- Define and discuss the concept of effort-reward imbalance (ERI), with special reference to the mechanisms by which it may affect health.
- Give examples of occupational groups with unfavorable versus favorable ERI ratios and summarize the new findings on the impact of ERI ratio on self-rated health.
- Discuss the implications of the new findings for occupational interventions to promote employee health, along with areas for future research into the health consequences of ERI.
Over the last two decades, the model of effort-reward imbalance (ERI) at work has become one of the leading theoretical approaches to understand the health-hazardous effects of an adverse psychosocial work environment. Originating from research in medical sociology in West Germany in the 1980s, the model focuses on the notion of social reciprocity as a fundamental principle of interpersonal behavior and social exchange.1,2 Social reciprocity is characterized by mutual investments based on the norm of return expectancy where efforts are equalized by respective rewards.1,2 Within the context of work, the ERI model posits that a “high cost/low gain” situation, in which individuals spend high effort while receiving low rewards at work, elicits emotional distress, which consequently affects health via both deteriorated health-related behaviors and psycho-neuroendocrionological and psycho-immunological pathways. It is further assumed, in the model, that ERI has in particular adverse health consequences when it co-occurs with a motivational disposition called “work-related overcommitment.”1,2
Siegrist et al2 developed a 17-item questionnaire to measure ERI that includes 6 items on effort and 11 items on reward (with subdimensions of financial and status-related reward, esteem reward, and job security reward). The ERI model has been tested most intensively in cardiovascular research, where it has been found to be associated with incident cardiovascular events in longitudinal studies.3 For other health outcomes, empirical evidence is less comprehensive. Studies are usually limited to cross-sectional analyses (for reviews see Refs. 4 and 5), and only recently researchers have started to investigate the effect of ERI on mental disorders6,7 and musculoskeletal disorders8 with longitudinal designs.
Self-rated health is considered as an excellent indicator of health status and a strong predictor for mortality.9–11 The empirical evidence for an effect of ERI on self-rated health is mixed. Most of the studies are cross-sectional in design and are therefore limited in drawing causal inference.4,5 To our knowledge, to date, only three studies have analyzed the effect of ERI on self-rated health prospectively. In the Whitehall II study (British civil servants), ERI predicted poor self-rated health in analyses by Stansfeld et al12 and Kuper et al13 but not in the most recent analysis by Kivimäki et al.14 In the French Gazel study (employees at France’s national gas and electricity company), Niedhammer et al15 reported that some operationalizations of ERI were associated with poor self-rated health whereas others were not. Finally, Kimimäki et al16 found in Finland that the association of ERI with self-rated health was stronger in local government employees compared with hospital workers.
Few studies have reported so far the distribution of ERI across occupational groups. In the British Whitehall II study, Kuper et al13 reported higher ERI ratios among civil servants of higher social grade, whereas a community sample from three cities of an industrialized urban region in Germany showed that middle managers had the highest and lower grade employees and the self-employed had the lowest ERI ratios.17 To our knowledge, no study has yet investigated the distribution of ERI in a representative sample of the general workforce. Furthermore, the prospective effects of ERI on self-rated health have not been studied in such a sample. Hence, we do not know which job groups are most highly exposed to ERI and we also do not know if findings on health effects of ERI can be generalized to the general working population. This study aims to fill this research gap by studying ERI in a representative sample of the Danish workforce. The objectives of this study are a) to investigate the distribution of effort, reward, and ERI by job group and to analyze to what extent job group explains the variance in effort, reward, and ERI; and b) to analyze whether ERI and its components are prospectively associated with risk of a decline in self-rated health.
Materials and Methods
Study Design and Population
The Danish Work Environment Cohort Study (DWECS) was established in 1990 and the cohort has since then been followed up every 5 years. The cohort has been supplemented with young people and immigrants at each follow-up. In each round, all people in the sample were contacted for interview, irrespective of their participation status in earlier rounds. The main purpose of DWECS is to monitor working conditions and health in the Danish workforce and to analyze the effect of work environment conditions on changes in health and labor market participation. A detailed description of the study design and population has been published elsewhere.18 For this article, we used the DWECS 2000 survey as the baseline and the DWECS 2005 survey as the follow-up data set.
The DWECS 2000 Sample.
In 2000, 11,437 Danish residents were approached to participate in DWECS. Of those, 8583 participated in the survey (response rate: 75%). Previous analyses have shown that the DWECS 2000 sample is representative of the Danish workforce.18 Data were collected by a telephone interview. Among the respondents, 5292 were employees at the time of the survey. We excluded 276 respondents, who were in job training or employed under special working conditions (eg, modified duty due to illness) and 39 respondents with a missing value on the effort-reward imbalance measure, resulting in a study sample of 4977 respondents. Of those, 2407 (48%) were women and mean age was 41 years (standard deviation: 11 years).
The DWECS 2000 to 2005 Cohort.
Of the 4977 employees in the DWECS 2000 survey, 3470 responded to the DWECS 2005 survey (follow-up response rate: 70%). Data were collected by a telephone survey, a self-administered questionnaire, or an internet survey. Because we aimed to analyze the effect of ERI in 2000 on a decline in health in 2005, we excluded 460 respondents, who had already “reduced health” at baseline (see the definition for reduced health given below). Further, we excluded 75 respondents with missing values on any variable in the multivariate analysis, resulting in a cohort of 2935 respondents.
Measurement of Effort-Reward Imbalance
Because DWECS did not include the original ERI questionnaire, we assessed effort and reward with proxy measures. DWECS did not contain items that could be used to approximate work-related overcommitment and consequently we did not include this construct.
The exact procedure for measuring ERI is described in the Appendix, which also includes a list of the effort and reward items. The effort score was calculated by summing up four items, derived from the Copenhagen Psychosocial Questionnaire.19 All items had five response categories ranging from “1 = never/hardly ever” to “5 = always.” The effort score was built by summing up the numeric values of the items.
The reward score was calculated by summing up seven items, representing the three subdimensions of the reward constructs that included two items on “financial and status reward,” three items on “esteem reward,” and two items on “job security reward.” The items had different response categories, but were scored in a way that for all items higher values indicated more rewards (see Appendix for details).
We constructed an “effort-reward imbalance ratio” (ERI ratio) with the effort score in the nominator and the reward score in the denominator. Hence, higher values of the ratio expressed a higher level of imbalance between high effort and low reward.
Measurement of Self-Rated Health
Self-rated health was measured with the question: “In general, how would you rate your health?” with the response categories “very good,” “good,” “fair,” “poor,” or “very poor.” We dichotomized responses in two categories: (1) “good health,” which included the responses “very good” and “good” and (2) “reduced health,” which included the responses “fair,” “poor,” and “very poor.” Study endpoint was “decline in self-rated health,” defined as moving from the “good health” category at baseline to the “reduced health” category at follow-up.
Measurement of Other Variables
In addition to ERI and self-rated health, we assessed gender, age, job group, occupational grade, indicators of health-related behaviors, and prevalence of depressive symptoms at baseline. To control for effects of the type of survey method on self-rated health, we also included a variable indicating whether the follow-up survey was administered as a self-administered questionnaire, on the internet, or on the telephone.
Employees were categorized into job groups based on information on job title classified by means of the extended Danish version20 of the International Standard Classification of Occupations (ISCO) and, in some cases, also based on type of industry. We further categorized employees into occupational grades, based on information on occupational position and vocational education, yielding five categories: I, executives and/or academics; II, middle managers and/or having more than 3 to 4 years of vocational education; III, other white-collar workers; IV, skilled blue-collar workers; V, semiskilled or unskilled blue-collar workers.
We included indicators of health-related behaviors into the analyses, because previous studies have shown associations between ERI and poor health behaviors.21 Indicators of health-related behaviors included smoking, alcohol consumption, leisure time physical activity, and body mass index (BMI). Smoking was categorized into “current smokers” versus “current nonsmokers.” Alcohol consumption was dichotomized in “no or moderate consumption” versus “heavy consumption,” with heavy consumption defined as drinking more than two (women) and three (men) units per day, respectively. A unit was defined as one small bottle of beer (33 cL), one glass of wine, or one shot of liquor. Leisure time physical activity was assessed with the question “When you should describe your leisure time physical activity in the last year, including commuting to or from work, to what group do you belong?,” with four response categories corresponding to sedentary, light, moderate, and strenuous physical activity. BMI was calculated based on self-reported height and weight and categorized into “underweight” (BMI <18.5), “normal weight” (18.5 to 24.9), “overweight,” (25 to 30), and “obese” (>30).
Prevalence of depressive symptoms were assessed with the five-item mental health scale from the SF-3622 with a cutoff point of 52, as suggested in the literature.23,24 We included this variable, because it has been argued that depressive symptoms and other states of negative mood might cause over-reporting of adverse working conditions.25,26
All analyses were conducted with the statistical program package STATA 8.0. Correlations were calculated with Pearson correlations. We plotted mean effort and reward scores against each other for all job groups with 20 or more respondents in DWECS 2000, and we calculated the mean ERI ratio for each job group. To determine the percentage of variance of the effort score, the reward score and the ERI ratio that was explained by job group, we calculated the adjusted r2 with an analysis of variance.
The effect of effort, reward, and ERI at baseline on a decline in health at follow-up was calculated by odds ratios (OR) and 95% confidence intervals (CI) with multivariate logistic regression models. Respondents with reduced self-rated health at baseline were excluded. The analyses included three models. Model 1 was adjusted for gender, age, good versus very good self-rated health at baseline, occupational grade, and type of survey method. Model 2 was additionally adjusted for smoking, alcohol consumption, leisure time physical activity, and BMI. Model 3 was additionally adjusted for prevalence of depressive symptoms at baseline. Finally, we repeated the analyses for each of the three subdimensions of the reward measure.
Distribution of ERI by Job Group
Figure 1 shows the effort and reward scores for each job group with 20 or more respondents. The highest effort scores were found among executives in the public sector, whereas municipal child minders had the lowest effort score. The highest reward scores were found among physicians and dentists, whereas mail carriers reported the lowest reward scores. The Pearson correlation between effort and reward score was 0.42.
Table 1 shows the distribution of the ERI ratio across job groups. The average ERI ratio across all job groups was 0.53 (standard deviation: 0.18). Job groups with a mean ERI ratio that was at least a half standard deviation higher than the average ERI ratio included executives in the public sector (mean ERI ratio of 0.65), social workers (0.63), managing clerks in the public sector (0.62), and medical secretaries (0.62). Job group explained 15% of the variance of the effort score, 4% of the variance of the reward score, and 7% of the variance of the ERI ratio.
ERI and Risk of a Decline in Health in 2005
Of the 2935 participants of the DWECS 2000 to 2005 cohort, who were in very good or good health at baseline, 465 (15.8%) showed a decline in self-rated health. Rates were slightly higher for men (17.1%) than for women (14.6%).
A one standard deviation increase in the ERI ratio at baseline was associated with a 16% increased risk of a decline in health at follow-up, after adjustment for gender, age, self-rated health at baseline, occupational grade, and type of survey method (Table 2; Model 1). When we analyzed the effort and reward scores separately, we found that both a one standard deviation increase in effort (OR = 1.12) and a one standard deviation decrease in reward (OR = 0.86) were associated with an increased risk of a decline in health. All effect estimates remained virtually the same after further adjustment for health-related behaviors (smoking, alcohol consumption, leisure time physical activity, BMI; Table 2, Model 2). Further adjustment for prevalence of depressive symptoms at baseline attenuated effect sizes; however, associations remained statistically significant for the ERI ratio (OR = 1.12) and the reward score (OR = 0.89).
When we stratified the analyses by gender, we found similar effect sizes for men and women. In Model 2, the effect sizes for ERI were OR = 1.16 (95% CI = 1.00 to 1.34) and OR = 1.19 (95% CI = 1.03 to 1.37) in men and women, respectively. In Model 3, the effect sizes were OR = 1.14 (95% CI = 0.98 to 1.32) and OR = 1.13 (95% CI = 0.98 to 1.30) in men and women, respectively (data not shown in table).
We repeated the analysis with the most adjusted model for each of the three subdimensions of the reward measure (Table 3). We found the strongest effect for “job security reward” with an OR of 0.82 for a one standard deviation increase. The effect size of financial and status reward (OR = 0.92) approached statistical significance, whereas esteem reward (OR = 0.97) was not associated with a decline in self-rated health.
The aims of this research report were a) to investigate the distribution of ERI in the Danish workforce and b) to analyze whether ERI predicts a decline in self-rated health. Here, we first discuss the results for these two research aims and then we address methodological issues regarding the construction of ERI in DWECS.
Distribution of ERI in the Danish Workforce
To our knowledge, this is the first study that has investigated ERI within a sample of a national workforce and therefore could compare the extent of ERI across a wide range of job groups. Effort and reward scores were well correlated with each other, indicating that reciprocity is in general high in the Danish working population. Nevertheless, there were some job groups, such as executives and managerial clerks in the public sector, social workers, and medical secretaries who showed, compared with the Danish average, a more unfavorable imbalance between efforts and rewards. Given the empirical evidence for the health-hazardous effects of ERI,4,5 including the findings on an increased risk of a decline in self-rated health in this study, these jobs might be considered for occupational interventions, in particular work organizational changes.27,28 On the other side, however, it has to be considered that job group explained only a relatively small fraction of the variance in ERI (and in particular in the variance of reward). This indicates that ERI scores aggregated at the job group level are only of limited use for occupational research and practice. Future studies are needed to investigate if aggregation at the workplace level (eg, work units, departments) might explain a greater proportion of the variance.
A limitation in the analysis is that—although the whole DWECS sample is representative for the Danish workforce18—we do not know if the respondents in the specific job groups are representative for people working in these jobs. This is especially a concern in those job groups with only 20 or 30 respondents and results therefore should be viewed with caution.
ERI as a Predictor for a Decline in Self-Rated Health
Epidemiological studies have shown that self-rated health is a reliable indicator of health status and a strong predictor for mortality in many countries, including Denmark.9–11 In our study, a one standard deviation increase of the continuous ERI ratio predicted a decline in self-rated health during the 5-year follow-up independently of health-related behavior and of prevalence of depressive symptoms at baseline. These results add to the evidence that ERI is a risk factor for poor self-rated health.12,13,15,16
The relative contribution of the effort and reward scores to risk of ill-health and the type of interaction between the two components have become a matter of research interest, recently.29 Our findings indicate that both high efforts and low reward contribute to a decline in self-rated health. When we analyzed the three subdimensions of the reward measure separately, we found the strongest effect for job security reward. This result is in line with a previous analysis of the DWECS 1995 to 2000 cohort, which showed that job insecurity was a predictor for a decline in health.30
It has been argued that associations between psychosocial working conditions and self-reported health outcomes might be biased by study participants who over-report both exposure and symptoms of ill-health, because of an underlying third factor, most notably negative affectivity.25 We accounted for this potential bias by adjusting the analyses for prevalence of depressive symptoms at baseline. However, it needs to be pointed out that it is controversially discussed in the literature, if one should or should not adjust for negative emotions in psychosocial occupational health studies. Negative emotions might be an intermediate step in the pathway from workplace exposures to incident poor health and therefore adjustment might result into an underestimation of the true effect of the workplace exposure.25,26,31,32
Methodological Issues Regarding the Construction of ERI Measures in DWECS
We had to use proxy measures for defining ERI, because the original ERI questionnaire was not used in DWECS. Bourbonnais33 has recently argued that proxy measures of ERI might result into an incomplete assessment of the construct and a misclassification of the exposure, which consequently might lead to either an overestimation or an underestimation of the association between ERI and health outcomes. We cannot rule out that this also applies to our study; however, we would like to point out that we used a comprehensive measure of ERI, which included four items to approximate effort and seven items to approximate reward. The seven reward items covered all three subdimensions of the reward concept that is financial/status reward, esteem reward, and job security reward. We consider this as a strength of our study, because previous studies that used ERI proxies usually had covered only one or two of the subdimensions.16,34,35
In the original ERI questionnaire, effort and reward are assessed by first asking the respondents about the occurrence of a specific aspect of effort or reward and then further asking the respondents to rate how much they feel “distressed” about this aspect.2 However, in this study, we measured occurrence and frequency of a effort and reward items, but not the level of self-reported distress. It has been controversially discussed in the literature, whether individual appraisal of the adversity of a psychosocial exposure should be measured or not. The advantage of assessing appraisal is to get a more complete picture about the importance of the exposure for the individual, whereas the disadvantage is the danger of blending exposure measurement (stressor) with outcome measurement (stress reaction).25,36
The Cronbach’s alphas of the effort and in particular of the reward score were only modest (see Appendix). Although Cronbach’s alpha is a popular estimate for reliability, the assumptions behind the approach are not often discussed. Cronbach’s alpha may overestimate reliability if some items show local dependency, for example, if they are too similar in content or wording.37 On the other hand, for scales that are multidimensional or contain items that are causal indicators, Cronbach’s alpha may underestimate reliability.38 Since both the effort and the reward scale contain causal indicator items and since the reward scale also consists of three subdimensions, Cronbach’s alpha may underestimate reliability for these scales.
With regard to the four effort items, it has to be considered that the type of effort they measure might be of different importance for different job groups. In another Danish workforce sample, Kristensen et al39 have demonstrated “differential item functioning” for various demand items, including the four items, we used as proxies for high effort. Based on these results, Kristensen et al suggest using different types of demand scales and especially to distinguish between high work pace (ie, pressure for intensification of work) and long working hours (ie, pressure for extensification of work). Accordingly, one might think about future studies with different effort scales that more precisely reflect actual workload for different job groups.
Regarding the seven reward items, special conditions of the Danish labor market have to be considered. The follow-up period of this study (2000 to 2005) was a time of economic prosperity with low unemployment rates in Denmark.40 Denmark has today the lowest income inequality (based on the Gini coefficient) among all Organisation for Economic Co-operation and Development (OECD) countries,41 and, despite weak employment protection laws, Danish employees reported the lowest level of job insecurity in a survey of 16 European nations.42
DWECS did not include items that would allow constructing a proxy measure for work-related overcommitment. This has to be regarded as a limitation of our study, because it is conceptually assumed that the presence of overcommitment heightens the health-hazardous effects of ERI.1,2
Conclusion and Future Research
This study has identified job groups with high effort-reward imbalance in Denmark and has shown that effort-reward imbalance is a risk factor for a decline in self-rated health. Future studies need to address the more specific health consequences of effort-reward imbalance and its components in the Danish workforce. In our research group, we are currently planning to link the DWECS data set to governmental registries on sickness absence, disability pensioning, hospitalization, and prescription of psychotropic medication to get information on specific health endpoints.
The analyses for this article have been supported by a grant from the Danish Working Environment Research Fund (grant number: 5-2006–04).
1.Siegrist J. Adverse health effects of high-effort/low-reward conditions. J Occup Health Psychol
2.Siegrist J, Starke D, Chandola T, et al. The measurement of effort-reward imbalance at work: European comparisons. Soc Sci Med
3.Kivimäki M, Virtanen M, Elovainio M, Kouvonen A, Vaananen A, Vahtera J. Work stress in the etiology of coronary heart disease—a meta-analysis. Scand J Work Environ Health
4.van Vegchel N, de Jonge J, Bosma H, Schaufeli W. Reviewing the effort-reward imbalance model: drawing up the balance of 45 empirical studies. Soc Sci Med
5.Tsutsumi A, Kawakami N. A review of empirical studies on the model of effort-reward imbalance at work: reducing occupational stress by implementing a new theory. Soc Sci Med
6.Netterstrøm B, Conrad N, Bech P, et al. The relation between work-related psychosocial factors and the development of depression. Epidemiol Rev
7.Godin I, Kittel F, Coppieters Y, Siegrist J. A prospective study of cumulative job stress in relation to mental health. BMC Public Health
8.Rugulies R, Krause N. Effort-reward imbalance and incidence of low back and neck injuries in San Francisco transit operators. Occup Environ Med
9.Idler EL, Benyamini Y. Self-rated health and mortality: a review of twenty-seven community studies. J Health Soc Behav
10.Burström B, Fredlund P. Self rated health: is it as good a predictor of subsequent mortality among adults in lower as well as in higher social classes? J Epidemiol Community Health
11.Osler M, Heitmann BL, Hoidrup S, Jorgensen LM, Schroll M. Food intake patterns, self rated health and mortality in Danish men and women. A prospective observational study. J Epidemiol Community Health
12.Stansfeld SA, Bosma H, Hemingway H, Marmot MG. Psychosocial work characteristics and social support as predictors of SF-36 health functioning: the Whitehall II study. Psychosom Med
13.Kuper H, Singh-Manoux A, Siegrist J, Marmot M. When reciprocity fails: effort-reward imbalance in relation to coronary heart disease and health functioning within the Whitehall II study. Occup Environ Med
14.Kivimäki M, Ferrie JE, Head J, Shipley MJ, Vahtera J, Marmot MG. Organisational justice and change in justice as predictors of employee health: the Whitehall II study. J Epidemiol Community Health
15.Niedhammer I, Tek ML, Starke D, Siegrist J. Effort-reward imbalance model and self-reported health: cross-sectional and prospective findings from the GAZEL cohort. Soc Sci Med
16.Kivimäki M, Vahtera J, Elovainio M, Virtanen M, Siegrist J. Effort-reward imbalance, procedural injustice and relational injustice as psychosocial predictors of health: complementary or redundant models? Occup Environ Med
17.Wege N, Dragano N, Erbel R, et al. When does work stress hurt? Testing the interaction with socioeconomic position in the Heinz Nixdorf Recall Study. J Epidemiol Community Health
18.Burr H, Bjorner JB, Kristensen TS, Tüchsen F, Bach E. Trends in the Danish work environment in 1990–2000 and their associations with labor-force changes. Scand J Work Environ Health
19.Kristensen T, Hannerz H, Høgh A, Borg V. The Copenhagen Psychosocial Questionnaire. A tool for the assessment and improvement of the psychosocial work environment. Scand J Work Environ Health
20.The Directorate of Labor. Dansk Fagkode
. Copenhagen: The Directorate of Labor; 1986.
21.Siegrist J, Rödel A. Work stress and health risk behavior. Scand J Work Environ Health
22.Ware JE Jr, Gandek B. Overview of the SF-36 Health Survey and the International Quality of Life Assessment (IQOLA) Project. J Clin Epidemiol
23.Berwick DM, Murphy JM, Goldman PA, Ware JE Jr, Barsky AJ, Weinstein MC. Performance of a five-item mental health screening test. Med Care
24.Strand BH, Dalgard OS, Tambs K, Rognerud M. Measuring the mental health status of the Norwegian population: a comparison of the instruments SCL-25, SCL-10, SCL-5 and MHI-5 (SF-36). Nord J Psychiatry
25.Kasl SV, Jones BA. An epidemiological perspective on research design, measurement, and surveillance strategies. In: Quick JC, Tetrick LE, eds. Occupational health Psychology
. Washington, DC: American Psychological Association; 2003:379–398.
26.Judge TA, Erez A, Thoresen CJ. Why negative affectivity (and self-deception) should be included in job stress research: bathing the baby with the bath water. J Organ Behav
27.Aust B, Ducki A. Comprehensive health promotion interventions at the workplace: experiences with health circles in Germany. J Occup Health Psychol
28.Semmer NK. Job stress interventions and the organization of work. Scand J Work Environ Health
29.Preckel D, Meinel M, Kudielka BM, Haug H-J, Fischer JE. Effort-reward-imbalance, overcommitment and self-reported health: is it the interaction that matters? J Occup Organ Psychol
30.Rugulies R, Aust B, Burr H, Bultmann U. Job insecurity, chances on the labour market and decline in self-rated health in a representative sample of the Danish workforce. J Epidemiol Community Health
31.Payne RL. Comments on ‘Why negative affectivity should not be controlled in job stress research: don’t throw out the baby with the bath waters.’ J Organ Behav
32.Spector PE, Zapf D, Chen PY, Frese M. Why negative affectivity should not be controlled in job stress research: don’t throw out the baby with the bath water. J Organ Behav
33.Bourbonnais R. Are job stress models capturing important dimensions of the psychosocial work environment? Occup Environ Med
34.Bosma H, Peter R, Siegrist J, Marmot M. Two alternative job stress models and the risk of coronary heart disease. Am J Public Health
35.Kouvonen A, Kivimäki M, Virtanen M, Pentti J, Vahtera J. Work stress, smoking status, and smoking intensity: an observational study of 46,190 employees. J Epidemiol Community Health
36.Frese M, Zapf D. Methodological issues in the study of work stress: objective vs subjective measurement of work stress and the question of longitudinal studies. In: Cooper CL, Payne R, eds. Causes, Coping and Consequences of Stress at Work (Wiley Series on Studies in Occupational Stress)
. Chichester, England: John Wiley & Sons; 1988:375–411.
37.Sireci SG, Thissen D, Wainer H. On the reliability of testlet-based tests. J Educ Meas
38.Bollen KA, R. L. Conventional wisdom on measurement: a structural equation perspective. Psychol Bull
39.Kristensen TS, Bjorner JB, Christensen KB, Borg V. The distinction between work pace and working hours in the measurement of quantitative demands. Work Stress
40.Kvist J. A Danish welfare miracle? Policies and outcomes in the 1990s. Scand J Public Health
42.Böckerman P. Perception of job instability in Europe. Social Indicat Res
Appendix: Documentation of the Assessment of Effort-Reward Imbalance in the Danish Work Environment Cohort Study
The following shows the four effort items (ef1, ef2, ef3, ef4) and the seven reward items (rw1, rw2, rw3, rw4, rw5, rw6, rw7) that were used to measure effort-reward imbalance in DWECS.
Effort Score (ef1 + ef2 + ef3 + ef4)
If a respondent has 1 missing value, the missing value is replaced by the mean value of the item from all other respondents. If a respondent has 2 or more missing values, the respondent is excluded from the analysis. The potential range of the effort scale was from 4 to 20. Cronbach’s alpha was 0.63.
Reward Score (rw1 + rw2 + rw3 + rw4 + rw5 + rw6 + rw7)
If a respondent has 1 or 2 missing values, the missing value is replaced by the mean value of the item from all other respondents. If a respondent has 3 or more missing values, the respondent is excluded from the analysis. If a respondent has answered “I have no colleagues” to item rw4 and/or “I have no supervisor” to item rw5, the respondent is assigned the mean value of this item from all other respondents. The potential range of the reward scale was from 7 to 27. Cronbach’s alpha was 0.53.
Effort-Reward Imbalance Ratio
The effort-reward imbalance ratio was calculated by dividing the “effort score” (numerator) by the “reward score” (denominator). Hence, a higher ERI ratio indicates a higher level of imbalance between high efforts and low rewards. Note, that Siegrist et al2 have suggested using a correction factor for the uneven number of effort and reward items, so that a ratio of 1.0 indicates a perfect balance between effort and reward. However, we could not use this correction factor in this study, because number of response categories differed between the items.