Replicating the essentially unidimensional model of the MACE work-to-family enrichment scale : going beyond goodness-of-fit indices

Abstract

ORIENTATION : Overreliance on goodness-of-fit (GoF) indices in confirmatory factor analysis (CFA) model fit evaluations appears to negatively influence the integrity and replicability of research findings in general, and on research to develop work-to-family enrichment (WFE) theory in particular. RESEARCH PURPOSE : The purpose of this study was to test for the conceptual replicability of the essentially unidimensional CFA model of the MACE Work-to-Family Enrichment scale (MACE-W2FE) using Bayesian structural equation modelling. MOTIVATION FOR THE STUDY : Multidimensional and second-order factor models are commonly reported for WFE instruments, but the more tenable essentially unidimensional model has remained largely untested, because of the limitations of GoF indices. RESEARCH APPROACH/DESIGN AND METHOD : Two independent cross-sectional study samples of 627 and 346 employees from various industry sectors was used. Bayesian structural equation modelling (BSEM) was applied to assess whether model misspecifications at local indicator level were substantive in terms of classical test theory, and justified the rejection of an essentially unidimensional CFA model (the breadth factor) for different MACE-W2FE versions. MAIN FINDINGS : In this study it was found that the essentially unidimensional model of the MACE-W2FE conceptually replicated across different studies, samples, MACE-W2FE versions and statistical theorems. PRACTICAL/MANAGERIAL IMPLICATIONS : The MACE-W2FE can be univocally scored as a single breadth factor for use in future research. CONTRIBUTION/VALUE-ADD : This study demonstrated the value of local indicator misspecification analysis using BSEM in countering deficient model testing in WFE studies.