Enoch O. Oladunmoye, PhD

How Many Factors Does a Psychological Scale Really Have?

Determining the number of latent factors underlying a set of questionnaire items is one of the most consequential decisions a researcher makes in exploratory factor analysis (EFA). An incorrect factor retention decision can produce an over factored, under factored, or conceptually incoherent measurement model, with downstream consequences for reliability, validity, scoring, and substantive interpretation. Despite its importance, many applied researchers continue to rely heavily on simple heuristics such as the eigenvalue greater than one rule or unaided visual inspection of scree plots. This Research Note reviews contemporary approaches to factor retention, including the Kaiser criterion, the scree test, parallel analysis, the minimum average partial correlation procedure, comparison data methods, the Hull method, and exploratory graph analysis. Particular attention is given to the principle that no single retention method is optimal under every data condition. Recent simulation and methodological literature indicates that the accuracy of factor retention methods varies with sample size, factor correlations, the number of indicators per factor, factor strength, missingness, and the underlying population structure (Auerswald & Moshagen, 2019; Goretzko, 2025). The Note proposes a practical, multi evidence decision framework for applied researchers and describes an implementation model for the PsychtrixWeb platform in which factor retention is treated as a converging body of psychometric evidence rather than a single automatic statistical command. The central recommendation is that researchers should combine empirical retention criteria with theoretical interpretability, item level evidence, and independent replication before finalising the dimensional structure of a scale. Keywords: factor retention, exploratory factor analysis, parallel analysis, eigenvalues, scree plot, dimensionality, psychometrics, scale development, latent factors, PsychtrixWeb