Factors Affecting Divorce: A Study of the Terman Sample

Within the past few years, renewed interest in understanding marital behavior has resulted in a number of studies which focus on an equation estimating the probability of divorce or remarriage. This paper reports on one such effort. It offers a brief rationale for and an estimation of probability functions for divorce rates at specific lengths of marriage duration for a very unrepresentative sample of American women -- a group of geniuses. The data are from the "Terman sample" of some 671 women selected in 1921 (together with a comparable group of men) by psychologist Lewis N. Terman. The sample was chosen from children enrolled in California schools in urban areas. It included children, preselected by their teachers, whose measured IQ was 135 or above. The sample thus represented students in the highest one percent of the school population in general intelligence. In another report I have compared the marital behavior of these Terman subjects to the relevant California population, controlling for the very high level of schooling and the somewhat constricted distribution of age at first marriage among the Terman subjects (Michael 1976). The Terman subjects generally exhibited the same qualitative relationships between marital patterns and such variables as age at marriage and schooling as the California population. However, one should keep in mind the very special nature of this sample when comparing results with other studies.