Objectives In order to make valid conclusions about individual change in work-related risk factors it is important to examine whether these factors are measurement invariant over time. We tested the measurement invariance of the effort-reward imbalance (ERI) scales using the ERI Questionnaire (ERI-Q). Additionally, we examined the criterion validity of the ERI scales.

Methods The sample used in this study was population-based and comprised 2128 participants (56.6% women) in full-time employment. Data on effort, reward and self-reported general stress were collected in 2007 and 2012. Measurement invariance was assessed separately for the effort and reward scales, with reward treated as a first-order and as a second-order variable. Criterion validity of the ERI scales was also examined using a single-item measure of general stress.

Results Effort and reward were found to be measurement invariant over time, that is, they measured the same latent variable across both time points. Furthermore, ERI and its components showed adequate criterion validity, and effort was additionally found to prospectively predict general stress 5 years later (β=0.072, 95% CI 0.013 to 0.131).

Conclusions Our results indicate that changes in the scores of the ERI scales are more likely caused by changes in perceptions of work characteristics than by changes in the construct of the scales. Additionally, the results support the criterion validity of ERI and its components.