Abstract
Schwartz' theory of ten basic human values has stimulated numerous studies using a variety of instruments. Confirmatory factor analyses (CFA) of the properties of some of the instruments have revealed that three pairs of values were excessively highly correlated. This led Davidov et al. (2008) to propose unifying values. To overcome the problems of loss of precision due to unifying distinct values, Knoppen and Saris (2009a,b) investigated the factorial structure of each of the ten values measured with the PVQ (Schwartz et al. 2001). They identified both cross-loadings and distinct sub-dimensions for the pairs of nondiscriminated values in two German student samples. They concluded that the original strategy for selecting items, maximizing theoretical coverage at the expense of item homogeneity, produced the poor discrimination between values. Our Study 1 examines whether the Knoppen and Saris findings generalize to a representative sample of the German population. With some notable exceptions, our findings replicate theirs. Study 2 uses 33 items from an experimental version of the PVQ to operationalize and test a full model of the 11 basic values. Following Knoppen and Saris, we included only one sub-dimension of each of the 11 values. This CFA model yielded a satisfactory fit with no estimation problems. We conclude that available indicators permit measuring the distinct values without the need to collapse factors. Limitations and implications of the research are discussed.
Beierlein, Constanze