I would like to to test Karasek's model of job strain in my current research on burnout in healthcare professionals. I have given this a lot of thought and have come up with several ways to do this and was wondering if anyone had any ideas.

For anyone unfamiliar with the model, job strain = job demands/(job control+skill utilisation). I have computed the scores for the sub-scales (job demands, job control, and skill utilisation) from questionnaire items, as per the original paper.

Initially I conducted regression analysis with job strain as the dependent variable, and the sub-scales as the independent variables. This didn't seem right, as job strain is bound to be associated with the sub-scales, seeing as it is a product of them. Would this be the incorrect way to test the model?

I then decided that in order to test the model I would need to see whether job strain could predict burnout, as this is the focus of my research. I have also measured job satisfaction and collegial/managerial support, both of which Karasek stated are moderators of psychological distress, and will include these variables in a stepwise regression analysis with job strain, with burnout as the dependent variable.

I am posting on this site to see whether anyone can pick holes in my plan, or suggest an alternative way in which I might test the model. I would also like confirmation that my initial attempt to test the model was seriously flawed!

Many thanks.