Highlight, copy & paste to cite:

Bernhard, C. & Sverke, M. (2003). Work Attitudes, Role Stress and Health Indicators Among Different Types of Contingent Workers in the Swedish Health Care Sector, Research and Practice in Human Resource Management, 11(2), 1-16.

Work Attitudes, Role Stress and Health Indicators Among Different Types of Contingent Workers in the Swedish Health Care Sector

Claudia Bernhard & Magnus Sverke


Despite contingent employment being frequently used as a tool for organisations to increase flexibility and cut labour costs, previous research has suggested that contingent employment is not an unitary phenomenon. By applying cluster analysis to data from the Swedish health care sector, the present study identified four types of contingent workers characterised by distinct profiles on demographic and work-related variables. Despite differences in organisational commitment, role overload, and job-induced tension, these groups of contingent workers reported rather similar levels of work attitudes, role stress, and well-being. The paper demonstrates the importance of combining variables in understanding the heterogeneity of contingent work and discusses how a pattern approach can provide valuable information for research on contingent employment and its consequences.


In recent years, many organisations have had to undergo a wide range of restructuring as a function of economic vexation as well as competitive pressures. Downsizing, outsourcing, and the creation of ‘contingent’ work arrangements - a term that was first coined in a conference on employment security in 1985 (Polivka 1996) and since then has been applied to a wide range of employment practices offering only temporary jobs - have become common strategies for organisations to reduce costs and increase flexibility (Pfeffer 1998). The shift towards contingent work arrangements can be illustrated by examining labour market data. According to the U.S. Bureau of Labor Statistics, the use of temporary agency workers increased from half a million in 1983 to two million ten years later (Beard & Edwards 1995). Most European countries are facing a similar trend, and today about 13 per cent of the European work force is employed in jobs of a temporary nature (OECD 2002). According to the 1997 survey of forms of employment (FOE), temporary employment accounts for six per cent of the total work force in Australia (OECD, 2002).

There are several reasons for the growth of the contingent work force. First, hiring contingent workers to fill labour gaps provides flexibility to meet the challenges of an increasing global competition (Pfeffer & Baron 1988). Second, contingent workers reduce labour costs, as they mostly do not get the same benefits or training as their core worker counterparts (Nollen 1996). Third, on a more general level, the economy in many industrialised countries is gradually shifting from the second sector (industry) to the third sector (service) with more low-paid and low-skilled jobs, where on-the-job training is not of such importance as in highly specialised industry work (Hoffmann & Walwei 2000).

Scant research about contingent work has been done, and consequently, knowledge about the new forms of working arrangements and their effects is limited. On a general level, there are numerous factors differentiating contingent workers from core full-time and part-time employees, which in turn, may lead to negative consequences for the individual, and in the long run also for the organisation (Pfeffer & Baron 1988). Due to the temporary nature of the employment contract, both the organisation and co-workers may invest less time and effort in assisting contingent employees in the understanding of their role responsibilities (Kochan, Smith, Wells & Rebitzer 1994). Because contingent workers have “less stake in the contingent job, their co-workers or the organization” (Pearce 1998:36), they are generally expected to express less favourable job attitudes and less commitment to the organisation. Further, contingent workers experience low control over the onset, contents, and termination of their work, and as a function, face a low degree of predictability (Beard & Edwards 1995). Contingent workers frequently find themselves in ‘new’ positions which may lead to increased role ambiguity and stress (Sverke, Gallagher & Hellgren 2000). In addition, an unpredictable work situation combined with an uncertain personal economy due to the temporary nature of work may result in impaired well-being (Krausz, Brandwein & Fox 1995, Martens, Nijhuis, van Boxtel & Knottnerus 1999).

However, the empirical evidence of research conducted to contrast the consequences of contingent work with those of traditional work arrangements is equivocal (McLean Parks, Kidder & Gallagher 1998). In this paper, it is argued that the ambiguous results stem from the fact that contingent workers do not represent a unitary category in terms of their contractual agreements, their working conditions and demographic characteristics.

As noted by McLean Parks et al. (1998), there exist several contingent work arrangements, which generally speaking, differ in terms of the duration of the contract, the number of contractual parties and the form of employment (some contingent workers are self-employed). Despite the efforts to define different categories of contingent employment contracts, research still has difficulties to explain similarities and differences in work-related experiences and health (Benavides, Benach, Diez-Roux & Roman 2000, Aronsson, Gustafsson & Dallner 2002). Besides a more detailed distinction of contingent work forms there seem to be other individual and contextual influences that explain the differing reactions of employees towards contingent work (Aronsson et al. 2002, Isaksson & Bellaagh 2002). Background variables such as gender (Ellingson, Gruys & Sackett, 1998), number of children (Isaksson & Bellaagh 2002), perceived financial dependency (Buttram 1996), type of profession (Isaksson & Bellaagh 1999), and psychological variables of personal choice and needs (Krausz et al. 1995, Isaksson & Bellaagh 2002) have been found to be significant predictors of work related attitudes, stress and health among contingent workers. However, labour market statistics of contingent work report age groups, educational levels and gender distributions do differ in the different categories of contingent work (Bergström & Storrie 2002). Hence, these variables may take on different meanings for different types of contingent employees. In other words, even if research suggests that such factors predict contingent workers’ attitudes and well-being, it is still unclear how the influence of each of them changes when they interact.

Against this background, it seems questionable whether studies using single variables as the main analytic unit can reveal much about what benefits or risks contingent work can have for the employees. Instead of a variable-oriented approach, it may be more appropriate to use a holistic approach with the individual “being the entity under study” (Baltes & Nesselroade 1979: 1). Following the theoretical rationale of a holistic, person-oriented approach, the individual is viewed to function as a totality formed by interactions among the elements involved (Bergman & Magnusson 1997). Each operating factor takes on meaning from the total functioning of the individual, and hence, the specific pattern of operating factors becomes the main analytic unit for an investigation of individual differences (Magnusson 1998). Although originating from developmental science, the pattern approach has also been found valuable in the context of working life research, for instance in the identification of patterns of psychological contracts (Shore & Barksdale 1998) and patterns of dual commitment to company and union (Sverke & Sjöberg 1994). A recent study succeeded in classifying ‘boundaryless’ versus ‘traditional’ contingent employees on the basis of their wages, job alternatives, employment preferences, and demographic variables, and found that the two types of contingent workers differed with respect to work attitudes and behaviour (Marler, Woodward Barlinger & Milkovich 2002).

Applying a pattern approach to a sample of Swedish contingent workers in the health care sector, the first objective of this paper was to find out how personal (age, gender, child responsibility, cohabiting) and work-related characteristics (profession, working time, organisational tenure) in combination with the formal employment contract can be used to identify distinct patterns of contingent workers. The second objective of the study was to investigate if such patterns of contingent workers differ also with respect to work attitudes, role stress, and health.

The Heterogeneity of Contingent Work

The typical contingent worker is often described as young, female, less educated, low-paid, and less experienced with the specific industry and the employer (Nollen 1996). However, as more and more individuals are being employed on a temporary basis, the contingent worker’s face has changed. The contemporary contingent worker is likely to possess a growing range of professional skills, since it has become more common to employ even professionals such as computer information specialists, engineers, legal staff, and nurses on a contingent basis (Davis-Blake & Uzzi 1993). Thus, over the years, contingent workers have become a more heterogeneous group, and research has identified several factors accounting for the variation.

First, contingent workers differ in their type of contractual agreement (McLean Parks et al. 1998, Aronsson et al. 2002). One common group of contingent workers are hired in various kinds of time-restricted employment (e.g., for the duration of a project or as a temporary substitute). Another group of contingent employees work on-call as ‘in-house temporaries’ to in fill vacancies. A third form of contingent work arrangements consists of leased workers and temporary firm workers. These contingent workers are employed by one organisation (a temporary agency), but they are temporarily ‘rented’ to a client organisation. Finally, various types of self-employment may be considered a form of contingent employment. For instance, working as an independent contractor means being hired by an organisation for a short time to supply specific skills. According to McLean Parks et al. (1998), differences in contract types are likely to affect the contract holder’s work-related experiences in different ways. For example, role stress and organisational commitment may differ, because different groups of contingent employees are not to the same extent integrated in the organisation and familiar with the work practices. Likewise, role conflicts may be more prevalent for leased workers and temporary help agency workers than for contingent workers who perform their job in the employing organisation (Gallagher & Sverke 2000). However, different experiences with contingent work may not only reflect contractual but also occupational differences. For instance, independent contractors and consultants often have a higher education and jobs requiring more skills and knowledge in comparison to seasonal workers. In turn, such differences may lead to different intrinsic work motivation, job satisfaction, and willingness to remain with the organisation (Nollen 1996, Walsh & Deery 1999).

There is some evidence gender is related with dimensions of contingent employment. Aronsson and Göransson (1998), for example, found women express more negative reactions to contingent work as compared to men. Generally, women seem to experience one and the same situation differently than do men (Diener, Suh, Lucas & Smith 1999), and besides that, mostly women take care of children and do a great part of the household work (Sverke et al. 2000). Moreover, women are over-represented in more shortterm and unpredictable contingent work contracts, such as on-call arrangements (Bergström & Storrie 2002). Failure to consider these issues might lead to misinterpretations of results, because more inter-role conflicts and impaired well-being among female contingent workers could be intertwined with the contractual work agreements and private life situations.

Consequently, the family situation of the contingent worker is another important matter. For example, having a partner as a social and economic ‘supporter’ can represent a resource for the contingent worker. As well, studies about subjective well-being have consistently found a significant relationship between marital status and well-being (Diener et al. 1999). Second, contingent workers without partners are first income-earners, which makes them much more dependent on a steady income than cohabiting contingent employees. As Buttram (1996) reports from his study about role-stress, strain, and satisfaction among contingent workers, an increased number of job assignments in interaction with a high level of economic dependency may lead to reduced well-being, impaired satisfaction, and more inter-role conflict.

The literature assumes a long-term relationship between employer and employee to be a prerequisite for a climate of trust and commitment, but typically, contingent workers have a short tenure with their present organisation (Kochan et al. 1994). By definition, contingent workers change jobs and organisations quite often. Thus, in contrast to permanent work, age is not automatically connected to a long organisational tenure and experience in an organisation. Still, older contingent workers are plausibly more experienced than younger ones, and hence, organisational tenure and age may take on different importance for outcome factors such as attitudes and well-being of contingent employees. Apart from that, age goes together with this form of employment in a clear-cut way such that individuals under 25 as well as over 54 tend to be employed on contingent employment contracts to a greater extent as compared to middle-aged individuals (European Foundation 1996).

The distribution of working hours is yet another factor differentiating the contingent work force. Martens et al. (1999) showed that irregular work hours as well as uncertainty of ongoing employment are connected with health complaints. Nevertheless, their study provides no information about the effects on individual well-being, if uncertain employment and irregular working hours occur in a combination.


Setting and Sample

Data were collected by mail questionnaires to the total staff of two Swedish emergency hospitals undergoing organisational restructuring. Completed questionnaires were received from 1,505 (61.3%) out of the 2,455 employees sampled for the survey. The present study is limited to 196 employees hired on contingent employment contracts. The mean age of the sample was 35 years (SD=9), and the majority (76%) were female. Around one third of the participants (35%) had children under age twelve living in their household, and nearly two thirds (64%) were married or cohabiting. Eighteen per cent of the contingent workers were physicians, 39 per cent worked as nurses, and more than 40 per cent of the contingent employees worked in other sections of the hospital (e.g., administration, service departments, or medical support specialties). More than half of the sample (58%) held a contract based on deputyship (i.e., direct-hire employees with a limited contract). The majority (72%) reported working in evenings and nights or in shifts. The length of service at the hospital ranged from less than a year (28%) to 24 years (0.5%), with an average organisational tenure of 4.5 years. Despite the rather long organisational tenure, the sample fits well into the description of the ‘average’ Swedish contingent worker as being rather young, predominantly female and working foremost in fixedterm or on-call contracts (Bergström & Storrie, 2002). Likewise, the sample is well comparable to Australian contingent workers, who are characterised as rather young, mostly female, working in fixed-term employment and primarily in the service sector (OECD 2002).


The postal questionnaire was designed to capture a range of demographic characteristics, experiences of the work environment, work attitudes, and well-being. The order of questionnaire items was randomised to avoid response set bias. Table 1 presents descriptive statistics, reliability estimates (Cronbach’s alpha), and inter-correlations for the study variables.

Characteristics of Contingent Workers

Age was measured as a continuous variable. Gender (1=female, 0=male), child responsibility (1=children younger than 12 living in the household), and marital status (1= cohabiting) were assessed with dichotomously scored variables. Organisational tenure was measured by asking how many years the participants had been working in the hospital. The variable daytime was constructed out of a question concerning work hours; participants stating that they worked daytime only were coded 1, while other response alternatives (evening or night-time only; irregular work hours/shift work) were coded 0. The variable deputyship was based on a question concerning the type of contingent employment contract; direct-hire employees with a limited contract were coded 1 while remaining forms (e.g., in-house temps contracted on an hourly basis) were coded 0. The various responses to a question on vocation were summarised to two dichotomous variables for the most frequent occupational groups: nurse (1=nurse, 0=other) and physician (1=physician, 0=other).

Work Attitudes

All work attitudes were assessed using Likert scales ranging from 1 (strongly disagree) to 5 (strongly agree). Job satisfaction was measured with a three-item index (Hellgren, Sjöberg & Sverke, 1997) capturing an overall contentment with the present job. The scale was adopted from Brayfield and Rothe (1951) and contains items like “I am satisfied with my job” (α = .85). Organisational commitment was assessed using a four-item short-form version of an affective commitment scale developed by Allen and Meyer (1990) (e.g., “This organisation has a great deal of personal meaning to me”). The coefficient alpha reliability was .68. Turnover intention was measured with a three-item scale devised to capture an overall turnover propensity (Sjöberg & Sverke 2000). The items (e.g., “I am actively looking for another job”), which were taken from different propensity-to-leave scales (Lyons 1971, Camman, Fichman, Jenkins & Klesh 1979), were slightly modified from the original scales so that all items were statements and not questions (α = .79).

Role Stressors

All role stress variables were assessed using Likert scales ranging from 1 (strongly disagree) to 5 (strongly agree). Role overload was assessed with three items (e.g., “I often have too much to do in my job”) developed by Beehr, Walsh and Taber (1976). The scale demonstrated a reliability estimate of .70. Role conflict was measured with a slightly modified version of fiveitem scale developed by Rizzo, House and Lirtzman’s (1970) five-item scale (e.g., “I receive incompatible requests from two or more people”). The scale demonstrated satisfactory internal consistency (α = .79). Role ambiguity was assessed using a combination of items from Caplan (1971) and Rizzo et al. (1970). The index contained four items (e.g., “There exist no clear, planned goals and objectives for my job”) with a reliability estimate of .69.

Table 1
Variable Intercorrelations, Means, Standard Deviations and Reliabilities (N = 196)
Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Mean 34.83 0.76 0.64 0.35 0.28 0.58 4.48 0.18 0.39 3.74 2.65 2.46 2.37 2.41 2.02 1.78 2.61 2.10
Standard deviation 9.17 0.43 0.48 0.47 0.45 0.50 0.44 0.38 0.49 1.01 0.90 1.30 0.92 0.90 0.78 0.44 0.93 0.68
Reliability (alpha) - - - - - - - - - 0.85 0.68 0.79 0.70 0.79 0.69 0.86 0.84 0.82
Characteristics of contingent workers
1. Age
2. Gender (female) -.09
3. Marital Status .09 .01
4. Child responsibility -.02 .06 .35
5. Daytime .23 -.04 -.01 -.02
6. Deputyship -.31 -.14 .01 -.03 -.00
7. Organisational tenure .57 -.12 -.01 -.10 .07 -.21
8. Physician .04 .21 .17 .13 -.08 .02 -.11
9. Nurse -.04 -.21 -.06 -.06 -.41 .03 .08 -.59
Work attitudes
10. Job satisfaction .25 .22 .07 .06 -.07 -.08 .22 .03 .12
11. Organisational commitment .21 -.18 .06 .06 .14 .05 .15 -.01 -.13 .59
12. Turnover intention -.24 .15 -.04 -.14 .11 .11 -.20 -.01 -.12 -.70 -.49
Role stressors
13. Role overload -.09 -.01 .01 -.01 -.02 .15 -.08 .22 -.16 -.20 .07 .21
14. Role conflict -.10 .16 -.08 -.01 -.09 -.08 -.09 .16 -.03 -.33 -.14 .33 .36
15. Role ambiguity -.07 .06 .09 -.02 .07 -.00 -.12 .14 -.13 -.42 -.28 .40 .46 .50
Health indicators
16. Mental distress -.07 -.04 -.08 -.07 .22 .10 -.10 .01 -.17 -.36 -.09 .37 .40 .34 .41
17. Job-induced tension -.12 .04 .05 -.10 .06 .19 -.06 .26 -.20 -.28 .01 .27 .48 .41 .33 .54
18. Somatic complaints -.08 -.02 .00 -.03 .11 .13 -.16 -.01 -.05 -.32 -.12 .34 .31 .26 .25 .60 .56

a. For r >= .16, p < .05; r >= .21, p < .01; r >= .26, p < .001.
b. Scale range: 0-1 (variables 2-6, 8-9 [for these variables the mean value symbolises the proportion scoring 1]), 1-5 (variables 10-15, 17-18), 1-4 (variable 16), years (variables 1, 7).

Health Indicators

Mental distress was measured with the General Health Questionnaire (Goldberg 1979), which consists of 12 questions where respondents indicate how often they have experienced certain symptoms (1=never, 4=always). The scale, which is deemed to represent a good indicator of non-psychiatric disorders, demonstrated adequate reliability (α = .86). We used a scale developed by House and Rizzo (1972) to assess job-induced tension. The seven items (e.g., “My job tends to directly affect my health”), which were scored on a five-point Likert scale, yielded a reliability estimate of .84. Somatic complaints were assessed with a ten-item index in which respondents indicated how often (1=never, 5=always) they had suffered from various symptoms (e.g., stomach problems, muscular tension, sleeping problems) in the past 12 months. The symptom list was developed by Andersson (1986) and modified by Isaksson and Johansson (1997). The coefficient alpha reliability was .82 in the present study.

Statistical Analysis

The first objective of this study, that is, to identify different patterns of contingent work, was addressed using cluster analysis. Cluster analysis is one of the methods naturally used for the purpose of classification (see Magnusson 1998), and has previously been used to distinguish different categories of contingent employment (Marler et al. 2002). All demographic (age, sex, marital status, child responsibility) and work related characteristics (daytime work, deputyship contract, organisational tenure, nurse, physician) were subjected to Ward’s (1963) hierarchical clustering method. In this procedure, where each individual initially represents a cluster, the two clusters that result in the smallest increase in the overall sum of squared within-cluster Euclidian distances are merged at each step until all cases go together in one single cluster. Because cluster analysis provides no statistical stop criteria indicating a ‘best’ solution, all solutions from one to six clusters were specified. To validate the results and choose the final solution, two criteria were used. First, the clusters were tested to determine if they could be statistically distinguished with respect to the input variables. Here, ANOVA was employed for the continuously scaled variable age. For all categorical input variables, chi-square tests were performed to analyse differences between the clusters. Second, the cluster solution should group all individuals in clusters that made sense and showed differences that could be interpreted.

The second objective of the study — to investigate if different patterns of contingent workers differ also with respect to work attitudes, role stress, and health — was addressed using multivariate analysis of variance procedures (MANOVA). Three separate MANOVAs were conducted to investigate whether the patterns of contingent workers identified in the cluster analysis differed in terms of work attitudes (job satisfaction, organisational commitment, and turnover intention), perceived role stress (role overload, role conflict, and role ambiguity), and self-reported health (mental distress, job induced tension, and somatic complaints). To examine multivariate differences in more detail, univariate follow-up tests were used, and post hoc tests (Bonferroni) were included in the analysis to test for between-group differences when the univariate effects were significant.


Different Patterns of Contingent Workers

When the various cluster solutions (one to six clusters) were analysed, the three-, four-, five-, and six-cluster solutions contained one small cluster (made up of seven elderly nurses) in addition to the larger clusters. Apart from this, the five-cluster solution revealed the most meaningful distinction between groups. Hence, this solution was chosen although the group consisting of the seven nurses was excluded from further analysis as it was too small for any statistical comparisons.

Table 2 presents descriptive statistics for each of the remaining four groups in all cluster input analysis variables. The F and chi-square tests show there were significant differences between the four groups in all variables other than child responsibility, thus indicating that this solution really generated distinct groups. In each of the four clusters, there are individuals with certain characteristics, which distinguish them from the other groups of contingent workers.

Table 2
Description of the Clusters (N = 189)
Variable Cluster 1
Younger nurses
Cluster 2
Cluster 3
Middle-aged nurses
Cluster 4
Support staff
n 69 34 44 42
% 36.5 18.0 23.3 22.2
Age (years) 29.2 36.8 40.1 36.8 32.29 ***
Child responsibility (%) 27.5 50.0 38.6 31.0 5.62       
Marital Status (%) 46.4 82.4 84.1 57.1 22.84 ***
Female (%) 75.4 55.9 86.4 81.0 10.60 *    
Daytime (%) 0.0 23.5 6.8 100.0 144.83 ***
Deputyship (%) 79.7 58.8 25.0 54.8 33.13 ***
Organisational tenure (years) 3.6 3.8 7.1 3.5 67.49 ***
Physician (%) 0.0 100.0 0.0 0.0 189.00 ***
Nurse (%) 87.0 0.0 100.0 26.2 122.07 ***

a. * p < .05 ; *** p < .001.
b. n = number of respondents.
Total N equals 189 since seven individuals in the total sample (N=196) were omitted from the analysis.

In the first cluster, labelled younger nurses (N=69), the average age was below 30 years, almost 90 per cent worked as a nurse while there were no physicians. Less than half the individuals lived with a partner and three quarters were female. None of them reported to work daytime only, and almost 80 per cent held a contract based on deputyship. Hence, this cluster could be characterised to comprise mainly younger, single nurses working on a deputyship contract. The second cluster, referred to as physicians (N=34), consisted solely of physicians, the majority of whom worked irregular work hours. These individuals had a higher average age (37 years), more than 80 per cent lived with a partner, and the percentage of women (55%) was the lowest from all four groups. Most of them were employed on a deputyship contract.

The third cluster, identified as middle-aged nurses (N=44), was the one with the highest average age (40 years) and contained only nurses. Again, as in the group of physicians, more than 80 per cent were living with a partner, but 86 per cent were women. This group had the longest average organisational tenure (7 years) as compared to the other clusters (less than 4 years). The third cluster was also the group with the fewest individuals employed on a deputyship contract (25%). The fourth cluster, labelled support staff (N=42), contained no physicians and only 26 per cent nurses; most employees in this group worked in support activities (e.g., medical secretaries, laboratory assistants, psychologists, physiotherapists). None of the respondents in this group worked shifts or during nights. The average age was about the same as in the group of the physicians (37 years), the percentage of women was almost the same as in the group of the middle-aged nurses (81%), and 57 per cent were cohabiting.

Differences in Work Attitudes, Role Stress and Health

The mean values for the work attitudes, role stress and health outcomes are presented and compared in Table 3.

Table 3
Tests for Mean Differences in Work Attitudes, Role Stress and Health Indicators (N = 189)
Outcome variable Cluster 1
Younger nurses
Cluster 2
Cluster 3
Middle-aged nurses
Cluster 4
Support staff
Univariate F Means comparisons
n 69 34 44 42
Work attitudes
Job satisfaction 3.66 3.85 3.84 3.65 0.49 n.s.
Organisational commitment 2.63 2.73 2.37 2.94 2.95* 3 < 4
Turnover intention 2.54 2.28 2.20 2.76 1.26 n.s.
Role stress
Role ambiguity 1.98 2.20 1.91 2.04 1.00 n.s.
Role overload 2.46 2.81 2.01 2.22 5.46*** 2 > 3, 4
Role conflict 2.45 2.74 2.25 2.23 2.57 n.s.
Health indicators
Mental distress 1.72 1.79 1.70 1.93 2.30 n.s.
Job induced tension 2.59 3.16 2.23 2.61 6.12*** 2 > 3
Somatic complaints 2.10 2.05 2.03 2.22 0.74 n.s.

a. * p < .05, *** p < .001.
b. For means comparisons (Bonferroni tests), the difference between groups is significant at the .05 level. Cluster 2 = 2, Cluster 3 = 3, and Cluster 4 = 4, n.s. = non-significantly different means.
c. n = number of respondents.

There were significant multivariate differences between the four clusters in all three MANOVAs conducted. This indicates that there were overall differences between the clusters of younger nurses, physicians, middle-aged nurses, and support staff with respect to work attitudes (Multivariate F[9,540]=2.57, p<.01), role stress (Multivariate F[9,540]=2.33, p<.05), and health indicators (Multivariate F[9,525]=3.83, p<.001). However, the univariate follow-up tests revealed that the four groups differed only with respect to organisational commitment (F[3,180]=2.95, p<.05), role overload (F[3,180]=5.46, p<.001), and job-induced tension (F[3,175]=6.12, p<.001). Bonferroni post hoc tests for these significant univariate results showed that the cluster comprising support staff contingent workers reported significantly higher organisational commitment than did middle-aged nurses. Furthermore, physicians reported a level of role overload that was significantly higher than for middle-aged nurses and support staff. For job-induced tension, the post hoc test revealed a significant difference between physicians and middle-aged nurses, physicians reporting a significantly higher level of tension.

Concluding Discussion

The literature on contingent work has identified a number of factors that contribute to the heterogeneity of contingent workers, and this study attempted to find out whether a combination of these factors shape certain patterns that distinguish subgroups of contingent employees. The focus on the combination of individual characteristics was chosen on the rationale that the complex processes of individual functioning are not understandable by investigating single variables in isolation from the context of others (Magnusson 1998).

With the aid of cluster analysis, four homogeneous subgroups within the sample of contingent employees from the Swedish health care sector were identified. The four clusters — labelled younger nurses, physicians, middle-aged nurses, and support staff — differed significantly from one another in all the cluster variables except for child responsibility. Although the denominations direct attention to age and profession, the overall patterns of individual characteristics, rather than these isolated variables, appeared to differentiate between different contingent workers and split the larger sample in subgroups. For example, there were two groups of nurses. Many of the middle-aged nurses lived with a partner, worked as in-house temps and had an average organisational tenure of seven years, in contrast to the cluster of younger nurses who more often held a deputyship contract and lived alone. Likewise, the proportions of individuals cohabiting were equally high among middle-aged nurses and physicians, but the two groups differed in terms of age, profession, contract form, daytime work, organisational tenure, and gender distribution. Further, the proportion of women was almost the same for middle-aged nurses and the support staff, but there were no other similarities. Moreover, the clusters of support staff and physicians showed similarities in average age and deputyship contract, but besides their profession, the groups differed in gender, daytime work, and marital status. These findings suggest that within the heterogeneous group of contingent workers, there are subsets of individuals sharing similar backgrounds, which make them different from other contingent employees. For future research, therefore, it may be valuable to investigate subgroups or patterns of contingent employees rather than to classify workers into core and peripheral, or to differentiate them solely on the basis of their formal contract type (see Marler et al. 2002).

The empirical evidence on individual consequences of contingent work is mixed (McLean Parks et al. 1998). With the aim to shed more light on this issue, the present study investigated if different patterns of contingent work are associated with different consequences in terms of work attitudes, role stress, and well-being. Generally, a particular ‘group at risk’ or a group with consistently more beneficial consequences was not identified. Overall, there were multivariate differences between the clusters in work attitudes, role stress, and health, but the only significant univariate differences concerned organisational commitment, role overload, and job-induced tension. Thus, although there were some significant and interpretable differences in outcome variables between the clusters, the four groups typically expressed more similarities than differences in their work related experiences and well-being.

The few existing differences do not seem connected to a single factor, but rather to specific combinations of factors inherent in the profiles characterising the different clusters of contingent workers. For instance, the results show that middle-aged nurses, with the longest organisational tenure of all four clusters, reported the lowest organisational commitment, and thus, differed significantly from the group of support staff. In contrast to the argumentation of Kochan et al. (1994), tenure and commitment were non-significantly correlated, which may indicate that tenure can take on different meanings when combined with other factors. For example, organisational tenure might be inversely related to commitment if experiences with the employer are negative. Due to their longer tenure, the group of middle-aged nurses more likely than the other groups of contingent workers had witnessed the whole process of organisational restructuring taking place at the two hospitals, and their experiences might explain their low level of organisational commitment.

While Sverke et al. (2000) found no significant differences between core and contingent workers in role overload, the present study revealed differences among different types of contingent employees, with physicians reporting higher role overload than middle-aged nurses and support staff contingent employees. Since the younger nurses did not differ significantly from the other groups, profession in combination with work experience might explain the difference. These results might indicate that high job demands in connection with low experience signify a high demand-low control situation (Karasek & Theorell 1990), which might increase stress experiences such as role overload. A low degree of control is almost inevitably inherent in contingent work (Beard & Edwards 1995), and hence, inexperienced contingent workers with high job demands can be expected to report more stress than core workers under the same conditions.

Concerning health indicators, physicians reported a significantly higher level of job-induced tension than middle-aged nurses. The cluster of middle-aged nurses was somewhat special in terms of age, tenure in the organisation, and type of contract. Physicians typically have more responsibilities at work than all other groups, and their work tasks also differ from the rest. So, age and tenure, which again might be connected to a certain experience, seem to help to ‘calm down’. Combined with that, hourly-based contingent work (which characterised the majority of the middle-aged nurses) typically involves fewer working hours per week as compared to deputyship contracts (which was characteristic of the other groups), which gives more time to recover from daily hassles.

To summarise, the pattern approach appears valuable for detecting the complex relationship between different natures of contingent work and their consequences. Nevertheless, the question arises why, although dissimilar in pattern, the four subgroups of contingent workers generally express more similar than dissimilar reactions. First, this finding might partly be explained by the contingent workers’ employment conditions. Compared to the broad variation of theoretically possible work arrangements (e.g., McLean Parks et al. 1998), only a limited number of contingent employment contract forms has been prevalent in the sample for this study. More importantly, the work arrangements under comparison (in-house temporary or fixed-term contracts) are still fairly similar since they involve an employee-employer relation of a certain duration. Apart from that, these employment forms are more typical for women, whereas forms of self-employment and project work are more prevalent for men and highly educated specialists (Bergström & Storrie 2002) who may represent another subgroup of contingent workers. Therefore, contrasting more dissimilar contingent work relationships will be necessary in future studies to give a more complete picture about contingent work and its consequences.

Second, some similarities found in this study may have appeared due to the special context of the investigation in the health care sector. The majority of the participants worked in caring occupations, which may have affected the results. As noted by Sverke et al. (2000) temporary health care workers may experience positive intrinsic factors of job motivation and be more accustomed to role characteristics associated with the profession rather than with the organisation, thus potentially explaining similarities between different patterns of contingent work in the sample.

Moreover, sector-specific growth rates of contingent work and different characteristics of contingent workers in certain sectors (e.g., young women and ethnic minorities are over-represented in the service sector) might well be of importance (see Bergström & Storrie 2002). Because contingent work arrangements lack standardised definitions (McLean Parks et al. 1998), negotiations are largely made according to the demands existing within one organisation. Thus, the experiences of contingent employees might depend upon how the organisation treats and defines their work arrangements. For instance, although all four subgroups in the present research differed with respect to organisational tenure, the average tenure was over four years. It is possible that tenure accounts for parts of the similarities given that the number of job changes is related to factors such as role ambiguity (Buttram 1996). To conclude, pattern oriented studies for a broader range of different organisations outside of the health care sector are necessary extensions to the present study.

Third, it would be fruitful to use other characteristics to examine how the heterogeneous group of contingent workers can be subdivided in homogeneous patterns. A growing body of literature suggests that the voluntary choice of contingent employment arrangements is one of the key factors to understand employees’ reactions (e.g., Krausz et al. 1995, Ellingson et al. 1998). Subjective preferences also partly reflect individual characteristics and life situations (Marler et al. 2002). Thus, research including the issue of volition in identifying patterns of contingent work is especially warranted. Other important subjective perceptions could involve the psychological contract of contingent workers (McLean Parks et al. 1998). For example, subjective time frames of contingent employees, that is, perceptions of defined and limited assignments versus perceived promises for renewed contracts, may represent important influences on well-being and work attitudes.

Thus, despite the limitations of this study, the results highlight the importance of examining how several variables in combination form distinct profiles of contingent workers. Apart from identifying potential variables that contribute to differences among workers employed on contingent contracts, this study also indicates that more research is needed to investigate which combinations of variables shape different patterns within the heterogeneous group of contingent workers — and which impact these combinations have when the variables are not examined in isolation. As has been demonstrated, those differences between clusters that occurred in terms of work attitudes, role stress, and well-being cannot be explained solely by linear relations between variables, but rather by a combination of variables. In connection with the increased flexibility of the labour market, contingent work arrangements are likely to grow and affect the daily lives of growing numbers of individuals. The use of a pattern approach can provide valuable information on the different natures of contingent work arrangements and bring an answer to the question under which circumstances contingent employment leads to positive or negative consequences for the employees and employers.


Claudia Bernhard, a Ph.D candidate, is working at the National Institute for Working Life in Stockholm, Sweden. Her main research interests include contingent work arrangements and the role of psychological contracts for employees’ well-being.

Magnus Sverke, Ph.D, is an Associate Professor at the Department of Psychology, Stockholm University, Sweden. His research interests include organisational change and its effects on employees, downsizing and job insecurity, labour market flexibility and new employment contracts, employee attitudes and well-being, and union member attitudes and behaviour.


Allen, N.J., & Meyer, J.P. (1990). The measurement and antecedents of affective, continuance and normative commitment to the organization. Journal of Occupational Psychology, 63(1), 1-18.

Andersson, K. (1986). Utveckling och prövning av ett frageformulärsystem rörande arbetsmiljö och hälsotillstånd [Development and testing of a questionnaire concerning work environment and health], Rapport No. 2:1986, Yrkesmedicinska Kliniken, Örebro.

Aronsson, G., & Göransson, S. (1998). Tillfälligt anställda och arbetsmiljödialogen: En empirisk studie [Contingent workers and the work-environment dialogue: An empirical study]. Arbete och Hälsa, 1998:3.

Aronsson, G., Gustafsson, K., & Dallner, M. (2002). Work environment and health in different types of temporary jobs. European Journal of Work and Organizational Psychology, 11(2), 151-175.

Baltes, P.B., & Nesselroade, J.R. (1979). History and rationale of longitudinal research. In J.R. Nesselroade & P.B. Baltes (Eds.), Longitudinal research in the study of behavior and development (1-39). New York: Academic Press.

Beard, K.M., & Edwards, J.R. (1995). Employees at risk: Contingent work and the psychological experience of contingent workers. In C.L. Cooper & D.M. Rousseau (Eds.), Trends in organizational behavior (109-126). New York: Wiley.

Beehr, T.A., Walsh, J.T., & Taber, T.D. (1976). Relationship of stress to individually and organizationally valued states: Higher order needs as a moderator. Journal of Applied Psychology, 61(1), 41-47.

Benavides, F.G., Benach, J., Diez-Roux, A.V., & Roman, C. (2000). How do types of employment relate to health indicators? Findings from the second European Survey on Working Conditions. Journal of Epidemiology and Community Health, 54(7), 494-501.

Bergman, L.B., & Magnusson, D. (1997). A person-oriented approach in research on developmental psychopathology, Development and Psychopathology, 9(2), 291-319.

Bergström, O., & Storrie, D. (2002). Contingent employment in Europe and the United States. Cheltenham: Edward Elgar.

Brayfield, A.H., & Rothe, H.F. (1951). An index of job satisfaction. Journal of Applied Psychology, 35(5), 307-311.

Buttram, R.T. (1996). Working without a net: A theoretical and empirical analysis of worker responses to temporary employment, Dissertation, Tulane University.

Camman, C., Fichman, M., Jenkins, D., & Klesh, J. (1979). The Michigan Organizational Assessment Questionnaire, Ann Arbor, MI: University of Michigan (Unpublished manuscript).

Caplan, R.D. (1971). Organizational stress and individual strain: A social-psychological study of risk factors in coronary heart diseases among administrators, engineers and scientists. Institute for Social Research, University of Michigan, University Microfilms No.72/14822, Ann Arbor, Michigan.

Davis-Blake, A., & Uzzi, B. (1993). Determinants of employment externalization: A study of temporary workers and independent contractors. Administrative Science Quarterly, 38(2), 195-223.

Diener, E., Suh, E.M., Lucas, R.E., & Smith, H.L. (1999). Subjective well-being: Three decades of process. Psychological Bulletin, 125(2), 276-302.

Ellingson, J.E., Gruys, M.L., & Sackett, P.R. (1998). Factors related to the satisfaction and performance of temporary employees. Journal of Applied Psychology, 83(16), 913-921. European Foundation. (1996). Precarious employment and working conditions in the European Union. Dublin: European Foundation for the Improvement of Living and Working Conditions.

Goldberg, D. (1979). Manual of the General Health Questionnaire. National Foundation for Educational Research, Windsor.

Gallagher, D.G., & Sverke, M. (2000). Contingent employment contracts: Are existing theories still relevant?. Proceedings of the 12th World Congress of the International Industrial Relations Association, Tokyo, Japan, May 29 - June 2, 2000, Proceedings Volume I, Track I: Invited Papers, 131-140.

Hellgren, J., Sjöberg, A., & Sverke, M. (1997). Intention to quit: Effects of job satisfaction and job perceptions. In F. Avallone, J. Arnold & K. De Witte (Eds.), Feelings work in Europe (415-423). Milano: Guerini.

Hoffmann, E., & Walwei, U. (2000). Strukturwandel in der Erwerbsarbeit: Was ist eigentlich noch normal? [Structural changes in paid employment: What is still normal?]. IAB Kurzbericht, 14(25.10.2002), 1-7.

House, R.J., & Rizzo, J.R. (1972). Role conflict and ambiguity as critical variables in a model of organizational behavior. Organizational Behavior and Human Performance, 7(3), 467-505.

Isaksson, K., & Bellaagh, K. (1999). Anställda i uthyresföretag: Vilka trivs och vilka vill sluta? [Temporary firm employees: Who are satisfied and who want to quit?]. Arbete & Hälsa, 1999:7.

Isaksson, K., & Bellaagh, K. (2002). Health problems and quitting among female “temps”. European Journal of Work and Organizational Psychology, vol. 11(1), 27-45.

Isaksson, K., & Johansson, G. (1997). Avtalspension med vinst och förlust [Gains and losses of early retirement]. Stockholm: Folksam.

Karasek, R., & Theorell, T. (1990). Healthy work. New York: Basis Books Inc.

Kochan, T.A., Smith, M., Wells, J.C., & Rebitzer, J.B. (1994) Human resource strategies and contingent workers: The case of safety and health in the petrochemical industry. Human Resource Management, 33(1), 55-77.

Krausz, M., Brandwein, T., & Fox, S. (1995). Work attitudes and emotional responses of permanent, voluntary and involuntary temporary help employees: An exploratory study. Applied Psychology: An International Review, 44(3), 217-232.

Lyons, T.F. (1971). Role clarity, need for clarity, satisfaction, tension, and withdrawal. Organizational Behavior and Human Performance, 6(1), 99 -110.

Magnusson, D. (1998). The logic and implications of a person-oriented approach. In R. B. Cairns, L. R. Bergman, & J. Kagan (Eds.), Methods and models for studying the individual (33-64). California: Sage.

Marler, J.H., Woodward Barlinger, M., & Milkovich, G.T. (2002). Boundaryless and traditional contingent employees: Worlds apart. Journal of Organizational Behavior, 23 (special issue), 425 - 453.

Martens, M.F.J., Nijhuis, F.J.N., van Boxtel, M.P.J., & Knottnerus, J.A. (1999). Flexible work schedules and mental and physical health: A study of a working population with non-traditional working hours. Journal of Organizational Behavior, 20(1), 35-46.

McLean Parks, J., Kidder, D.L., & Gallagher, D.G. (1998). Fitting square pegs into round holes: Mapping the domain of contingent work arrangements onto the psychological contract. Journal of Organizational Behavior, 19 (special issue), 697-730.

Nollen, S.D. (1996). Negative aspects of temporary employment. Journal of Labor Research, 17(4), 567-582.

OECD (2002). Employment outlook. Organization for Economic Co-operation and Development: Paris.

Pearce, J.L. (1998). Job insecurity is important, but not for the reason you might think: the example of contingent workers. In C. L. Cooper & D. M. Rousseau (Eds.), Trends in Organizational Behavior (31-46). New York: Wiley.

Pfeffer, J. (1998). The human equation: Building profits by putting people first, Boston: Harvard University Press.

Pfeffer, J., & Baron, J.N. (1988). Taking the workers back out: Recent trends in the structure of employment. In L.L. Cummings & B.M. Staw (Eds.), Research in Organizational Behaviour (257-303). Greenwich,CT: JAI Press.

Polivka, A.E. (1996). Contingent and alternative work arrangements, defined. Monthly Labor Review, October 1996, 3-9.

Rizzo, J.R., House, R.J., & Lirtzman, S.I. (1970). Role conflict and ambiguity in complex organizations. Administrative Science Quarterly, 15(2), 150-163.

Shore, L.F., & Barskdale, K. (1998). Examining degree of balance and level of obligation in the employment relationship: A social exchange approach. Journal of Organizational Behavior, 19 (Special issue), 731-744.

Sjöberg, A., & Sverke, M. (2000). The interactive effect of job involvement and organizational commitment on job turnover revisited: A note on the mediating role of turnover intention. Scandinavian Journal of Psychology, 41(3), 247-252.

Sverke, M., Gallagher, D.G., & Hellgren, J. (2000). Alternative work arrangements: Job stress, well-being, and work attitudes among employees with different employment contracts. In K. Isaksson, C. Hogstedt, C., Eriksson & T. Theorell (Eds.), Health effects of the new labor market (145-168). New York: Plenum.

Sverke, M., & Sjöberg, A. (1994). Dual commitment to company and union in Sweden: An examination of predictors and taxonomic split methods. Economic and Industrial Democracy, 15(4), 531-564.

Walsh, J., & Deery, S. (1999). Understanding the peripheral workforce: Evidence from the service sector. Human Resource Management Journal, 9(2), 50-63.

Ward, J. (1963). Hierarchical grouping to optimise an objective function. Journal of American Statistical Association, 58(3), 236-244.