Using data from the National Survey of Families and Household, attitudes and behaviors measured at the first interview are used to predict the disruption of unions between 1987-88 and 1992-94. The analysis is structured in terms of three conceptual distinctions. An index of "conservative family attitudes" is constructed from items relating to approval of premarital sex, childbearing, cohabitation, independence in marriage, and divorce. Religion is considered in terms of denomination, frequency of attendance and fundamentalism. Couples' relationships are characterized in terms of behaviors likely to affect union stability, and in terms of evaluations of their relationships. These measures are then entered sequentially into multivariate hazard models predicting separation before the second interview.
Strong effects on union disruption are found for each of the variables examined, net of background measures, and net of other variables within the same conceptual set. The effects of patterns of interaction are largely captured by the evaluation measures. Two sets of the union evaluations are examined separately. The first set includes a measure of relationship happiness, perceptions of whether life would be better or worse were separation to occur, and whether the relationship was seen as unfair to the respondent. The second set is composed of two variables that are more directly a part of the disruption process: whether a relationship was seen as "in trouble," and perceptions of the likelihood of separation. Conservative attitudes toward family issues have an indirect effect through relationship quality, as well as a strong direct effect on disruption net of the other variables. Effects of religiosity are substantially mediated by the value measures, but also maintain independent effects. These results are consistent with expectations that both secularization and value change involving an increasing tolerance of new family forms are likely important in the transformation of family life, albeit reciprocally related to new family behaviors.