Factor Analysis (R-Technique) of Ipsatized Data May Be Misleading

Abstract

Despite cautions in the literature against factor analyzing (R-technique) non-independent or ipsative data, several such studies have reported apparently meaningful results. This paper examines the results of factor analysis for six different sets of contrived data with known properties. Comparisons between the known factor structure and obtained results suggest that the number of factors initially extracted will be significantly under- or overestimated and that less than 60% of factored items will be correctly loaded. These findings provide empirical support for rational notes of caution in the use of factor analytic procedures as applied to ipsative data.