The role of cohabitation in remarriage: a replication

Abstract

Previous research has demonstrated that the effects of postdivorce cohabitation on the quality of intimate relationships are similar to those of premarital cohabitation. Thus, like premarital cohabitation in first marriage, postdivorce cohabitation in general and multi-partnered postdivorce cohabitation in particular delay remarriage and diminish relationship quality (i.e. undermine happiness and stability). However, such findings are based on survey data from the 1980s, such as the first wave of the National Survey of Families and Households. Using pooled samples from the National Survey of Family Growth (1995, 2002), the present study determines whether such findings can be replicated with more recent nationally representative data collected from remarried women between the ages of 15 and 44 (n=1915). Multivariate regression analyses reveal two major findings. First, as anticipated, premarital and postdivorce cohabitation, including multi-partnered forms of cohabitation, delay remarriage. Moreover, serial cohabitation also significantly delays remarriage. Second, remarriages preceded by either postdivorce cohabitation or serial cohabitation exhibit a higher likelihood of remarital disruption (i.e. dissolution or separation). Consequently, while premarital and postdivorce cohabitation have become common in the USA, postdivorce and serial cohabitation continue to exert disruptive effects on remarriages among young American women. Implications and directions for future research are discussed.

Keywords:

• cohabitation
• premarital cohabitation
• postdivorce cohabitation
• remarriage
• remarriage dissolution
• marital stability

Notes

1. Prior literature on religion, cohabitation and marriage suggests that religious persons are less likely to cohabit (Thornton et al. 1992, 2007), and that they marry as well as remarry sooner than their nonreligious counterparts (Xu et al. 2005, Brown et al. 2009). For these reasons, we included respondents’ denominational affiliation, worship service attendance, and religious salience (i.e. importance of religion) in our initial analyses. However, because of the lack of statistical significance, they were removed from our final analyses. The removal of these variables did not significantly modify our results.