An Empirical Comparison of Factor, Image, Component, and Scale Scores

Abstract

The relationships among principal component, image component, three types of factor scores, and a scale score method were examined. The scores were compared over several levels of variables (p), component saturations (aij), sample sizes (N), variable to component ratios (p/m), and factor pattern rotations. Scores were compared on the same (convergent) components, and also on different (divergent) components. These comparisons indicated that there were virtually no differences among score methods, although comparisons within score methods, i.e. between component scores and among factor scores, generally indicated slightly higher relationships than did comparisons between the respective score methods. Also, the scale score method, while correlating the lowest, was still very representative of the component and factor scores. When scores did depart from each other, it usually occurred in one of the conditions of low component saturation, low sample size, low p/m ratio, or a combination of these conditions.