State-Trace Analysis Meets Personality Measurement: Identifying and Fixing Hidden Inconsistencies in the Big Five Questionnaires

Factor analysis falls short in addressing a pivotal question within personality measurement: the determination of whether a set of items can be logically reduced to a single latent factor. This study advocates for the application of state-trace analysis, an underutilized method from mathematical psychology, to decisively address this question. State-trace analysis introduces a simple but rigorous criterion for unidimensionality: monotonicity between item pairs. Identification of items violating this criterion within a factor is straightforward. This paper illustrates exemplary analyses within the framework of the five-factor model, focusing on the International Personality Item Pool-NEO-120 ($N=618,000$) and the NEO Personality Inventory--Revised (N₁=857, N₂=500) questionnaires. The findings demonstrate that sustaining a five-factor model necessitates alterations to many items. This underscores the potency of state-trace analysis in advancing personality measurement beyond current methodologies. The paper concludes by discussing strategies to promote broader adoption of this method and how future designs in personality research can be tailored to effectively incorporate state-trace analysis.