Premarital Cohabitation and Marital Dissolution: An Examination of Recent Marriages

Abstract

An ongoing question remains for family researchers: Why does a positive association between cohabitation and marital dissolution exist when one of the primary reasons to cohabit is to test relationship compatibility? Drawing on recently collected data from the 2006 – 2008 National Survey of Family Growth, the authors examined whether premarital cohabitation experiences were associated with marital instability among a recent contemporary (married since 1996) marriage cohort of men (N = 1,483) and women (N = 2,003). They found that a dichotomous indicator of premarital cohabitation was in fact not associated with marital instability among women and men. Furthermore, among cohabitors, marital commitment prior to cohabitation (engagement or definite plans for marriage) was tied to lower hazards of marital instability among women, but not men. This research contributes to our understanding of cohabitation, marital instability, and broader family change.

The increase in cohabitation is well documented, such that nearly two thirds of newlyweds have cohabited prior to their first marriage (Kennedy & Bumpass, 2008; National Center for Family & Marriage Research, 2010b). Cohabitation allows young adults to test their relationship (Bumpass, Sweet, & Cherlin, 1991; Manning & Smock, 2009), which should help determine whether they are compatible before they marry; many researchers, however, have found a positive association between cohabitation and marital dissolution (e.g., Jose, O’Leary, & Moyer, 2010; Kamp Dush, Cohan, & Amato, 2004; Stanley, Rhoades, & Markman, 2006). The bulk of the work documenting a positive influence of cohabitation on marital instability rests on data collected from women over 10 years ago (National Survey of Families and Households, 1987 – 1988, see http://www.ssc.wisc.edu/nsfh/; National Survey of Family Growth [NSFG], 1995, 2002, see http://www.cdc.gov/nchs/nsfg.htm). Recent research suggests that as cohabitation becomes more common, its effect on marital instability may weaken for more recent marriage cohorts (Hewitt & De Vaus, 2009; Reinhold, 2010).

We drew on recently collected data from the NSFG (2006 – 2008; see http://www.cdc.gov/nchs/nsfg.htm) to examine the relationship between cohabitation and marital dissolution among men and women. We first assessed whether cohabitation was associated with marital instability among recently married men and women. Our approach is consistent with the diffusion perspective, which argues that a weaker cohabitation effect exists among recent marriage cohorts with higher rates of premarital cohabitation. We also considered how commitment to marriage at the outset of cohabitation is tied to the relationship between cohabitation and marital instability. Given that most prior studies in the United States have relied on national data on women’s marriages formed over 10 years ago, our work provides an important update. The findings from this article will help move forward our understanding of marital stability, cohabitation, and family change.

Background

The increase in cohabitation is well documented, with increasing percentages of young adults experiencing cohabitation. Furthermore, cohabitation has become the modal path to marriage, such that 44% of women cohabited prior to their first marriage in the late 1980s (1985 – 1989), and 67% have done so since 2000 (Bumpass & Lu, 2000; National Center for Family & Marriage Research, 2010b). At the same time, there has been a plateau in the divorce rate, with about one half of first marriages ending in separation or divorce (Raley & Bumpass, 2003; Stevenson & Wolfers, 2007). Thus, at the aggregate level, the rise in cohabitation is not associated with a similar growth in divorce (Goldstein, 1999).

Theoretically, cohabitation has been framed as comprising relationships in which marriages least likely to succeed were weeded out (Bumpass & Sweet, 1989). Both quantitative and qualitative data indicate that young adults view cohabitation as a way to test the relationship, and some hope to “divorce-proof” their marriages by spending time cohabiting with their future spouse (Manning & Smock, 2009). In 2008, nearly 70% of high school seniors reported that living together before marriage is a good way to test compatibility, and cohabitation rates have been increasing steadily over the last 30 years (National Center for Family & Marriage Research, 2010a). Determining whether one’s relationship is suited for marriage may be not the primary motivation for initially cohabiting but instead a latent motivation that develops during the course of the relationship (Sassler, 2004). Substantive findings indicating that cohabitation is perceived as a means of filtering out poor matches are in line with theoretical marital search models (Becker, 1981) that suggest that cohabitation provides the best way to garner information about prospective spouses (England & Farkas, 1986). Furthermore, Oppenheimer (1988) argued that premarital socialization helped improve marriage matches and that cohabitation “facilitates the kind of interaction that increases the knowledge of oneself and of a potential marriage partner and of the kind of mutual adaptations that are so essential to a stable relationship” (p. 583). It is ironic that most empirical studies find that couples who cohabited prior to marriage experience significantly higher odds of marital dissolution than their counterparts who did not cohabit before marriage (Jose et al., 2010).

The question addressing the underlying mechanisms that explain why cohabitation influences marital stability is not new. Newcomb and Bentler (1980b) studied 159 couples in Los Angeles and concluded that there were two (not mutually exclusive) approaches to help understand the cohabitation effect: (a) selection and (b) cohabitation experience. Thirty years later, virtually every study on cohabitation and marital stability has drawn on these same two overarching explanations. The selection approach argues that the same characteristics that predict cohabitation are associated with marital dissolution. Many studies have found that selection explains some of the effect of cohabitation on marital instability (e.g., DeMaris & MacDonald, 1993; Kamp Dush et al., 2004; Phillips & Sweeney, 2005). Research has supported the selection argument using statistical techniques (Lillard, Brien, & Waite, 1995), a rich set of covariates (Woods & Emery, 2002), or type of marriage indicator (covenant vs. standard; Brown, Sanchez, Nock, & Wright, 2006). An outgrowth of the selection argument joins commitment theory with the concept of inertia and argues that once couples start to cohabit they end up on a fast track toward marriage without sharing high initial commitment levels (Stanley et al., 2006). The second and related explanation for the cohabitation effect is that the cohabitation experience itself is tied to a waning commitment to marriage (e.g., Axinn & Barber, 1997; Clarkberg, Stolzenberg, & Waite, 1995; DeMaris & MacDonald, 1993; Kamp Dush et al., 2004; Stanley et al., 2006). Young adults report more supportive attitudes toward divorce and cohabitation after cohabiting (Axinn & Barber, 1997; Cunningham & Thornton, 2005). The consensus in the literature is that both general mechanisms have been operating.

An integration of the diffusion and selection arguments has emerged as selection processes into cohabitation have weakened with the growing prevalence of premarital cohabitation (Kamp Dush et al., 2004; Leifbroer & Dourleijn, 2006). The diffusion approach proposed by Leifbroer and Dourleijn (2006) states that as cohabitation has become more widespread, its effect on marital instability has declined. The pattern is U-shaped, such that European countries where cohabitation is more rare (e.g., Belgium) or very common (e.g., Finland) there was a negative effects of cohabitation on marital stability (Leifbroer & Dourleijn, 2006). Similarly, in Australia, premarital cohabitation appeared to reduce the risk of marital dissolution among recent marriages (De Vaus, Qu, & Weston, 2005; Hewitt & De Vaus, 2009).

Studies conducted in the United States have proposed that the increase in cohabitation and a growing acceptance of cohabitation indicate that couples who cohabit may be becoming less selective and comprise a more typical family pattern than in the past (Kamp Dush et al., 2004; Reinhold, 2010; Schoen, 1992). Reinhold (2010) relied on data from the 2002 NSFG and reported that recent marriage cohorts of women (married after 1993) did not experience a cohabitation effect. Unfortunately, in the 2002 NSFG a substantial and select subgroup of women (married women with stepchildren) were not asked about marital dissolution because of a routing error in the interview, so findings based on the 2002 data must be considered with some caution (National Center for Health Statistics, 2004; Reinhold, 2010). Other U.S.–based work has not found a trend in the cohabitation effect (Kamp Dush et al., 2004; Teachman 2002, 2003); however, these studies were based on data that reflected marriages formed prior to the mid-1990s (1981 – 1997, 1950 – 1994, and 1970 – 1995, respectively). Marriage cohort may be a proxy for changes in the institutionalized support for marriage as well as the spread of cohabitation. Thus, we expected that because cohabitation has become the majority experience prior to marriage, its effect may become weaker among more recently married women and men.

Because cohabitation has become increasingly common, attention must be paid to the heterogeneity among cohabitors. Commitment to marriage has been one way to distinguish cohabitors (Brown & Booth, 1996; Casper & Sayer, 2000; Guzzo, 2009; Kline et al., 2004; Manning & Smock, 2002), because it indicates who views their relationship as a clear step toward marriage. Guzzo (2009) reported that about two-fifths of cohabitors were engaged or had definite plans to marry their partner when they started cohabiting. Cohabitors with marriage plans experienced levels of marital quality and distress similar to those of married respondents who had not cohabited (Brown, 2004; Brown & Booth, 1996; Kline et al., 2004; Rhoades, Stanley, & Markman, 2009a; Stanley, Rhoades, Amato, Markman, & Johnson, 2010). Cohabiting couples without marriage plans experienced lower marital quality and higher marital distress (Kline et al., 2004; Rhoades et al., 2009a).

Only a limited body of research has specifically studied cohabitors’ marriage plans and marital dissolution, however. One recent study by Stanley and colleagues (2010) drew on data from women and men married in the 1990s from four states (Arkansas, Kansas, Oklahoma, and Texas) and measured initial couple commitment with an item asking whether the respondent was engaged at the start of cohabitation. They reported that cohabitors without marriage plans experienced a higher odds of marital dissolution than cohabitors with marriage plans (Stanley et al., 2010). Using national data, in the present study we evaluated whether the association between cohabitation and marital stability differs according to marriage plans. On the basis of prior work, we expected that married couples who cohabited with marriage plans would share marital stability similar to that of married women and men who had never cohabited and greater marital stability than men and women who cohabited without plans for marriage.

We used recently collected data to assess the influence of cohabitation on marital stability. The bulk of national U.S.–based research that has reported a positive effect of cohabitation on marital instability was based on data collected some time ago (NSFG 1995, 2002, NSFH 1987 <en> 1988) and may not reflect contemporary experiences. We assessed whether recently married (since 1996) men and women who cohabited with their spouse prior to marriage experience greater marital instability than their counterparts who did not cohabit. We capitalized on a key differentiation in cohabitation—commitment to marriage at the outset of cohabitation (engagement or definite plans for marriage)—to further evaluate the link between cohabitation and marital instability. We examined whether men and women who did not cohabit with their spouse prior to marriage experience greater marital stability than cohabitors who had marriage plans and cohabitors who did not have marriage plans.

Although the goal of this study was to specifically examine how cohabitation influences marital stability, we included key sociodemographic indicators available from men and women to assess whether these potential selection factors account for relationship between cohabitation and marital instability. Relying on marriages that occurred prior to 1995, Phillips and Sweeney (2005) found that cohabitation had a significant positive effect on marital instability among Whites but had no effect among Blacks and Mexican Americans. Women who have cohabited prior to marriage have greater numbers of sexual partners (outside of cohabitation) than women without premarital cohabitation experience (Cohen & Manning, 2010). Teachman (2003) reported that, among women who lived with their husband prior to marriage, their sexual history, not their cohabitation history, predicted marital instability. Women who married at younger ages faced higher marital dissolution rates (Teachman, 2002), and we have observed continual increases in the age at first marriage for both men and women (U.S. Census Bureau, 2009). Earlier work has suggested that women who had children prior to marriage face greater odds of marital instability (Graefe & Lichter, 2002; Teachman, 2002) and that cohabiting women are more likely to have children than single women (Manning, 2001; Reinhold, 2010). On a related note, fertility prior to marriage explains some of the effect of cohabitation on marital quality (Tach & Halpern-Meekin, 2009). Women from disadvantaged backgrounds and lower socioeconomic status experience greater cohabitation and marital instability (Kennedy & Bumpass, 2008; Raley & Bumpass, 2003; Teachman, 2002). Consistent with the selection perspective, these sociodemographic factors may partially explain the link between cohabitation and marital instability.

Method

We draw on data from the 2006 – 2008 NSFG, which contains a national probability sample of 7,356 women and 6,139 men ages 15 through 44. These data are ideal because the samples include both men and women as well as recently collected cohabitation histories and marriage histories. Our analytic sample is based on 2,003 women and 1,483 men who were ages 15 through 44 in 2006 – 2008 and who had ever been married since 1996. We based our analyses of marital dissolution on first marriages because sample size limitations and dissolution processes differ among second and subsequent marriages.

The core dependent variable was the timing of the divorce or separation of the first marriage. In our sample, approximately one-fifth of women and men had experienced the dissolution of their first marriage by the interview date. We measured the time from marriage to dissolution or interview in terms of months.

The key independent variable was cohabitation experience. We found that approximately 63% of women and men had cohabited prior to their first marriage, and most cohabitors lived with their spouse (96% of women and 93% of men). Thus, we limited out analyses to respondents who had lived with their spouse prior to marriage. We created a dichotomous indicator of cohabitation with spouse. A trichotomous indicator of cohabitation distinguished (a) men and women with no premarital cohabitation with spouse, (b) cohabitation and engaged or plans for marriage with spouse, and (c) cohabitation with no plans for marriage. The NSFG question was worded, “At the time you began living together, were you and [partner] engaged to be married or did you have definite plans to get married?” Intentions for marriage were not asked of respondents who did not cohabit with their spouse.

The focus of this article is on the association between premarital cohabitation and marital instability for a recent marriage cohort. Given the upper age limit of 44 in the NSFG, we limited our analysis to women and men who were married since 1996 (within 10 years of the interview). Thus, our analyses are by definition limited to men and women who first married before age 34 (if interviewed in 2006). We conducted supplemental analyses of respondents married in the earlier marriage cohort (1986 <en> 1995) who, by definition, married at younger ages than our contemporary marriage cohort, and our findings are consistent with those of prior studies.

The NSFG data allowed us to include several key characteristics that have been included in prior work on cohabitation and marital instability (Phillips & Sweeney, 2005; Stanley et al., 2010; Teachman, 2003). The distribution of the covariates is presented in Table 1. We include race and ethnic indicators of White, Black, Hispanic native born, and Hispanic foreign born. The number of noncohabiting sex partners was available only for women and was calculated by subtracting the number of premarital cohabiting partners from the reported number of premarital sexual partners. We relied on a continuous indicator of the number of noncohabiting sex partners and noted that most women (75%) had had noncohabiting premarital sexual partners. Premarital fertility was measured using an indicator of whether the respondent had a child before his or her first marriage; approximately one quarter (27% of women and 24% of men) had a child before they were married. Age at marriage is a continuous variable; the mean age for women was 24.9 for women and that for men was 27.2. The NSFG measure of education was based not on education prior to marriage but on educational attainment at the time of interview. We recognize that this is flawed, but the majority of women and men have completed their education by the time they marry (Martin, 2006). The distributions of women and men in regard to education were quite similar; for example, approximately 39% of women and 36% of men graduated from college. Mother’s education was divided into three categories: (a) less than high school, (b) high school, and (c) college and beyond. The modal mother’s education category for both women and men was a high school degree. We measured whether respondents had lived with both biological parents through age 14 and found that about two thirds (65%) of men and women had been raised by two biological or adoptive parents.

Table 1.

Distribution of Variables for Women (N = 2,003) and Men (N = 1,483) Married Since 1996

Variable	Women	Men
	% / M	% / M
Cohabitation with spouse
Yes	61.54	61.22
No	38.46	38.79
Cohabitation/engagement status
No premarital cohabitation	38.46	38.79
Cohabitation and engaged	34.77	33.40
Cohabitation and not engaged	26.77	27.82
Race/ethnicity
White	64.52	62.07
Black	10.85	10.06
Native-born Hispanic	7.54	8.17
Foreign-born Hispanic	9.70	9.51
Other	7.39	10.20
No. noncohabiting sex partners	4.03
Premarital birth
Yes	27.41	24.05
No	72.59	75.95
Age at marriage	24.91	27.19
Respondent’s education
Less than high school degree	14.19	14.47
High school degree	46.40	49.68
College degree	39.42	35.85
Mother’s education
Less than high school degree	21.81	22.86
High school degree	57.39	58.21
College degree	20.80	18.93
Childhood family structure until age 18
Two biological/adoptive parents	64.54	65.32
Not two biological/adoptive parents	35.46	34.68

Note: Results are weighted. Source: the 2006 – 2008 National Survey of Family Growth (see http://www.cdc.gov/nchs/nsfg.htm).

We used life tables and survival models to examine the timing of marital instability among women and men separately. We tested for gender interactions; the results are presented in Table A on the Journal of Marriage and Family web site (http://onlinelibrary.wiley.com/journal/10.1111/(ISSN)1741-3737). Unlike the gender-specific analyses, the combined models of men and women do not indicate a significant association between cohabitation and engagement and marital instability. We used basic bivariate contrasts and life tables to provide an initial portrait of the relationship between cohabitation and marital stability. Because of the sampling strategy, we applied techniques that accounted for design effects. We used Stata to estimate Cox event history models that account for complex design effects. The Cox event history techniques incorporate duration as part of the dependent variable. In Table A, we present zero-order models and those with the full array of covariates. We conducted further analyses to determine factors that may explain the effects of cohabitation on marital instability.

Results

Recently married women and men typically cohabited prior to marriage (see Table 1); approximately 61% of women and men married since 1996 cohabited with their spouse prior to marriage. Approximately one third of recently married women and men were cohabiting with a commitment to marriage at the start of cohabitation. About half of men and women who cohabited with their spouse prior to marriage had a commitment to marriage at the outset of cohabitation.

Life table models indicated that premarital cohabitation was not linked to marital stability for women or men (see online Figures 1 and 2; http://onlinelibrary.wiley.com/journal/10.1111/(ISSN)1741-3737). In Table 2 we present the zero-order and multivariate hazard models that predicted marital dissolution. The dichotomous indicator showcased that premarital cohabitation was not significantly tied to marital instability at the zero order for women and men. In the full model, the results were similar (results not shown). For example, among women, the cohabitation coefficient was .83 (p = .265). Furthermore, the data in online Table A show that the interaction term of gender and premarital cohabitation and marital instability was not statistically significant. Thus, cohabitation appears to have a similar nonsignificant relationship to marital instability for men and women.

Table 2.

Cox Models Predicting Hazards for First Marriage Dissolution for Women (N = 2,003) and Men (N = 1,483) Married Since 1996

	Women				Men
	Zero Order		Full		Zero Order		Full
Variable	Hazard	SE	Hazard	SE	Hazard	SE	Hazard	SE
Cohabitation with spouse (reference [ref.]: No)
Yes	1.03	0.13			1.22	0.24
Cohabitation/engagement status (ref: No premarital cohabitation)
Cohabitation and engaged	0.75†	0.13	0.57*	0.13	1.22	0.30	0.99	.19
Cohabitation and not engaged	1.40*	0.23	1.07	0.23	1.22	0.27	0.98	0.23
Race/ethnicity (ref: White)
Black	1.70*	0.35	1.26	0.30	1.06	0.27	0.83	0.21
Native-born Hispanic	0.96	0.22	0.54†	0.17	2.35**	0.60	1.84*	0.43
Foreign-born Hispanic	0.86	0.17	0.53*	0.14	0.85	0.25	0.79	0.24
Other	1.07	0.40	0.98	0.35	0.75	0.32	0.86	0.34
No. noncohabiting sex partners	1.04**	0.01	1.03**	0.01
Premarital birth (ref: No)
Yes	2.03***	0.31	1.45†	0.30	2.24***	0.43	1.84**	0.40
Age at first marriage	0.93**	0.02	0.94**	0.02	1.00	0.03	1.00	0.03
Respondent’s education (ref: High school degree)
Less than high school degree	0.90	0.21	0.81	0.17	0.53**	0.12	0.51**	0.11
College degree	0.25***	0.05	0.33***	0.07	0.46*	0.14	0.59*	0.17
Mother’s education (ref: High school degree)
Less than high school degree	1.32	0.25	1.26	0.25	0.98	0.21	1.21	0.22
College degree	0.69†	0.15	0.84	0.19	0.88	0.27	1.12	0.34
Childhood family structure age 14 (ref: not two biological/adoptive parents)
Two biological/adoptive parents	0.51***	0.07	0.74*	0.11	0.61*	0.12	0.69†	0.13

Note: Source is the 2006 – 2008 National Survey of Family Growth (see http://www.cdc.gov/nchs/nsfg.htm).

^†p < .10.

^*p < .05.

^**p < .01.

^***p < .001.

The variable distinguishing plans for marriage indicated that premarital cohabitation was positively associated with marital instability when women were not engaged to their spouse at the outset of cohabitation. Women who were engaged and cohabited had marginally lower hazards of marital dissolution than women who did not cohabit. In addition, women who were engaged had significantly lower hazards of marital dissolution than women who were not engaged (results not shown).

The next model included the remaining covariates and indicated that women who premaritally cohabited with their spouse and were not engaged shared similar odds of marital instability as women who never cohabited. Women who were engaged and cohabited continued to experience lower hazards of marital instability. Further investigation indicated that the negative effect of cohabiting without engagement appeared to be explained by risk factors of marital instability (premarital fertility, family structure, educational attainment, or number of premarital sex partners). Greater proportions of women who had a premarital birth, were from single or stepparent families, and/or had lower levels of education and greater numbers of sex partners experienced cohabitation with their spouse prior to marriage. Additional analyses indicated that the positive association between cohabitation with commitment and marital stability existed only among select subgroups of women who faced greater risks of dissolution (i.e., women who were Black, had a premarital birth, had less than a college degree, were raised in single or stepparent families, or had more than the median number of sex partners; see online Table B). Finally, women who cohabited and were engaged still had significantly lower hazards of marital dissolution than women who cohabited and were not engaged (results not shown).

Many of the remaining covariates were related to marital dissolution. We found that Blacks had higher hazards of marital dissolution than Whites, and in the full model Hispanics had lower hazards of marital instability. The number of noncohabiting sex partners was positively associated with the hazard of marital instability. Premarital fertility was tied to higher levels of marital instability at the zero order, but not in the full model. Women who married at later ages had lower hazards of marital instability. Women with some post–high school education had a lower odds that their first marriage would end in dissolution. As expected, women from intact families experienced lower hazards of dissolution.

The zero-order models in Table 2 for men show that cohabitation prior to marriage was not associated with marital dissolution, regardless of engagement status. In the adjacent column, we present the multivariate results. Cohabitation appeared to not be significantly associated with marital instability in the full model. Men who cohabited and were engaged shared similar hazard rates as men who cohabited and were not engaged (results not shown). In further analyses, we tested how the cohabitation and engagement effect differed for men and women (see online Table A). As expected, cohabiting and being engaged was more protective against marital instability among women than among men. Women and men who cohabited and were not engaged shared similar hazards of marital instability. The remaining covariates indicated that native-born Hispanic men had higher marital dissolution rates than Whites. Premarital fertility was tied to greater marital instability. Higher levels of men’s education were associated with lower marital instability, and men raised in a two-biological parent family experienced lower rates of marital dissolution.

Discussion

Cohabitation has become an integral part of the marriage process. Young Americans believe that cohabitation helps one select good spouses who will ensure stable marriages, and concerns about divorce motivate some young adults to cohabit (Manning & Smock, 2009). Newcomb and Bentler (1980a) concluded that “It seems clear from the data that the impact of premarital cohabitation on a subsequent marriage is not a simple or direct relationship, but rather is multifaceted” (p. 23). Thirty years later, we draw similar conclusions.

Our work shows that, since the mid-1990s, whether men or women cohabited with their spouse prior to marriage is not related to marital stability. Consistent with a diffusion perspective, our results suggest a reduced effect of cohabitation on marital instability, as cohabitation prior to marriage becomes widespread. Although we found that the simple relationship between premarital cohabitation and marital instability no longer exists, we recognize that it may simply be more difficult to detect (e.g., for only some subgroups). We have updated and expanded on Reinhold’s (2010) findings, and we support his argument about an important marriage cohort distinction in the role of cohabitation on marital instability. These findings call for new ways of framing research on cohabitation and marital instability.

We recognize that there is variation in the meaning of cohabitation, and our study builds on previous research on cohabitation and marital instability in that we examined commitment when starting to cohabit, with a focus on both men and women. Just about half of men and women who cohabited prior to a recent marriage had a commitment to marriage when they started living together, and half did not. We found that marital commitment prior to cohabitation was tied to lower hazards of marital instability among women, but not among men. These nationally representative findings for women generally mirror Stanley et al.’s (2010) results reported for a combined sample of men and women. We found that women and men who were engaged at the outset of cohabitation fared no worse, and some women fared better, than their counterparts who did not cohabit. Men who cohabited without engagement shared rates of marital dissolution similar to those of men who did not cohabit and, once premarital experiences were accounted for, women who cohabited without a commitment to marriage shared odds of marital dissolution similar to those of women who never cohabited.

The gender distinctions in the influence of commitment may exist because there are shifts in the plans for marriage during the course of cohabitation, and there are gender differences in regard to the meaning of cohabitation (Guzzo, 2009; Huang, Smock, Manning, & Bergstrom-Lynch, 2011; Manning & Smock, 2002; Sassler, 2004). Our static and retrospective indicator is not capturing the dynamic nature of commitment that could be better assessed with prospective data that reflect both partners’ mutual plans for marriage (Rhoades et al., 2009a; Rhoades, Stanley, & Markman, 2009b; Stanley et al., 2006). Among cohabiting couples, men’s plans for marriage determine the transition to marriage (Brown, 2004) but, once married, men’s plans for marriage may not influence the quality and stability of marriages. It also is possible that women in the most stable marriages may more often retrospectively report engagement at the start of cohabitation, whereas men’s retrospective reports of commitment to marriage may be less influenced by the state of the marriage. Men could be less precise reporters of their relationship experiences, and plans for marriage may take on different meaning for men and women (Huang et al., 2011). Furthermore, the concepts of being “engaged” and having “definite” plans for marriage may seem straightforward, but there is variation in the interpretation of these terms and differences in couples’ interpretation of commitment to marriage or plans for marriage (Manning & Smock, 2009). Further research that considers additional possible ways to capture the heterogeneity of cohabitation and gender differences in the influence of cohabitation on marital quality and stability is warranted.

To date, no study has found a protective influence of cohabitation on marital instability. Among subgroups of women facing the greatest risk of divorce, being engaged at the start of cohabitation appears to be protective and tied to significantly lower odds of marital instability. These findings speak to the importance of recognizing socioeconomic variation in the potential role of cohabitation in regard to marital quality and stability (e.g., Phillips & Sweeney, 2005; Tach & Halpern-Meekin, 2009) and acknowledging there may be potentially positive cohabitation experiences that carry over into marriage. As cohabitation becomes more common, this issue is becoming more salient, with growing variation in the role of cohabitation and marital stability. Furthermore, as cohabitation increasingly becomes a context in which children are raised (Lichter, in press), future research should specifically consider the parental relationships to children (biological and stepchildren) as well as the timing and planning of pregnancy. Extensions of commitment theory that recognize potential socioeconomic variation in the meaning of cohabitation, marriage, and commitment will help move forward our understanding of cohabitation and marital quality and stability.

This study has some additional limitations. First, the social background measures are quite narrow in scope. We had indicators only of parental family structure and mother’s education, although measures such as delinquency or substance use may be important to consider as well (Lonardo, Manning, Giordano, & Longmore, 2010; Woods & Emery, 2002). We think it is important to expand the concept of premarital fertility to distinguish fertility in cohabiting unions and fertility of single women and men. Our purpose was to replicate prior studies, so we did not include this factor, but we believe it is an important step for future research. Second, we did not account for selection processes into marriage from cohabitation. There may be cohort shifts in the factors tied to the transition from cohabitation to marriage. Third, our work did not include period indictors of attitudes or norms toward cohabitation. Although there is a growing acceptance of cohabitation, we could not discern the mechanisms tied to the diffusion process. Qualitative research showcases the socializing influence of families and peers on cohabitation attitudes in the United States (Manning, Cohen, & Smock, 2010), and survey-based findings in Japan and Germany have highlighted the importance of peer experiences (Nazio & Blossfeld, 2003; Rindfuss, Choe, Bumpass, & Tsuya, 2004), but additional research needs to examine variation and diffusion processes with larger samples in the United States. Our analysis focused on divorce within a short time frame, but the event history techniques did include observations among respondents who had not yet experienced divorce, and a substantial share of marriages ended within the exposure time under consideration. Another limitation is that we did not consider the many ways that varying dimensions of premarital relationships may influence marital instability, including serial cohabitation, time spent in relationships, and sexual partnerships. We believe an important next step is to recognize the wide range of premarital experiences and further explore variation in their influence on the timing to marriage, marital quality, and marital stability. Relying on recently collected national-level data of both men and women may offset some of these shortcomings.

This article provides new estimates of the association between premarital cohabitation and the timing of marital instability for women as well as men. Using recently collected data, we demonstrated that the influence of cohabitation on marital stability is not as straightforward as prior work may suggest, with potentially important distinctions according to gender and indicators of disadvantage. This research provides a starting point for a more nuanced understanding of the effects of cohabitation on marital stability. Additional work that considers the relationship between cohabitation and marital quality, as well as marital stability, is warranted. It will be important to replicate these findings with additional national data sources. The influence of cohabitation on marital stability has declined; this may be due in part to the increase in cohabitation. In a broader sense, the findings from this work contribute to our understanding of marital stability and recent family change.