NEW APPROACHES FOR MEASUREMENT SCALES ANALYSIS IN HRM: AN ILLUSTRATION IN THE FIELD OF PSYCHOSOCIAL RISK ANALYSIS

Abstract

Being able to measure accurately and reliably multidimensional latent constructs reflecting attitudes such as job satisfaction, work involvement, or well-being is both a managerial imperative and a methodological challenge for HRM. Recent research has shown that traditional validation procedures for measurement scales have limitations, mostly related to very restrictive assumptions about measurement models (e.g. not taking into account cross loadings in factorial models, or the inability to simultaneously take into account global and specific effects of multidimensional constructs). New methodological developments have open the possibility to overcome some of these restrictions and to propose more realistic models with better psychometric properties and stronger predictive power. This article aims to illustrate these developments, which are still mostly unacknowledged in the Frenchspeaking management literature. We present a mixed variable-centered and person-centered approach. As an illustration, we have selected a measurement model based on a French translation of the Health and Safety Executive Management Standards Indicator Tool (HSE MSIT), used in several countries to assess management standards and working conditions promoting the reduction of psychosocial risks (PSRs). The measurement model used in this article is an exploratory bi-factor model (Bi-ESEM) with six specific factors, tested on a sample of 1066 French workers. In terms of a variable-centered approach, this model accurately predicts the prevention of PSRs such as absenteeism, presenteeism and turnover intentions. In terms of a person-centered approach, it highlights four contrasted profiles of health and safety management policies perceptions (the ignored, the relaxed, the involved, and the high potentials). The benefits of this approach for the researcher and the manager are then discussed.