Abstract
This paper compares moral values in 12 European countries based on data from three waves of the European Values Survey: 1981, 1990, and 1999. We question the validity of comparisons across cultures and in time if the structural equivalence in the measurement of values is not tested. A procedure is suggested to research equivalence in measurement within a latent-class factor approach. Multiple group analyses revealed that assuming a homogeneous measurement model is not justified. Rather patterns of partial homogeneity emerged in which country and period influenced moral issues independently from the effect of two group variables on the latent-class factors. As a result, country differences generally decreased when direct effects of country on the items were introduced, however without contradicting the comparisons from the homogeneous measurement model. Period differences decreased less after allowing direct effects of the former group variable on the items.