Many reviews of the literature and considerations of methodological problems in specific areas are published after considerable experimentation and at a time when interest in the field has lagged. This paper will attempt to anticipate problems of method in a newly developing research approach in order to contribute to maximal use of research energy in attacking the problems involved. The approach consists of administering attitude questionnaires to parents whose children differ in behavior. In studies to be considered it is indicated or implied that variation in the independent variable as measured by parental attitude measures has contributed in mediated or direct form to variation in the dependent variable or criterion measure of child behavior. In further discussion, the term criterion variable will be used to refer to measures of the behavior of children, whether such measures result from the use of experimental and control groups, or from measurements of a continuous variable in a single group. Studies to be discussed are postdictive rather than predictive. The study design thus parallels the retrospective clinical case study. It might be called a "reverse leap-frog" design. Inferences from obtained differences in parental attitudes involve a backward extrapolation in time, conceptually "leaping over" the actual behavior of parents with the children. The nature of an earlier developmental process is inferred from data obtained at a later point in time. Accordingly it is difficult to assert that obtained differences between parental attitudes measured after the development of children reflect factors operating at the time the children were developing. On the other hand, obtained differences in attitudes between parent groups cannot be ignored. It is tempting simply to suggest longitudinal studies of the effects of measured parental attitudes on child development. However, the productivity of such cumbersome, slow-moving projects may be heavily dependent on the preliminary hypothesis development carried out most rapidly in the retrospective type of study.