related to different smoking habits. Percept Mot. Skills 40, 515-523, 1975. 27. Mausner B. and Platt E. S. Smoking: A behavioral Analysis. Pergamon Press, New York, 1971. 28. Nesbitt P. D. Smoking, physiological arousal and emotional response. J. Pers. Sot. Psychol. 23, 137-144, 1973. 29. Caplan R. D., Cobb S. and French J. R. P. Jr Relationship of cessation of smoking with job stress, personality, and social support. J. appl. Psycho/. 60, 21 I-219, 1975. 30. Conway T., Ward H., Vickers R. and Rahe R. Occupational stress and variation in cigarette, coffee, and alcohol consumption. J. Hlth Sot. Behuv. 22, 155-165, 1981. 31. Westman M., Eden D. and Shirom A. Job stress, cigarette smoking and cessation: The conditioning effects of peer support. Sot. Sci. Med. 20, 637-644, 1985. 32. Green K. Job characteristics and health-related behaviors among chemical plant workers. Dissertation: Johns Hopkins School of Hygiene and Public Health, 1988. 33. Green K. L. and Johnson J. V. The effects of psychosocial work organization on the prevalence of cigarette smoking among chemical plant employees. Am. J. pubf. HIth 80, 1368-1371,

1990.

34. House J. S. Occupational Stress and the Mental and Physical Health of Factory Workers. The Institute of Social Research, The University of Michigan, Ann Arbor, 1980. 35. Frese M. Occupational socialization and personality development: An underemphasized research perspective in industrial psychology. J. Occup. Psychol. 55, 209-224,

1982.

36. Wilensky H. L. Work, careers, and social integration. Int. Sot. Sci. J. 12, 543-560,

1960.

37. Garde11 B. Worker participation and autonomy: A multilevel approach to democracy at the workplace. 1982.

38. Elden M. Political efficacy at work: The connection between more autonomous forms of workplace organization and a more participatory politics. Am. Polit. Sci. Rev. 75, 43-58,

1981.

39. Svensson L. Grupper och Kollektiv. (Groups and Collectives.) Swedish Center for Working Life, Stockholm, 1986. (In Swedish.) 40. Kohn M. L. and Schooler C. Occupational experience and psychological functioning: An assessment of reciprocal effects. Am. Sot. Rev. 38, 97-118, 1973. 41. Magrmsson D. and Endler N. S. (Eds) Personality at the Crossroads. Erlbaum. Hillsdale. NJ. 1977. 42. Kohn M. L. and Schooler C. Work and Personality: An Inquiry into the Impact of Social Stratification.

Ablex, Norwood, NJ, 1983. 43. Miller K. A. and Kohn M. L. The reciprocal effects of job conditions and the intellectuality of leisure-time activities. In Work and Personality (Edited by Kohn M. L. and Schooler C.), pp. 217-241. Ablex, Norwood, NJ, 1983. 44. Karasek R. A. The impact of the work environment on life outside the job. Doctoral dissertation, Massachusetts Institute of Technology, distributed by the Institute for Social Research. Universitv of Stockholm. 1976. 45. Karasek R. A. ‘Job social&ion and job strain: The implications of two related psychosocial mechanisms for job design. In Working Life. A Social Science Contribution to Work Reform (Edited by Garde11 B. and Johansson G.), pp. 75-94. Wiley, London, 1981. 46. Karasek R. A. and Theorell T. Healthy Work. Stress, Productivity and the Reconstruction

of Working Life.

Basic Books, New York, 1990. 47. Karasek R., Garde11 B. and Lindell J. Work and non-work correlates of illness and behavior in male and female Swedish white collar workers. J. Occup. Behav. 8, 187-207, 1987. 48. Seligman M. Helplessness. On Depression, Development and Death. Freeman, San Francisco, 1975.

49. Bandura A. Self-efficacy mechanisms in human agency. Am. Psycho/. 37, 122-147, 1982, 50. McCarthy W. J. and Brown E. R. Health enchancement, illness prevention and illness containment behaviors: A continuum of declining impact on health self-efficacy. Summary paper, Annual Meeting of the American Association for the Advancement of Science, 1985. 51. Hacker W. Perceptions of and reactions to work situations: Some implications of an action control approach. In Towards a Psychology of Situations: An Interactional Perspective (Edited by Magnusson D.), pp. 113-134, Erlbaum, Potomac, MD, 1981. 52. Volpert W. Psychische Regulation von Arbeitstltigkeiten. (Mental regulation of work behavior.) In Enzyklopiidie der Psychologie (Edited by Rutenfranz J. and Kleinbeck U.). Hogrefe, Gottingen, 1986. 53. Kim J. 0. and Mueller C. W. Inrroduction to Factor Anal.vsis. Sage, London, 1978. 54. Harrell F. E. The logist procedure. In SUGI Supplement Library User’s Guide, 1983 edn, pp. 181-202. SAS Institute, Cary, NC, 1983. 55. Karasek R. A., Theorell T., Schwartz J. E., Schnall P. L., Peiper C. F., and Michela J. L. Job characteristics in relation to the prevalence of myocardial infarction in the US Health Examination Survey (HES) and the Health and Nutrition Examination Survey (Hanes). Am. J. publ. Hlth 78, 910-918, 1988. 56. Peterson C. and Stunkard A. J. Personal control and health promotion. Sot. Sci. Med. 28, 819-828, 1989.

57. Hall E. Women’s Work: An inquiry into the health effects of invisible and visible labor. Doctoral Dissertation, Karolinska Institute, Stockholm, Sweden, 1990.