Multivariate Analysis Of

ABSTRACT The term "ipsative" refers to measurement based on intra-individual comparisons. The research literature in the social sciences contains many cautions about using ipsative data in multivariate analysis. The purpose of this paper is to identify the problems associated with the multivariate and regression analyses of ipsative data and to provide recommendations for avoiding these problems. Ipsative scales have been mainly employed by researchers in the area of personality measurement, vocational choice, and the assessment of values and attitudes where the scale values are interrelated. Due to this interrelatedness of the scales, many problems can arise when multivariate and/or regression analysis procedures are used to analyze the data obtained from ipsative measurement. Such an analysis would produce a singular correlation or covariance matrix that cannot be inverted. The results of the analysis can be inflated or deflated relationships that are artifacts of the forced distributions. While counselors, clinical psychologists, and other clinicans may have valid uses for ipsative measures, these same measures may not be appropriate for research purposes where multivariate and regression analyses are used. This study recommends that multivariate or regression analysis should not be performed on data obtained with ipsative measurement. Other alternatives are to use normative rather than ipsative measurement, or to reduce the interrelatedness of the scales by deleting one of the ipsative scales or adding a normative scale. (JAZ)