Factoring the Personal Profile System for Construct Validity: Three Analyses Under Different Standardization Assumptions

Three types of data were factor analyzed using principal components extractions with orthogonal and oblique rotations to test publisher claims for construct validity of the Personal Profile System (PPS). Behavioral descriptor data from 1,045 senior non-commissioned Air Force officers were factored as raw data, mean corrected data, and standardized z-scores (correlations). The most efficacious solution was produced with standardized z-scores generating four factors accounting for 86% of the total variance. The measure of sampling adequacy for every descriptor exceeded 0.922. The first factor was general with approximately equal loadings on each of the dominance, influencing, steadiness, and compliance dimensions. The second factor was biscalar, with dominant loadings on steadiness and compliance descriptors; the third was essentially uniscalar with generally weak loadings on influencing, its closest PPS dimension. All descriptors loaded on at least one factor at 0.30 or higher, accommodating a marginally acceptable theoretical degree of psychometric and measurement properties and indicating four-factor construct relevance. Results do not completely support previous PPS publisher claims for instrument dimensionality and scaling properties. Five tables present study data, and there is a list of 17 references.