A test of the Job Demands-Resources model with alternative measures of strain and well-being

Since the beginning of the 21st century, the Job Demands-Resources model (Demerouti, Bakker, Nachreiner & Schaufeli, 2001) has given rise to numerous studies aimed at a better understanding of positive and negative aspects of wellbeing at work. The existence of a motivational and an energetic process leading to, respectively, engagement and strain has been demonstrated several times, as well as cross-links between the two.

However, these studies mainly rely on two measurement tools: the MBI-GS as a measure of strain and the UWES as a measure of engagement. The aim of our study is to test the model using alternative measures. Strain was measured using the NOSI subscale, and engagement using the POSI subscale of the PNOSI, a new tool measuring positive and negative wellbeing. Structural and external validity of this tool have been demonstrated previously (Barbier, Peters & Hansez, submitted). According to the JDR model, we hypothesise that high job demands would lead to strain and then to health problems. Our second hypothesis is that job resources would lead to engagement and then to low intent to leave.

Structural equation modelling was performed using Lisrel 8.80 on 954 data collected in a Belgian public institution. As regards to measurement model, results show that a six-factor model (demands, resources, strain, engagement, health problems and intent to quit or IQ) shows good fit to data. We first tested the traditional JDR model, but it showed bad fit. Fit was improved when adding a path from resources to strain, from strain to IQ and from engagement to health problems. This alternative model showed better fit than the initial one.

This study makes three main contributions. First, it tests the well-established JDR model using alternative measures of positive and negative sides of well-being. Second, it asks questions as to the respective outcomes of strain and engagement. More specifically, the path between engagement and IQ became not significant when a path from strain to IQ was added. That is, IQ seems to be more a product of strain than of (lack of) engagement. Finally, it emphasizes the important role of resources. It may be that the link between resources and low IQ would be mediated by low level of strain rather than by high level of engagement. Similarly, the relation between engagement and health problem is positive, which raises questions as to the pre-supposed positive effects of engagement.