Problems in making inferences from a sample to a population and from one cohort to other cohorts are discussed. It is concluded that in most cases a longitudinal design using repeated measures is preferable to an independent samples design. (Author)

Presents a theoretical discussion of the role of cohort effects in the study of human development. The paper examines three problems associated with cohort analysis. These involve the interrelated problems of defining boundaries, defining distinctive experiences, and analyzing typical response patterns. (BD)

Evaluates empirical studies on child development. Suggests that most such research consists of studies of a priori, nonempirical, logical relations between concepts, whose definitions guarantee the relationship studied. Argues that hypotheses are empirical if variables involved are semantically and logically independent. Research that is not based…

Comments on the ideas espoused by Smedslund (PS 522 552) in this issue. Agrees to the idea of spending more intellectual energy in distinguishing between a priori and empirical hypotheses but emphasizes that concepts are not always accurate reflections of reality and that even empirical disconfirmation of an a priori hypothesis sometimes can…