A Nomological Approach to Structuring Personality Characteristics and Their Measures: Grouping Big Five Facets by Similarities in Their Nomological Networks

While trait measures are often structured in an intercorrelation grouping approach (IGA) based on their covariations, we propose a nomological grouping approach (NGA) that bases structure on nomological similarities. The NGA is demonstrated with 15 BFI-2 facets and 96 different nomologicals, using LISS panel data. We subjected four facets × nomologicals matrices to factor analyses: unpartialized (UP), hetero-domain partialized (HDP: variance controlled from facets of other domains), auto-domain partialized (ADP: variance controlled from facets of the same domain), and fully partialized (FP: variance controlled from all other facets). The UP matrix yielded a well-interpretable 3-factor solution and the HDP matrix a 5-factor solution (essentially the Big Five), while the ADP and FP matrices did not yield sensible factor structures. Using Procrustes rotations, the UP solution, but not the ADP and FP solutions, could be coerced into a Big Five solution. Some, but not all, NGA factor solutions were similar to IGA factor solutions. Taken together, facets of the same domain cohered based on highly similar nomological networks, but removing shared variance in BFI-2 facets dissolved their nomological similarities. We discuss the NGA as a framework for establishing taxonomies or conducting psychometric analyses.