Validity: Q vs. R

Abstract

This study deals with political ideology, in the form of Q sorts of 80
self-referent statements provided by more than 50 respondents, the P set
being cross-sectional. The resulting data set was split up into three
equivalent P samples, each of which was submitted independently to Q
factor analysis. Also, it was split up into four equivalent item
samples, each of which was submitted independently to R factor analysis.
Comparison of Q analysis loadings with R analysis z-scores and of Q
analysis z-scores with R analysis loadings shows a limited amount of
cross-validity. With respect to (meta-)reliability, the three cases of Q
factor analysis are strikingly similar; the fours cases of R factor
analysis, however, turn out to be rather dissimilar.