A refutation of Karasek's demand – discretion model of occupational stress with a range of dependent measures.

Ben Fletcher, Fiona Jones

Research output: Contribution to journalArticlepeer-review

112 Citations (Scopus)


Karasek's (1979) demand–discretion model of occupational stress has been highly influential but has been criticized as confounding social class variables (Payne and Fletcher, 1983), ignoring the curvilinearity of relationships (Warr, 1991), and because it has inconsistently predicted heart disease (Ganster and Fusilier, 1989). The present study examines the ability of the model to predict raised blood pressure (a risk factor in the development of CHD), anxiety, depression, job and life satisfaction, in a large heterogeneous sample of over 3000 people. It examines the predictability of the model for men and women, taking occupational classification (manual and non-manual workers) into account, to allow some control over social class confounding. It also examines nonlinear relationships in the data.
The study showed that demands and discretion do separately predict psychological strain, job and life satisfaction, though they account for a very small percentage of the variance. There was no evidence of interactive effects which are central to the model. Where the model was able to predict blood pressure, relationships were consistently in the opposite direction to that hypothesized. There was also little evidence of curvilinear relationships. Interpersonal support was found to add significantly to the predictability of the dependent variables. The findings cast doubt on Karasek's model and suggest that emphasis should be shifted to those models which are more encompassing of support factors
Original languageEnglish
Pages (from-to)319-30
JournalJournal of Organizational Behavior
Issue number4
Publication statusPublished - Jul 1993


Dive into the research topics of 'A refutation of Karasek's demand – discretion model of occupational stress with a range of dependent measures.'. Together they form a unique fingerprint.

Cite this