A cocktail fallacy in research with composite variables: a commentary on job strain and effort–reward imbalance (ERI) theories with updated meta-analyses on health outcomes

Job strain and effort–reward imbalance are two widely used theories in occupational stress research. In this paper we show that a large body of research based on these theories needs to be reevaluated. We take a non-technical approach to show that the use of single parameter composite variable models to test the job strain and effort–reward imbalance theories is insufficient. Further, we describe how to use standard two or three parameter models to investigate associations proposed by these theories. We present a method for recalculating estimates from a particular composite variable model so they represent proper estimates from a three parameter interaction model, and apply that method on estimates from five large studies on health outcomes. We also present an updated meta-analysis on job strain and coronary heart disease (CHD) with over 400k subjects and >6.8k CHD cases using such recalculated estimates. The results show no support for the job strain theory, indicating instead that low job control alone is associated with an increased risk of CHD. We conclude that further research on the job strain and effort-reward imbalance theories using single parameter composite variable models is untenable and that a constructive avenue for occupational stress research is to better understand the association between low job control and adverse health outcomes.