Marital happiness over the marital career: A specification and test of the developmental model

This study examines the direction of change in marital happiness over the course of the marital relationship. Previous research has led to the acceptance of a U-shaped, curvilinear pattern of marital quality over time in spite of ambiguous findings and the problematic use of data that are primarily cross-sectional and often based on nonprobability samples. Using data from a large, representative four-wave panel sample, this study presents a direct comparison of cross-sectional and longitudinal analyses of trends in marital happiness. In the first stage of analysis, data from the first wave of the panel study are analyzed. In this cross-sectional analysis, a U-shaped curvilinear relationship between marital happiness and marital duration is replicated, with marital happiness declining from the beginning of marriage through the first several years, then leveling off and increasing through the remaining years of marriage. The second stage of analysis utilizes a fixed effects pooled time-series model to examine the relationship between marital happiness and marital duration in the multiple-wave panel data. The longitudinal analysis reveals declining marital happiness for respondents at all marital durations, showing no support for an upturn in marital happiness in the later years of marriage. Instead, the curvilinear relationship between marital happiness and marital duration is indicated by a flat S-shaped pattern, with steeper declines in marital happiness during the earliest and latest years of marriage. When the effects of other life course variables are controlled, including the number of children in the household at various ages, the curvilinearity of the relationship is reduced, but a significant negative effect of marital duration on marital happiness remains. This study provides evidence that the U-shaped pattern of marital happiness is likely a product of cross-sectional research and is not typical of American marriages. Implications of this study emphasize the need for longitudinal data in studies of developmental change in family relationships