Studying the process of change becomes tricky from a methodological point of view if change has its origin in both the intrapersonal and societal spheres. This paper examines the problems which arise when discussions of change measure only changing means without considering the fact that constructs of change are continually redefined. For example, as a person ages, the contention that he or she becomes more conservative is as plausible as the assumption that in rich societies economic crises lead to a reorientation toward conservative values. To disentangle intrapersonal and societal origins of change, at least two multi-cohort panels have to be started at different points in time. Such longitudinal studies, however, also involve pitfalls when measuring change. Both the means of a value preference change over time as does the understanding of the value itself. Means of change are often artifacts, due to the fact that content-wise redefinitions of constructs are overlooked. In some cases, the latter are responsible for mean differences that would not have reached statistical significance had qualitative changes been treated properly in analytic strategies. In other words, researchers must consider that mean differences may be due to qualitative changes in meaning that the examined construct underwent. (Contains five tables which present statistical summaries.) (RJM)