'Shared job strain': A new approach for assessing the validity of job stress measurements

While problems of self-report measures of work stress have long been recognized, those of more 'objective' measures are often underestimated. Combining both in structural equation models yields more valid estimates, yet correlations with indicators of well-being or strain rarely exceed .30. To decide whether this is due to insufficient validity of instruments or to the multi-causal aetiology of well-being, the concept of 'shared job strain' is introduced. This is a latent variable, with individual symptoms of strain of two workers holding the same job as indicators. Thus, it represents the strain that these two workers have in common, while truly individual variance is removed. It should, therefore, show much higher correlations with job stressors than do individual symptoms of strain. To estimate stressors, self-reports of the two workers and of two independent observers are used as indicators. Four stressors explained two-thirds of the variance in 'shared job strain'. It is concluded (a) that estimating latent job stressors on the basis of self-report and observer indicators yields highly valid measurement and (b) that the substantive argument is supported. There probably is an upper limit of 15 to 20 per cent variance in total strain symptoms that can be explained by job stressors.