1
COMPARING AND COMBINING THE DEMAND-CONTROL-SUPPORT
MODEL AND THE EFFORT REWARD IMBALANCE MODEL TO PREDICT
LONG-TERM MENTAL STRAIN
2
Abstract:
A purpose of this prospective study was to compare the predictive power of baseline
measures of the Effort-Reward Imbalance (ERI) model and the Demand-ControlSupport
(DCS) model in regard to long-term mental strain. The study further aimed to
investigate whether combining the models would improve the predictive power in
regard to long-term mental strain for two occupational groups, managers/professionals
(n=660) and manual workers (n=385). Both models explained small albeit significant
proportions of variance in mental strain for both occupational groups, the slightly
higher explanatory power found for ERI model compared to the DCS model may in
part have been due to the lower consistency in the measurements of the latter model.
As expected both models explained larger proportions of variance in mental strain for
manual workers than for the professional group. When introduced after DCS model in
the regression equation the ERI model explained additional variance in mental strain
whereas the DCS model failed to add explanatory power to mental strain when
introduced last in the equation. A conclusion from the study is that combining the two
models may increase the explanatory power to predict work-related mental strain.
3
Work-related stress has been firmly associated with a broad spectrum of negative
health outcomes and with impeded wellbeing. In order to reduce the complexity of the
relationships between psychosocial work environment and the health reactions of the
individual, several generic models have been developed in order to identify crucial
elements in the pathway between the psychosocial working conditions and health
outcomes (Van Vegchel, de Jonge, Bosma & Schaufeli, 2005). In relation to health
Sigriest & Marmot (2004) define the psychosocial environment in terms of the
opportunities available to support the individuaľs need of positive self-experience.
Self-efficacy, the individuaľs belief in his/her ability to accomplish tasks, and selfesteem,
the continued positive experience of a person's self-worth, are according to
Siegrist & Marmot (2004) two aspects of self-experience that are of crucial
importance to health and wellbeing. Two models that have received special attention
in the recent past, the Demand-Control/(Support) (DC/S) model and the EffortReward
Imbalance (ERI) model, analyse the relationship between the psychosocial
opportunity structure provided by work and health outcomes, by the pathway of workrelated
self-experience, according to Siegrist and Marmot (2004).
The Demand-Control-Support Model. The DCS model comprises three dimensions of
psychosocial working conditions (Karasek and Theorell, 1990; Karasek, Brisson,
Kawakami, Houtman, Bongers, and Amick, 1998). Psychosocial work demand relates
to how hard and intense the job holder has to work and includes for example time
pressure and quantitative workload. Control or decision latitude comprises two
distinct but closely related components. Task authority reflects the scope of the job
holder's authority to make decisions at work, while skill discretion relates to the level
and variety of the skill required for the work tasks and the long-term possibilities to
acquire new skills in the work role. The third dimension, later added to the model,
4
work-related social support, mainly refers to emotional and instrumental support from
colleagues and immediate superiors (Karasek and Theorell, 1990).
In psychological terms the model is characterised by a stimuli approach, where
the external environment is assumed to determine the emotional reactions and
behaviour of the individual, while psychological traits or subjective factors are not
taken into consideration (Karasek et al., 1998, Karasek and Theorell, 1990). The
Strain Hypothesis derived by the model claims that adverse long-term psychosocial
working conditions, high psychosocial demands in combination with low control
opportunities, will reduce the job holder's self-efficacy and create high psychophysiological
strain, which in turn will exert a negative impact on health, possibly
mediated through neuropsychological pathways (Frankenhaeuser and Johansson,
1986; Karasek and colleagues, 1979; 1990; 1998). High-strain job, characterized by
high demands and low control, are mainly to be found in machine paced work and in
low-status service work (Karasek and Theorell, 1990).
By exclusively reviewing longitudinal studies based on the DC(S) model,
with a large range of outcome health and well-being measures, de Lange, Taris,
Kompier, Houtman & Bongers (2003) concluded that there was good evidence for
longitudinal causal main effects between psychosocial working conditions included in
the DC/S model and health and well-being outcomes, especially self-reported
outcomes. On the other hand, this review provided only limited support for the high
strain assumption of the model.
The Effort-Reward Imbalance Model. The Effort-Reward Imbalance (ERI) model has
its origin in medical sociology and represents a more recent approach to the relation
between work and health (Siegrist, 1996; Marmot, Siegrist, Theorell, & Feeney,
1999). The ERI model takes not only the work content into account, but also the work
5
role in a social perspective, as well as the job holder's individual coping pattern and
need for control (Calnan, Wainwright, & Almond, 2000; Siegrist, 1996).
The ERI model conceptualizes work in a social exchange perspective and the
work role is seen as a link between the social opportunity structure and self-regulatory
need of the person (Siegrist, 1996; 2000). The ERI model shifts the attention from
control in work, to the social and economical rewards from work. For most people the
work role provides the main reciprocal exchange link between the individuaľs selfregulatory
system (e.g. self-esteem) and the social reward system (Siegrist, 1996). The
reward dimension of the ERI model embraces three so called transmitter systems by
which rewards are distributed - status control, economical rewards and esteem
(Siegrist 1996; 2000). The concept of status control has a central role in the theory
behind the model, and refers to the degree of strength and security the work role
provides to support the self-regulatory system of the individual (Siegrist, 1996). A
loss of status control is thought of as being strongly associated with negative feelings,
Siegrist (1996) suggests that low status control is a more threatening and therefore
potentially more stressful condition than low task control.
The effort dimension of the ERI model distinguishes between extrinsic and
intrinsic efforts (Siegrist, 1996). The extrinsic component concerns job demands and
obligations while the intrinsic component refer to over-commitment, defined as "a set
of attitudes, behaviours and emotions reflecting excessive striving in combination with
a strong desire of being approved and esteemed" (Siegriest, 2004, p. 2). High efforts
are thus thought of as occurring either from high extrinsic job demand and/or from the
coping pattern of the individual worker.
The ERI model predicts that the working conditions most likely to elicit
emotional and physiological strain reactions are those with a lack of reciprocity
6
between the (extrinsic and/or intrinsic) efforts invested in work ­ and the rewards
gained from work (Siegrist, 1996; 2002). An important prerequisite for the model is
that the social rewards are unevenly distributed within the labor market, and that the
really exhausting jobs often are the worst rewarded. Therefore an effort-reward
imbalance is most likely to occur in occupations with low status control, e.g. manual
jobs (Siegrist, 1996). Two recent reviews (Tsutsumi & Kawakami, 2004; van Vegchel
et al., 2005), concludes that the imbalance hypothesis between high extrinsic efforts
and low rewards has gained considerable support for a wide range of health outcomes.
The findings were more contradictory on the moderating role of over-commitment in
the relation between ERI and health (van Vegchel et al., 2005).
A conceptual comparison between the DCS model and the ERI model. In a wider
sense both models integrate sociological and psychological theories to conceptualise
and analyse the relationship between the psychosocial working conditions and health
outcomes and there are overlapping features between them. The Demand dimension
of the DCS model and the Extrinsic Effort dimension of the ERI model both mainly
refer to time pressure and quantitative workload. Although the Reward dimension of
the ERI model is attending towards more general social and economical opportunities
and outcomes of the work, the esteem dimension has an obvious overlap with social
support dimension of the DCS model, conceptual as well as in operational terms.
The intrinsic effort component of the ERI model refers to the individual workrelated
coping pattern of over-commitment and inability to withdraw from work (de
Jonge, Bosma, Peter & Siegriest, 2000; Siegrist, 1996; van Vegchel et al., 2005). By
inclusion of individual characteristics in the ERI model, the subjective meaning of
work experience is taken into consideration in the work stress process (Calnan et al.,
2000). Furthermore, the ERI model takes labor market factors and the work role in a
7
macro-social perspective into account (Calnan et al, 2000; Marmot et al., 1999; Peter
et al., 2002; Siegrist, 1996). While the DCS model is focused on the immediate work
content and implies that work task control is critical to counter job strain, the ERI
model is mainly focused on the broader concept of fairness and reciprocity in the
social exchange process (Marmot et al., 1999; Siegrist 1996). By providing
information of different but related aspects of the relationship between the
psychosocial work environment and health, the two models may complement each
other (Siegrist & Marmot, 2004). These authors claims that the DCS model, with iťs
focus on the content of the work tasks links the psychosocial working conditions and
self-efficacy, whereas the focus of ERI model on the reward structure "defines a link
between the opportunity structure at work and the working persons' experience of
self-esteem" (p. 1466).
Due to the complimentary features of the DC(S) and ERI models and the
comprehensive and firm empirical support for the two models, there have been calls
in the literature for further prospective studies to evaluate the predictive power on
work-related ill-health by combining the two models to capture a broader range of
potential job stressors (de Jonge et al., 2000; Marmot et al., 1999; Peter et al., 2002;
Siegrist & Marmot, 2004). A number of previous studies have reported improved
predictive power on combining the models (Bosma, Peter, Siegrist, & Marmot, 1998;
de Jonge et al., 2000; Ostry, Kelly, Demers, Mustard, & Hertzman, 2004). On the
other hand, in a recent cross-sectional study the ERI model was consistently found to
be more strongly related to self-reported measures on wellbeing compared to the DCS
Model (Calnan et al., 2004).
Different types of working conditions may be of importance for the emergence
of work-related stress in different occupations (Marmot et al., 1999, Sparks and
8
Cooper, 1999). It has been suggested that the two models may be associated with
distinct contributions in explaining work stress in different occupations (Marmot et
al., 1999). Both models do although predict the most adverse working conditions to be
found in low-status manual service or blue-collar occupations (Karsek and Theorell,
1990; Siegrist, 1996; Theorell and Karasek, 1996). The DCS model was focused on
job strain among industrial workers and the model has shown its strongest power in
predicting adverse health outcomes for manual workers (Karasek & Theorell, 1990).
Also in regard to the ERI model, Siegrist (1996; 2000) implies that low status control
and therefore low rewards are most commonly associated among persons in lowqualified
occupations, mainly for manual workers. Van Vegchel et al., (2005) claims
that the ERI model mainly has been applied in the human service sector and calls for
further studies with other occupational groups to generalize the applicability of the
findings.
The first purpose of the present study was to compare the predictive power of
the DCS model and the ERI model in regard to long-term predictive power on mental
strain. A second purpose of the study was to examine whether combining the DCS
model and the ERI model will improve longitudinal predictive power on mental
strain. Although previously reported findings on combining the models are somewhat
contradictory, the two models do to some extent refer to different sources of work
stress. It may therefore be reason to expect a somewhat improved predictive power by
combining the models. A third purpose of the study was to compare the predictive
power of the combined model for two different occupational groups: managers/
professionals and manual workers. Based on the conceptual content of both models
we presume that they, separate or combined, will predict long-term mental strain
better for manual workers than for the professional group. This study may add to
9
existing knowledge by providing comparative data of the long-term influence on
mental health by the components of the two recently most attended models of workrelated
stress, for two occupational groups with heterogeneous working conditions
and therefore potentially distinct sources of work-related stress.
Method
Respondents. The respondents were all participating in a large longitudinal
study on work stress and work-related musculoskeletal and mental disorders
(Devereux, Rydstedt, Kelly, Weston, & Buckle, 2004). The study involved in all
twenty British companies and public organizations - with a large variety in size and
location, within a large variety of industrial sectors. The companies were selected
strategically to include the largest possible variety of professions and working
conditions into the study (the authors do although not claim the sample to be
statistically representative for the British workforce). The selection procedure of
participants varied between the companies. In some of the (mainly) smaller
companies the entire workforce was drawn to the sample, whereas in other cases the
participants were randomly selected by the researchers from the staff lists.
About 8000 participants were included in the initial sample, and the response
rate was about 40%. Manual workers were over-represented in the non-respondent
group. For the comparison a short questionnaire with some core questions on
musculoskeletal complaints and perceived job stress were distributed to a sub-sample
of non-respondents, whom, compared to the participants, reported higher job stress
(21% of the non-respondents compared to 12% of the participants found their jobs
"very" or "extremely" stressful). The second and third wave of data collection
included only those respondents that participated in the baseline study. After attrition
(mainly due to turnover and downsizing in some of the companies) the sample for the
10
final wave of data collection consisted of 2640 persons. About 86% of the participants
at Time 1, or 2270 persons, completed the questionnaire at follow-up. The attrition
did not affect the age and gender distribution in the sample.
Based on the job title reported in the questionnaire the participanťs sub-major
occupational groups were classified according to the Standard Occupational
Classification 2000 (SOC2000). For this study two major occupational categories was
compared in the analyses, managers/professionals (major occupational groups 1-2,
SOC2000) and manual workers (major occupational groups 8-9, SOC2000). Only
participants with complete data for all the actual variables were selected.
The professional group consisted of 660 persons, whereof 30% were females
and 70% males, with an average age 41.5 years (Sd 9.8). Three out of ten (30%) of the
participants in the professional group were "managers and senior officials", while
38% were classified as "science and technology professionals", 15% as "business and
public service professionals", 17% as "teaching and research officials" and 1% as
"health professionals". The group of manual workers consisted of 385 persons,
whereof 15% females and 85% males, with an average age of 42.1 years (Sd 9.7). The
largest sub-major groups were "transport and mobile machine drivers" (42%) and
"process, plant and machine operatives" (35%). Moreover 16% of the manual workers
were in "elementary trades, plant and storage related occupations" and 7% held
"elementary administrative and service occupations".
Independent variables. The Demand-Control-Support model was assessed by
the Job Content Questionnaire (JCQ) modified for the Whitehall II study (Stansfeld,
Head & Marmot, 2000). Demands were measured by four items with  coefficient of
0.66. Work-related control was measured by 16 items with  coefficient of 0.87, ten
11
aimed at decision authority in the work situation and six items aimed at skill variety.
Work-related social support was measured by seven items with  coefficient of 0.79.
The ER questionnaire (Siegrist, 2004; Siegrist, Starke, Chandola, et al., 2004)
was used for assessing the ERI model. Subjects rated extrinsic efforts and rewards on
a five point scale. Efforts were measured by 5 items (excluding the item of physical
demands) with the  coefficient 0.78. Rewards were measured by 11 items, with the 
coefficient of 0.86. Intrinsic efforts (over-commitment) were measured by six items
with an  coefficient of 0.86.
Outcome variable The short version of the General Health Questionnaire (12)
was used as an indicator of mental strain (Goldberg and Williams, 1998), with the 
coefficient 0.86.
Statistical analyses. To control for the baseline value of the outcome variable
hierarchic multiple regression analysis was used, where the baseline value of mental
strain (GHQ) was introduced in the first step of the equation. For the first equation the
components of the DCS model were introduced in the second step, while the ERI
model was introduced in the third step, to examine whether additional predictive
power could be gained by combining the models. For the second equation the models
were introduced in the reversed order that is the ERI model was introduced before the
DCS model. Separate analyses were made for the professional group and the manual
workers. Since the interaction effects in the DCS model did not add significantly to
the explained variance, they have been excluded from the equations.
Results
Table 1 presents the inter-correlations between the dimensions of the two
models, the baseline and outcome measures of mental strain (GHQ), separately for the
professional group and manual workers.
12
INSERT TABLE 1 ABOUT HERE
Table 1 show that the professionals reported significantly higher job demands
(t=8.1, p<.01) but also higher control (t=30.2, p<.01) and work-related social support
(t=6.3, p<.01) than the manual workers. The professionals also reported higher
Intrinsic Efforts (t=5.2, p<.01) but also more satisfying Rewards from work ((t=6.9,
p<.01) than the manual workers. The professionals reported higher mental strain at
baseline as well as in the follow-up (t= 2.5/2.3, p<.05).
As further revealed in Table 1, there was a correlation between Demands in
the DCS model and Extrinsic Efforts (r=.64 for the professional group and .49 for
manual workers) as well as between Demands and Intrinsic Efforts in the ERI model
(r=.48 for the professional group and .33 for manual workers respectively), thus,
indicating a certain degree of overlap between the models. The correlation between
control and rewards was .38 for the professional group and .35 for manual workers,
while the correlation between support and rewards was 0.56 for the professional group
and .53 for the manual workers. This indicates a slightly higher overlap between the
dimensions of the models in this study compared to what was reported by Siegrist
(2004). The correlation between baseline and outcome mental strain was considerably
weaker for the professional group (.29) than for the manual workers (.48).
Initial introduction of the DCS model to predict long-term mental strain. Step
II in Table 2 shows the relation between the components of the DCS model at baseline
and perceived mental strain in the follow-up. After controlling for baseline mental
strain the DCS model accounted for 1.4% of the variance in outcome mental strain for
the professionals, and 4.4% for the manual workers. The beta weights reveals that Job
Demands related significantly to the outcome measure of mental strain for the
professionals whereas for the manual workers Social Support significantly predicted
13
the outcome. Thereafter, when introduced in the third step the ERI Model also
accounted for a significant proportion of the variance in mental strain for the
professional group (R2
=.021) as well as for the manual workers (R2
=.027).
INSERT TABLE 2 ABOUT HERE
In regard to the separate components of the ERI model, Rewards predicted
variance in mental strain for both occupational groups, as shown by the beta-weights
in Table 2. Intrinsic effort was related to mental strain only in the professional group.
Initial introduction of the ERI model to predict long-term mental strain. Step II
in Table 3 shows the relation between the components of the ERI model at baseline
and perceived mental strain in the follow-up, after control for the baseline measure of
GHQ. The ERI model accounted for significant proportions of the variance in mental
strain for the two occupational groups, 3.6% for the professionals and 6.1% for the
manual workers. When introducing the DCS model in the third step, it did not
significantly contribute to predicting the variance in outcome mental strain for any of
the occupational groups.
INSERT TABLE 3 ABOUT HERE
In regard to the separate components of the ERI model Rewards related
significantly to mental strain for both groups whereas Intrinsic Efforts related
significantly to mental strain among only in the professional group.
Discussion
When initially introduced both the DCS model and the ERI model related
significantly to long-term mental strain for both occupational groups. When the DCS
model was introduced in the final step of the equation, after the ERI model, it did not
explain any additional variance in mental strain. On the other hand, when the ERI
model was introduced in the final step of the equation it contributed significantly to
14
the explained long-term variance in mental strain for both occupational groups. For
the manual workers the latter analytical strategy gave the highest total proportion of
explained variance in mental strain - 7.0%. For the professional group the combined
model did not predict mental strain better than the ERI model alone. At least for
manual workers the present results suggested that combining the two models may
improve long-term prediction of mental strain by psychosocial work characteristics.
In terms of explained variance the ERI model emerged as the more powerful
predictor of mental strain, alone or in combination with the DCS model. This is in line
with recently reported findings (Calnan et al., 2004; Ostry et al., 2004). Due to some
remaining concerns about the fairness of the present comparison no far reaching
conclusions about their relative strength in predictive power of the models should
although be drawn from this study. For fair tests of comparative effects between
models Cooper & Richardson (1986) claims that the models should have been
measured with equivalent strength. This includes distributional equivalence or the
range of values the measurements of the models takes in the population. The measures
covered about the full range of possible values for the different components of both
models and the standard deviations of the measurements were relatively comparable,
which thus indicates acceptable distributional equivalence in the study. Equivalent
strength also concerns procedural equivalence, that "the competing theories, factors
or variables are operationalized, manipulated or measured with equal care and
fidelity" (Cooper & Richardson, 1986, p. 179). Even though both models were
measured by well established scales the alpha coefficient for the subscale of the
demand component in the DCS scale was markedly weaker (0.66) than in the subscale
of the ERI model measuring the extrinsic effort component (0.78). Even though this
consistency measure from the present study was in level with what was reported by
15
Stansfield et al. (2000) on the same measure (0.67), the relatively low reliability of
this subscale may have caused a disadvantage for the DCS model and therefore
rendered the comparison between the models with some degree of unfairness.
The long-term predictive power of the combined models never went beyond a
small effect size (Cohen, 1988). From the point of view of content and
methodological limitations small longitudinal stressor-strain relationships, even in
case of multiple correlations such as in this study, are although to be expected (Zapf,
Dormann & Frese 1996). The multi-causal antecedents to mental strain and
unavoidable measurement bias are among the most prominent reasons for the limited
effect sizes of these relations (Zapf et al., 1996). Controlling for the initial level of the
outcome variables may also have led to an underestimation of the magnitude of the
long-term relationships. Most of the participants have been exposed to the same type
of potentially adverse psychosocial working conditions for a prolonged period of
time. It therefore seems reasonable to assume that also the initial level of mental strain
was related to the psychosocial working conditions in the actual job. The effect sizes
may also have been reduced by sample attrition, since the attrition analysis a slightly
higher level of job strain among the non-respondents compared to the participants.
Both models accounted for higher proportion of variance in mental strain for
manual workers than for professionals and managers. This is consistent with
previously reported findings as well as with the conceptual content of both the models
­ primarily intended to assess work stress in manual low status occupations (Marmot
et al, 1999; Karasek & Theorell, 1990; Siegrist, 1996). In particular the Reward
component of the ERI model, referring to low returns from work in terms of money,
esteem and status control were shown to be related to long-term mental wellbeing for
the manual workers. Siegist and Marmot (2004) suggests that lower socioeconomic
16
status may increase the negative health effects of adverse psychosocial working
condition due to more frequent exposure and/or due to increased vulnerability.
The main reason to develop general work stress models has been to identify
core elements in the complex stressor-strain relationship (van Vegschel et al., 2005),
which is of central importance for the theoretical development of this field of
research. On the other hand Sparks and Cooper (1999) argues that the use of generic
models embracing a limited range of potential stressors possibly have led to the
attention away from other work characteristics of importance for well-being and
therefore argues for job-specific models. The present results indicate that what
psychosocial factors predict mental strain seems to some extent to be dependent on
the occupations and organisations being investigated. Even though general models
best identifies core sources of work-related strain more job-specific models are also
needed to explain the different sources of strain in different occupations.
While the ERI encompasses individual traits of the job holder, the DCS model
adopts a stimuli approach, where psychological traits or subjective factors are not
taken into consideration in the perception of the psychosocial work environment.
Nevertheless Siegrist & Marmot (2004) suggests that both the models offer
complementary information on the pathways between the psychosocial work
conditions and health by the mediation through individual psychological processes,
self-efficacy for the DCS model and self-esteem for the ERI model. To further
understand the role of psychosocial working conditions for health outcomes it will of
importance to further illuminate and analyze the so far still hypothetical mediating
link of psychological processes in the long-term work-related stressor-strain
relationship.
17
References
Bosma, H., Peter, R., Siegrist, J., & Marmot, M. (1998). Two alternative job
stress models and the risk of coronary heart disease. American Journal of Public
Health, 88(1), 68-74.
Calnan, M., Wadsworth, E., May, M., Smith, A., & Wainwright, D. (2004).
Job strain, effort-reward imbalance and stress at work: Competing or complementary
models? Scandinavian Journal of Public Health, 32(2), 84-93.
Calnan, M., Wainwright, D., &Almond, S. (2000). Job strain, effort-reward
imbalance, and mental distress: A study of occupations in general medical practice.
Work and Stress, 14(4), 297-311.
Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences,
(second edition). Hillsdale: Erlbaum Associates.
Cooper, W. h., & Richardson, A. J. (1986). Unfair comparisons. Journal of
Applied Psychology, 71(2), 179-184.
De Jonge, J., Bosma, H., Peter, R., & Siegriest, J. (2000). Job strain, effortreward
imbalance and employee wellbeing: A large-scale cross-sectional study. Social
Science and Medicine, 50, 1317-1327.
Devereux, J., Rydstedt, L., Kelly V., Weston, P., and Buckle, P. (2004). The
Role of Work Stress and Psychological Factors in the Development of
Musculoskeletal Disorders: The Stress and MSD Study. Contract Research Report
273/2004. HSE Books.
Goldberg, D.P., & Williams, P. (1988). A User's Guide to the General Health
Questionnaire. Windsor: NFER-Nelson
Karsek, R. (1979). Job demands, job decision latitude and mental strain:
Implications for job redesign. Administrative Science Quarterly, 24, 285-308.
18
Karasek, R., Brisson, C., Kawakami, N., Houtman, I., Bongers, P., & Amick,
B. (1998)). The Job Content Questionnaire (JCQ): An instrument for internationally
comparative assessments of psychosocial job characteristics. Journal of Occupational
Health Psychology, 3(4), 322-355.
Karasek, R.A., & Theorell, T. (1990). Healthy Work: Stress, Productivity and
the Reconstruction of Working Life. New York, NY: Basic Books.
Marmot, M., Siegrist, J., Theorell, T., & Feeney, A. (1999). Health and the
psychosocial environment at work. In M. Marmot and R G Wilkinson (Eds.) Social
Determinants of Health,105-131. Oxford University Press.
Ostry, A. S., Kelly, S., Demers, P. A., Mustard, C., & Hertzman, C. (2003). A
comparison between the effort-reward imbalance and demand control models.
BioMed Central Public Health, 3(10).
Siegrist, J. (1996). Adverse health effects of high-effort/low reward
conditions. Journal of Occupational Health Psychology, 1, 27-41.
Siegrist., J. (2000). The effort-reward imbalance model. Occupational
Medicine: State of the Art Reviews, 15(1), 83-87.
Siegrist, J. (2004). The Model of Effort-Reward Imbalance. www.uniduesseldorf.de/MedicalSociology.
(Accessed the 27 July 2005)
Siegrist, J., & Marmot, M. (2004). Health inequalities and the psychosocial
environment ­ two scientific challenges. Social Sciences & Medicine, 58, 1463-1473.
Siegriest, J., Starke, D., Chandola, T., Godin, I., Marmot, M Niedhammer, I.,
& Peter, R. (2004). The measurement of Effort-Reward Imbalance at work: European
comparisons. Social Science & Medicine, 58(8), 1483-1499.
Standard Occupational Classification 2000. (SOC2000). Office for National
Statistics. London: The Stationary Office.
19
Sparks, K., & Cooper, C. L. (1999). Occupational differences in the workstrain
relationship: Towards the use of situation-specific models. Journal of
Occupational and Organizational Psychology, 72, 219-229.
Stansfeld, S., Head, J., and Marmot, M. (2000). Work-Related Factors and IllHealth:
The Whitehall II Study. Norwich, UK: HSE Books: Contract Research Report
266.
Tsutsumi, A., & Kawakami, N. (2004). A review of empirical studies on the
model of effort-reward imbalance at work: Reducing occupational stress by
implementing a new theory. Social Science & Medicine, 59, 2335-2359.
Van Vegchel, N., de Jonge, J., Bosma, H., & Schaufeli, W. (2005). Reviewing
the effort-reward imbalance model: Drawing up the balance of 45 empirical studies.
Social Science & Medicine, 60, 1117-1131.
Zapf, D., Dorman, C., & Freese, M. (1996). Longitudinal studies in stress
research: a review of the literature with reference to methodological issues. Journal of
Occupational Health Psychology, 1(2), 145-169.
20
Table 1.
Inter-Correlations between Baseline Working Conditions and Mental Strain at
Follow-Up. Manual Workers (n=385) Above the Diagonal; Managers/
Professionals (n=660) Below the Diagonal.
1 2 3 4 5!!
6 7 8
Manual M 2.65 2.37 2.86 2.06 406 2.08 1.82 1.95
workers Sd 0.58 0.48 0.59 0.66 0.72 0.58 0.31 0.44
Professionals
M Sd.
1 2.93 0.52 - 13 -16 .49 -26 .33 .19 .18
2 3.18 0.37 .0! - .41 -08! .35 -05! -13 -15
3 3.08 0.51 -22 .35 - -23 .53 -17 -17 -28
4 2.14 0.67 .63 -09 -31 - -57 .57 .37 .34
5!!
4.36 0.63 -17 .38 .56 -35 - -28 -32 -37
6 2.27 0.54 .48 -00! -24 .51 -13 - .46 .34
7 1.88 0.35 .22 -16 -32 .33 -27 .37 - .48
8 2.02 0.47 .17 -07! -16 .22 -19 .24 .29 -
!
p> .05
!! Higher value indicates higher rewards
1. Demands (DCS) 5. Rewards (ERI)
2. Control (DCS) 6. Intrinsic efforts (ERI)
3. Support (DCS) 7. Mental strain at baseline
4. Extrinsic efforts (ERI) 8. Mental strain at follow-up
21
Table 2.
Hierarchic Multiple Linear Regression Analysis. Control for Baseline Mental
Strain. The DCS Model Followed by the ERI Model - in Relation to Mental Strain
for (1) the Professional Group (N=660 and (2) Manual Workers (N=385).
1 2
I. Baseline DV
R2
.083** .234**
II. DCS model
R2
.097** .278**
R2
.014** .044**
 weights:
Demands .10* .07
Control -01 -01
Support -06 -19**
III. ERI model
R2
.118** .305
R2
.021** .026**
 weights:
Extrinsic efforts .05 .04
Rewards -11* -14*
Intrinsic efforts .12** .10
*p<0.05 **p<0.01
22
Table 3
Hierarchic Multiple Linear Regression Analysis. Control for Baseline Mental
Strain, The ERI Model Followed by the DCS Model - in Relation to Mental Strain
for (1) the Professional Group (N=660) and (2) Manual Workers (N=385).
1 2
I. Baseline DV
R2
.083** .234**
II. ERI model
R2
.118 .295**
R2
.036** .061**
 weights:
Extrinsic efforts .06 .03
Rewards -10* -21**
Intrinsic efforts .13** .11
III. DCS model
R2
.119 .305**
R2
.000 .010
 weights:
Demands .01 .01
Control .01 -00
Support .01 -12*
*p<0.05 **p<0.01