The state of the art of latent variables measurement: The case of self-perceived employability among senior executives in the French private sector

Abstract

This article summarizes the evolution of the different latent variable measurement models used in organizational scholarship. We start by revisiting the traditional models used since the 1980s, then present the most recent findings over the last decade, including the rediscovery of bifactor models and the development of exploratory structural equations modeling (ESEM). These last two models represent a generalization of the traditional confirmatory factor analysis model (CFA ICM), and demonstrate both a better fit to the data and a better construct validity (convergent, discriminant, predictive). These methods are then illustrated by operationalizing the “self-perceived employability” construct. The traditional CFA model is significantly less well suited to the data than the ESEM models, and displays much lower construct validity and reliability. The final result yields a bifactor ESEM model with an overarching general employability factor and ten domain-specific factors, all of which predict that a senior executive job-seeker is more likely to feel employable than a professionally active senior executive, except for the professional self-efficacy and competency factors. Some theoretical and empirical biases of the traditional CFA model are then highlighted, and managerial as well as academic openings are outlined for future studies.