Much research on cohabitation has focused on transitions from cohabitation to marriage or dissolution, but less is known about how rapidly women progress into cohabitation, what factors are associated with the tempo to shared living, and whether the timing into cohabitation is associated with subsequent marital transitions. We use data from the 2006–2013 National Survey of Family Growth to answer these questions among women whose most recent sexual relationship began within 10 years of the interview. Life table results indicate that transitions into cohabitation are most common early in sexual relationships; nearly one-quarter of women had begun cohabiting within six months of becoming sexually involved. Multivariate analyses reveal important social class disparities in the timing to cohabitation. Not only are women from more-advantaged backgrounds significantly less likely to cohabit, but those who do cohabit enter shared living at significantly slower tempos than women whose mothers lacked a college degree. In addition, among sexual relationships that transitioned into cohabiting unions, college-educated women were significantly more likely to transition into marriage than less-educated women. Finally, although the tempo effect is only weakly significant, women who moved in within the first year of their sexual relationship demonstrated lower odds of marrying than did women who deferred cohabiting for over a year. Relationship processes are diverging by social class, contributing to inequality between more- and less-advantaged young adults.
Relationship formation, particularly when accompanied by nonmarital parenting, has emerged as a topic of academic and policy interest (Glenn 2002; Hymowitz et al. 2013; Sawhill 2014; Stanley et al. 2006). Some have argued that contemporary young adults are drifting into sexual relationships and shared living without adequate commitment to their partners (Sawhill 2014; Stanley et al. 2006). Various studies have suggested that rapid relationship progression reduces dedication, is associated with lower marital quality (should couples wed), and adversely affects parenting (Cherlin 2009; Glenn 2002; Sassler et al. 2012; Stanley et al. 2010). To date, however, little is known about how rapidly relationships progress or whether this tempo is associated with subsequent union transitions.
As of the early years of the twenty-first century, more than one-half of young American adults had ever cohabited, and the majority of those who married first lived with their partners (Kennedy and Bumpass 2008). The basic facts about contemporary union formation—that is, what proportion of adults cohabit or marry; how that varies by age, race, and educational attainment; and shifts in the factors conditioning transitions from cohabitation into marriage—are well known (Kennedy and Bumpass 2008; Lichter et al. 2006; Manning et al. 2014). Less often studied is how rapidly sexual relationships transition into cohabitation or marriage, or from cohabitation to marriage. This omission is curious, given evidence showing growing social class divergence in how relationships progress. Even as cohabitation has become more common across all social classes, it remains most prevalent among the economically disadvantaged and moderately educated (Guzzo 2014; Lichter et al. 2010; Manning et al. 2014; Sassler and Miller 2017). Marriage, in contrast, has become increasingly selective of the economically advantaged (Goldstein and Kenney 2001; Lichter et al. 2006; McLanahan 2004). Yet, the majority of those who wed have lived with their partners before the marriage (Manning 2013), raising the question of whether cohabitation operates in the same way for more- and less-advantaged individuals. A better understanding of the factors that expedite or delay the transition of sexual relationships into cohabitation or direct marriage, and from cohabitation to marriage, is needed.
In this article, we use data on young women’s most recent sexual relationships (including those that are current at the time of the interview) from the 2006–2010 and 2011–2013 National Survey of Family Growth (NSFG) to build on previous work in three important ways. Our primary objective is to determine whether the timing to cohabitation is associated with the tempo to marriage among women whose most recent relationship involved a cohabitation spell. To answer this question, we first provide estimates of the timing from sexual involvement to cohabitation and marriage for women’s most recent sexual relationship, using multiple-decrement life table techniques. Second, we update previous studies that explored the factors shaping the transition into either cohabitation or marriage. Third, among those women who first transition into cohabitation, we assess whether the timing to cohabitation is associated with the tempo to marriage. Our results shed additional light on the role that cohabitation plays in the increasingly divergent family behaviors of those from more- and less-privileged backgrounds.
Research on the Tempo of Relationship Progression
Family scholars have increasingly stressed the need to pay closer attention to how relationships unfold (Cherlin 2009; McLanahan 2004; Sassler 2010). Premarital sexual involvement has been normative for several generations. Among cohorts of women turning 15 between 1964 and 1993, at least 86 % had premarital sex by age 25 (Finer 2007). Furthermore, two-thirds of women entering first marriages in the early years of the twenty-first century cohabited with their partners prior to the wedding (Manning 2013). Much attention has focused on aspects of sexual experience—including age at sexual debut, number of sexual partners, and cohabitation experience—but less attention has been paid to how rapidly sexual relationships transition into shared living, whether via cohabitation or marriage, and whether the tempo to cohabitation shapes subsequent transitions into marriage.
Data limitations have largely prevented researchers from distinguishing among respondents with different patterns of sexual relationship progression (Halpern-Meekin and Tach 2013). Despite numerous studies exploring the timing from cohabitation to marriage (Brown 2000; Lichter et al. 2006; Sassler and McNally 2003), or childbearing to marriage (Carlson et al. 2004; Harknett and Kuperberg 2011), to our knowledge, only two published papers have addressed the sequencing of different stages of sexual relationships. Using data from married couples in a nationally representative internet survey conducted in 2010 for the National Center for Family and Marriage Research (NCFMR) by Knowledge Networks, Halpern-Meekin and Tach (2013) explored the association between relationship progression and relationship quality, examining discordance in couples’ retrospective reporting of premarital relationship stages. Women who cohabited prior to marriage reported spending an average of 11 months dating before they began to spend the night together, compared with almost 24 months for women who married without cohabiting.1 Among women who cohabited, the transition from spending the night to living together was rapid (approximately 3 months), but the tempo from shared living to marriage considerably slower (almost 22 months) (Halpern-Meekin and Tach 2013: table 2).2 No information was presented about whether tempos varied by social class, race, or parental status at the time of the relationship’s start.
The second published study concentrated more specifically on relationship tempos. Focusing on young adults (18 to 24), Sassler and Joyner (2011) explored whether racially heterogamous couples proceeded more rapidly into sexual intimacy and from sex to cohabitation, marriage, or dissolution than did homogamous couples. They provided detailed information on the distribution of timing into sexual involvement using data from the 2002 National Longitudinal Study of Adolescent Health (Sassler and Joyner 2011: table 2) and from sexual involvement to marriage, cohabitation, or dissolution (using Add Health and the 2002 NSFG data). Although the largest proportion of women involved in sexual relationships subsequently progressed into cohabiting unions, the paper did not present survival estimates of the timing from sexual involvement to marriage, or from cohabitation to marriage (Sassler and Joyner 2011: table 3). The results indicated that minority women with white partners progressed significantly more rapidly into cohabitation than did women in racially homogamous unions, or white women partnered with minority men. The relative youth of the sample, however, resulted in too few events to examine transitions from cohabitation to marriage or dissolution. Using more recent data from the NSFG, Sassler and colleagues examined how rapidly sexual relationships formed in the prior year progressed into cohabiting unions or dissolutions (Sassler et al. 2016), although given the short window, they were unable to explore subsequent transitions. Nonetheless, they reported that indicators of social class disadvantage expedited transitions into cohabitation.
Qualitative researchers have also sought clarification on how relationships progress, with an emphasis on transitions into cohabitation. Several studies have suggested that entry into cohabitation occurs gradually. Jamison and Ganong (2011) interviewed college-educated daters (n = 22) and described how respondents gradually began spending nights together over time (stayovers, in their terminology), progressively increasing the number of nights. Manning and Smock (2005) drew on interviews with a diverse sample of 115 young adults with current and past cohabitation experience, finding that couples reported spending more and more nights together with increasing relationship duration. However, Sassler (2004) and Sassler and Miller (2011, 2017) noted social class differences in the tempo of relationship progression; these studies were based on two class-diverse samples—one of cohabiting individuals in New York City (n = 30), and the second of cohabiting couples in Columbus, Ohio (n = 122). Among those with a high school diploma or postsecondary schooling but no bachelor’s degree, one-half began cohabiting within six months of the relationship’s start. The college-educated, in contrast, were romantically and sexually involved for longer periods (on average, one year) before moving into shared living (Sassler and Miller 2011), consistent with findings from Jamison and Ganong (2011). Highly educated couples also more frequently reported concrete marriage plans than their less-educated counterparts, suggesting that their transition from cohabitation into marriage occurred more rapidly (Sassler and Miller 2011, 2017). Qualitative studies, with their small sample sizes, are not adequate to explore population-level associations, but their findings highlight the need to further examine variation in sexual relationship progression.
Explanations for Variation in Relationship Tempos and Transitions
Social class of one’s family of origin, race, and occupational and educational pursuits shape young adult’s life course trajectories in important ways. Much of the research on transitions into cohabitation or marriage, or from cohabitation to marriage or dissolution, has focused on individual-level economic determinants of union formation, such as whether school enrollment, educational attainment, or employment and earnings shape transitions into cohabitation or marriage (for a review, see Smock et al. 2005). Early studies of transitions into cohabitation or marriage often included controls for family background (e.g., Carlson et al. 2004; Clarkberg 1999; Manning and Smock 1995; Thornton 1991) but did not focus extensively on social class differences in relationship processes.
In recent years, greater attention to the diverging destinies experienced by youth from more- and less-privileged social class backgrounds has highlighted this research gap (Lichter et al. 2006; McLanahan 2004). A growing body of literature has explored how the social class of youth’s family of origin shapes adolescent and young adult romantic relationships. Youth from less-advantaged backgrounds often enter into sexual relationships at younger ages and engage in more relationships than do young adults from more-privileged family backgrounds (Cavanagh 2007; Cavanagh et al. 2008; Pearson et al. 2006; Sandberg-Thomas and Kamp Dush 2014). We hypothesize that social class also influences the tempo of entering into shared living, and that this may be associated with subsequent union transitions. We operationalize social class of family of origin in terms of family structure through adolescence, maternal marital status at birth, and maternal educational attainment because together, these indicators represent the relative social and economic position of families in the United States (Lareau 2003; McLanahan 2004).
Economic disadvantage among the parental generation affects young adults’ relationship experiences in various ways (Amato 2005; Cherlin 1995; Graefe and Lichter 1999; McLanahan 2004). Disadvantaged status is frequently transmitted across the generations: women who grew up in poor families engage in partnering and parenting at younger ages than their more-advantaged counterparts (Amato 2005; Cavanagh et al. 2008; Cooksey et al. 2002; Pearson et al. 2006; Sassler et al. 2016). Children from single-parent or stepparent families also receive less financial support to achieve academically than do those from intact married-parent families (Astone and McLanahan 1991; Ermisch and Francesconi 2001). Although experiencing family disruption or parental repartnering accelerates young adults’ departure from the parental home (Aquilino 1991; Cherlin 1995; Goldscheider and Goldscheider 1998; Teachman 2003), youth from economically disadvantaged backgrounds often lack the resources necessary to maintain independent households, perhaps due to low-paying jobs or the absence of parental assistance for residential autonomy (Sassler and Miller 2011). Those growing up in middle-class families, in contrast, are often encouraged to delay forming serious relationships until later in the life course (Cavanagh 2011). Youth born to highly educated mothers are expected to attend and complete college prior to embarking on intimate coresidential relationships (Furstenberg 2008; Lareau 2003). Although they may not defer sexual relationships, evidence suggests that they maintain separate homes during the early stages of dating (Jamison and Ganong 2011; Sassler and Miller 2011), perhaps because they receive parental assistance in paying for schooling and housing (Rindfuss and VandenHeuvel 1990).
Extending the literature on variation in transitions into cohabitation versus marriage, we anticipate that indicators of economic disadvantage will not only increase the odds of entering into cohabitation versus marriage but also expedite transitions from sexual relationships into cohabitations while deterring marriage. Markers of family advantage, in contrast, are expected to delay entrance into cohabiting unions. Research highlights that youth from more-advantaged social class backgrounds are more likely to marry directly. However, whether they also experience more rapid transitions from sexual relationships into marriage is unclear, given the lengthy process of acquiring educational and occupational credentials needed for transmission of middle-class status.
Markers of advantage have also been found to elevate the likelihood of marriage among cohabitors: cohabiting women who have completed college are significantly more likely to marry than less-educated women (Lichter et al. 2006). The qualitative research on cohabitors has also found that less-advantaged cohabitors often stress that particular factors—such as completing school, obtaining stable employment, or paying off loans—must be in place before a marriage can ensue (Sassler and Miller 2011; Smock et al. 2005). We therefore expect that, conditional on cohabiting, more-advantaged women will be more likely to transition into marriage, and will do so more rapidly, than will their less-advantaged peers.
Data and Methods
Data come from the 2006–2010 and 2011–2013 NSFG, an ongoing nationally representative cross-sectional sample of U.S. men and women between the ages of 15 and 44. A unique strength of the NSFG is its collection of information on the month and year that the respondent and her most recent sexual partner first had sexual intercourse, so long as either the partner was current as of the interview date or the relationship ended within the 12 months prior to the interview date. The overwhelming majority of women reported being in a sexual relationship in the past 12 months.3 We limit our sample to those women whose most recent or current sexual relationship began within 10 years of the interview: 10,045 women from the 2006–2010 NSFG, and 4,588 women from the 2011–2013 NSFG. We use the date of first sexual intercourse as the starting point for the relationship. Dates were also obtained for the month and year when respondents began living with their partners4 and married.
Of the 14,633 women aged 18–44 in the combined 2006–2010 and 2011–2013 NSFG whose sexual relationship began within 10 years of the interview date, we restrict our analysis to women who fell into one of the following categories: (1) currently dating but not coresiding; (2) currently cohabiting; (3) currently married; or (4) separated within the last 12 months from their most recent partner and not yet in a new sexual relationship (n = 12,681). We exclude 1,057 women who had been married previously given that retrospective recall bias is lower for first marriages (Bumpass and Lu 2000; Kuo and Raley 2016). Also excluded are 587 women who had not ever had sex prior to moving in with their partners, and another 225 women who had missing information on the date of move-in or who reported dates that resulted in a negative duration between the date of first sex with the most recent partner and the date of move-in.5 Finally, we eliminate 122 women who had missing information for their mother’s educational attainment, a key covariate in our analysis.
Our final sample contains 6,086 women whose most recent or current sexual relationship had started within 10 years of their survey date and who were between the ages of 18 and 44 at the time of the interview: 545 women who married directly and were still with that partner; 2,831 women who had entered into a cohabiting union; and 2,710 women who were sexually involved but did not enter into shared living with their partners. For the second stage of our analysis, we limit our sample to the 2,831 women whose most recent sexual relationship involved a cohabiting union, and we explore whether their unions transitioned into marriage (n = 1,158). Even though our 10-year time frame surpasses the 5-year duration used by others (Bumpass and Lu 2000; Hayford and Morgan 2008; Kuo and Raley 2016), these studies began their clock when respondents began cohabitating rather than when the sexual relationship started. A five-year window simply does not provide enough observations of transitions from sexual involvement into cohabitation and then to marriage.6 We employ life table methods and event-history analysis to control for the start year and duration of current or most recent sexual relationships.
Multiple-Decrement Life Table Estimates
We first provide life table estimates of transitions of respondents’ most recent sexual relationships into coresidential unions, distinguishing between direct marriage and cohabitation. Multiple-decrement life tables are used to estimate the likelihood of cohabitation or marriage (Preston et al. 2001). The cumulative proportion of women entering cohabitation or marriage within three years of initiating the sexual relationship are presented. Relationships that we do not observe transitioning into cohabitation or marriage during this window are censored at the interview date or the breakup (the date of last sexual encounter), whichever comes first. Next, we examine life table estimates of transitions out of cohabiting unions to estimate the likelihood of marriage, presenting the cumulative proportion of women marrying within three years of moving in together, censoring at the interview date or the breakup.
We next fit discrete-time event-history models that examine various relationship progression stages. Our first analysis examines transitions out of sexual relationships, treating entrance into cohabitation or marriage as competing risks relative to continuance in sexually intimate noncohabiting relationships. Because the data are measured in months, our approach is quite similar to a continuous-time hazard model (Allison 1984; Kalbfleisch and Prentice 2002). Respondents contribute person-months to the data until they experience an event—marriage or cohabitation—or are censored (either at the interview date or the breakup). Our second analysis limits the sample to women whose most recent relationship involved a cohabiting union and examines transitions to marriage. We assume that our outcomes are distinct events influenced by different underlying mechanisms (Allison 1984).
Our first model takes the following functional form:
where Pijt is the conditional probability of experiencing direct marriage or cohabitation (j = 1 for cohabitation, or j = 2 for marriage; j = 0 for censored cases) for woman i at year t since the start of the sexual relationship, given that she has not yet experienced an event or been censored prior to year t. αij is a set of dummy variables to control for time dependence (in six-month increments). After exploring various measures of duration to ascertain temporal patterns, we opt for the interval measure because it captures the time effect more accurately and has the best model fit.7 We include a set of M time-constant variables as well as N time-varying variables (measured at time t – 1).
Our second model takes a logistic regression form, where Pit is the conditional probability of experiencing a marriage for a cohabiting woman i at month t since the start of the cohabitation, given that she has not yet experienced an event or been censored prior to month t. Here, we model duration since cohabitation date as a quadratic because this model provides the best fit over other models that specify the duration in 6-month or 12-month intervals. Our primary independent variables of interest (described in the next subsection) are time-constant, but we also include some time-varying control variables.
Independent Variables: Social Class Measures
Our key independent variables are proxies of social class, measured by respondents’ childhood family characteristics. Family structure is measured with dummy variables for whether respondents always lived with biological or adoptive parents (reference) or ever lived with a single mother, a stepfather, or no biological parents. We also assess the educational attainment of the respondent’s mother: less than a high school diploma, high school diploma (reference), some postsecondary education but no degree, or a college degree or more. Another measure indicates whether respondents’ parents were married at their birth (married = 0, unmarried = 1). Last, we determine whether the respondent had completed high school at the start of the sexual relationship. Other indicators (e.g., the age of the respondent’s mother at her birth and the respondent’s number of siblings) were tested but not included because they never reached conventional levels of statistical significance.
Our second analysis, which is limited to those whose most recent relationship transitioned into cohabitation, includes family structure during childhood but omits our indicators of maternal education and maternal marital status at birth because they never reached conventional levels of significance. Our measure of social class now includes information about the respondents’ own education when they began cohabiting in order to reduce concerns of endogeneity of educational attainment to cohabitation and marriage decisions. This is an imperfect measure because pursuing postsecondary education has become increasingly protracted (Bound et al. 2012), and many of our respondents were young and may obtain additional schooling and, possibly, degrees after the cohabitation date. The NSFG data on school enrollment, however, are not detailed enough to enable time-varying measures of school attendance and completion. Nonetheless, a large proportion of our respondents (71 %) had finished their education prior to entering their cohabiting union. We therefore examine respondents’ own educational attainment as a social class indicator of transitions from cohabitation into marriage, comparing respondents with some postsecondary schooling but no degree (reference) with those with less than a high school diploma, a high school diploma, or a college degree (or more). Respondents who did not have a high school diploma at the time of the interview, or who reported that the high school diploma date was after the move-in date, are coded as having less than a high school diploma. They are classified as high school graduates if their highest degree as of the interview date was a high school diploma. Women who reported having some postsecondary schooling at the time of their interview, or who reported that their college degree date was after the move-in date, are classified as having some college education. Finally, respondents are considered college graduates if their college graduation date was prior to the move-in date.
Other background measures include the respondents’ race and ethnicity, categorized as non-Hispanic white (reference), black, Hispanic, and other. We include dummy variables to determine how old the respondent was at the start of her most recent sexual relationship: younger than 18, 18–20, 21–24, 25–29, or 30 or older. We also include a measure of when the relationship began: 1996–2000 (reference), 2001–2003, 2004–2006, 2007–2009, or 2010–2012. These controls address concerns of recall bias that may be introduced by including relationships formed up to 10 years prior to the interview date. Our last measures account for respondents’ sexual and relationship histories. We include a continuous measure of the number of prior sexual partners, top-coded at 20 partners (which represented the 99th percentile), as well as several dichotomous indicators of whether the respondent had ever cohabited with another partner or was a parent at the start of the most recent sexual relationship (1 = yes for all measures). Our final measure is a lagged time-varying indicator of whether the respondent was pregnant (1 = yes).
Most of the same measures are used in our second analysis, with a few exceptions. We construct a measure of the time to shared living. Of those who entered into a cohabiting union, more than one-half (59 %) had moved in with their partner within one year of starting their sexual relationships, consistent with the findings of some qualitative studies (Sassler 2004; Sassler and Miller 2011). By the end of the second year of sexual involvement, more than three-quarters of respondents who entered a cohabiting union (78 %) had moved in with their partner.8 We therefore create a dummy variable indicating those who entered their cohabiting union within the first year of their relationship; those who deferred cohabiting for one year or longer serve as the reference group. We also specify the month since the respondents began their cohabiting union, using a linear and quadratic term, because preliminary results suggest a nonlinear relationship between time cohabiting and marriage. Our last new measure introduces a linear term for the year the cohabiting union began, which spans 1996–2012.
All means for the descriptive analyses rely on NSFG person weights estimated for the entire sample; our multivariate models are also weighted using the NSFG person weights. Means are presented in Table 1. A sizable proportion of our respondents experienced some form of family disadvantage. Less than two-thirds of the women had grown up in intact, married-parent families (column A), and similar proportions had grown up with either a low-educated mother or a college-educated one (18.9 % vs. 21.4 %). However, those who married directly (column B) differ from women who cohabited (column C) and from women in a nonresidential sexual relationship (column D). More than three-quarters of those who married directly grew up in an intact family, compared with lower proportions of both other groups of women. Direct marriages were more likely to occur among those with college-educated mothers and who had a college degree themselves. Compared with those who formed cohabiting unions or were dating but not coresiding with their partners, women who married directly were older, on average, at the time of their interview, and their relationships more often had begun in the late 1990s and early 2000s. Women who entered into cohabiting unions also differed from those who remained in dating relationships. Those who cohabited were younger when they began their sexual relationship and were more likely to be white and college-educated. Dating women were significantly more likely than cohabiting and married women to have begun their relationship recently, to be parents, and to have previously lived with a cohabiting partner. They also reported the highest mean number of lifetime sexual partners.
Among our initial sample of women whose most recent relationship began within 10 years of the interview date, a relatively small proportion—only 9.0 %—married directly. The remaining sample is roughly evenly divided between women who cohabited with their most recent partner (46.5 %) and women who had not (yet) coresided with their most recent partner (44.5 %). The mean duration from sexual involvement to direct marriage was approximately 19 months, shorter than reported by Halpern-Meekin and Tach (2013), although transitions into cohabitation occurred at a similar tempo (approximately 13 months).
Life Table Estimates
Our life table estimates of relationship trajectories reveal that transitions into cohabitation are most common early on in sexual relationships (Table 2, panel A). Nearly one-quarter of women had begun cohabiting within six months of becoming sexually involved, whereas only 3.4 % had married directly. By the 12-month mark, the proportion of women entering into cohabiting unions was more than five times greater than the proportion marrying directly (33 % vs. 6 %). Transitions into cohabitation continue to outpace transitions into marriage in subsequent months; nearly one-half of women’s most recent sexual relationships (47.3 %) became cohabiting unions by 24 months, still four times greater than the share marrying directly (10.8 %). Overall, of the most recent sexual relationships that persisted at least three years, less than one-third remained in a sexual relationship without transitioning into either cohabitation or directly to marriage.
Panel B of Table 2 provides information about the experiences of women whose most recent sexual relationship transitioned into cohabiting unions. Transitions from cohabitation to marriage occurred far more gradually than did transitions into cohabiting unions. Only 16.2 % of never-married women who entered into a cohabiting union in their most recent sexual relationship had married within 12 months of moving in with their partners. In fact, by the third-year mark, the proportion of women who had married their cohabiting partner had still not approached the proportion that remained cohabiting or had broken up (44.2 % vs. 55.5 %, respectively).
Union Transitions in Sexual Relationships
The risk ratios of cohabitation and marriage relative to continuing in a dating relationship are presented in Table 3, in three sequential models. The first (Model A) examines the tempo of relationship transitions as a function of the intervals since the start of the sexual relationship. Next, we include controls for social class indicators (Model B). Last, we include controls for respondents’ race and relationship history (Model C). Relative risk ratios (or the exponentiated coefficients of the parameter estimates) can be interpreted as the change in the risk ratio of entering a cohabiting or marital union associated with a one-unit increase in the independent variable. These analyses serve to identify key temporal and social class differences in union formation processes.
Results from Table 3 (Model A) reveal the temporal patterns of transitioning into cohabitation or direct marriage, confirming the life table results. Most cohabiting unions were formed quickly, within the first year of sexual involvement. Those who did not enter into shared living early on tended not to cohabit. The relative risk of cohabiting for women who have been sexually involved for 6 or fewer months was 49.1 % greater than for women who had been dating for 13 to 18 months. In fact, the odds of cohabiting decreased substantially with increased duration in a relationship; the relative risk of cohabiting was 38.4 % greater among those sexually involved for 6 or fewer months than it was for women who dated for 7 to 12 months (relative risk ratio = 1.491 / 1.077), whereas those involved for longer than three years were significantly less likely to cohabit than those involved for more than one year but less than 18 months. In fact, at the shortest duration (less than 6 months), women were significantly more likely to cohabit than to marry. The pace of entrance into marriage is less consistent, and none of the timing coefficients in Model A reach statistical significance, suggesting no evident trend in timing to marriage from sexual involvement.
The tempo patterns regarding entrance into cohabitation and marriage are largely unchanged upon controlling for family social class (Model B). As hypothesized, women who experienced family disruption or reformation—living with a single mother or in a stepfamily by adolescence—were substantially less likely to marry than to enter into a cohabiting union or remain in a nonresidential sexual relationship. However, women who experienced parental union instability were no more likely to enter cohabiting unions than to remain in nonresidential dating relationships. Women whose mothers were unmarried at birth were significantly less likely to cohabit or marry, versus staying sexually involved, than women born to married mothers. In contrast, women who grew up in more-privileged families—such as those with college-educated mothers—were significantly less likely to cohabit than to remain in a nonresidential sexual relationship and were more likely to marry directly than to cohabit. Those who completed high school before starting their relationship, as well as those with the least-educated mothers, were also significantly more like to marry directly than to remain dating or cohabit.
The temporal associations observed in the reduced model remain after accounting for racial background and women’s sexual and relationship history (Model C), with one notable exception: at longer durations, relationships are more likely to transition into marriage than cohabitation or remaining in a sexual, nonresidential union. The relative risks of marrying directly are 54.8 % greater among women who have been dating for three or more years relative to those involved for one year to 18 months. Other measures operate as hypothesized after race and sexual history measures are controlled, although fewer indicators reach conventional levels of significance. The effect of maternal marital status and maternal education operate as hypothesized. Women born to unmarried mothers are significantly less likely to marry directly, relative to both cohabiting and remaining dating, and those born to college-educated mothers are significantly less likely to enter cohabiting unions than to remain dating or marry directly.
Our other controls reveal how relationship tempos are shaped by background characteristics. Black women were significantly less likely than white women to both marry directly and cohabit, relative to remaining in a dating relationship. Age at the start of the sexual relationship is strongly associated with both entrance into direct marriage and entrance into cohabitation, although the pattern differs. The odds of direct marriage increase with women’s age at the start of the sexual relationship, but the pattern is curvilinear for transitions into cohabitation, with the youngest and oldest women significantly less likely to enter a cohabiting union than women in their 20s, relative to remaining in a dating relationship. Furthermore, the oldest women were significantly more likely to marry directly than to cohabit. We find no clear association between the year when the sexual relationship began and transitions into either cohabitation or direct marriage, although women whose relationship began during the Great Recession (2007–2009) were significantly less likely to cohabit compared with relationships that began in the late 1990s. Our indicators of sexual history operate as expected, with greater relationship experience—more prior sexual partners, having previously cohabited, and being a mother prior to the start of the relationship—strongly associated with a lower risk of entering directly into marriage. Consistent with the demographic literature, conception is associated with both transitions into cohabiting unions as well as direct marriage, relative to dating.
One of our goals is to examine whether tempo into cohabitation differs by social class. Based on Model C of Table 3, we model the timing to cohabitation, relative to remaining sexually involved, separately by maternal educational attainment, to allow for the effect of duration to differ across categories. Our approach is similar to a single regression with interactions between our social class indicators and duration since first sexual encounter. Rather than six-month intervals used in Table 3, we instead examine a quadratic monthly measure of duration (shown in Fig. 1) for ease of presentation. We then graph conditional predicted probabilities of entering cohabitation or marriage by months since the initiation of the sexual relationship.
Not only are women who grow up with college-educated mothers less likely to cohabit, but those who do cohabit enter into shared living at significantly slower tempos than do women who grew up with mothers who lacked a college degree, as shown in Fig. 1. Entry into cohabitation occurs the most slowly among women whose mothers had a college degree. After more than one year of sexual involvement, the cohabitation trajectories of women with the most-educated and less-educated mothers begin converging; not until approximately 19 months after the start of their sexual relationship do women with highly educated mothers catch up with those with moderately educated mothers. Figure 1 also demonstrates that transitions into cohabitation have the highest probability of occurring early in sexual relationships. The longer a woman has not entered a coresidential union with her sexual partner, the lower her odds of subsequently doing so.
Transitions From Cohabitation: Does Tempo to Cohabitation Matter?
We turn now to our second multivariate analysis. Because previous studies have extensively explored the factors associated with marital transitions of cohabitors, we focus mainly on our temporal measures. Previously, we showed how rapidly many women entered cohabiting unions. Results from Table 4 (Model A) reveal that cohabitors who moved in with their partners within the first year were, in fact, less likely to transition into marriage than those who took longer, although net of other controls, this effect is negative but only marginally significant (p < .053). After women are living with a partner, they are more likely to marry with each month since moving in albeit at a declining rate. The odds ratio for our linear measure of months since the start of the cohabitation is greater than 1, and the quadratic term is almost 1 (with rounding), and both terms are significant.
Accounting for social class of family of origin and own educational attainment (Model B) weakens the association between duration to cohabitation and marriage to insignificance, although the temporal effects of timing since entrance into cohabiting unions remain the same. Even after we account for the tempo into shared living and duration since the start of the cohabitation, cohabiting women’s educational attainment is strongly associated with transitions into marriage. This finding is consistent with previous studies (Kuo and Raley 2016; Lichter et al. 2006).
Accounting for race and relationship history (Model C) increases the significance of our tempo measures, although the effects of family structure and respondents’ education are weaker, perhaps because of the associations among family structure, race, and the onset and progression of sexual relationships (Cavanagh 2007, 2011). Women who moved in within the first year of their sexual relationship had odds of marrying their partners that were marginally significantly lower than for women who deferred cohabiting for more than one year. The temporal pattern of transitioning from cohabitation into marriage remained the same: the probability of marrying increased with each month since moving in, but at a slower rate.
Educational attainment remains a strong predictor of transitions from cohabitation into marriage after controlling for other factors, such as race and relationship history. Less-educated women are significantly less likely to marry their cohabiting partners than are women who have completed college. Women college graduates have relative risks of marrying that are 79 % greater than women who have some postsecondary schooling but no degree, relative to remaining in their cohabiting union. Unlike earlier studies of cohabiting women’s transitions into marriage (e.g., Brown 2000; Sassler and McNally 2003), where women’s educational attainment was not a strong predictor of union transitions, our results reveal further widening of disparities between college educated women and others. We also interact our tempo measures by educational attainment (results not shown) to determine whether timing to cohabitation or since the start of living together differentially influenced union transitions. The negative association between moving in with a partner within the first year of the start of the relationship and marriage is significant only for women who had graduated from college.
Racial minorities exhibit significantly lower odds of transitioning from cohabitation to marriage than do their white counterparts; we find no significant differences between black and Hispanic cohabitors in their odds of marrying. We also find an interesting temporal effect: cohabiting unions that began in more recent years were significantly less likely to transition into marriage than those that began in the late 1990s. Furthermore, cohabitors who conceived after moving in with their partner were more likely to marry.
To further illustrate tempo variations in relationship transitions, we model the logistic regressions shown in Table 4 by the respondents’ educational attainment, allowing duration to differ across categories of education. Figure 2 presents monthly predicted probabilities of marrying separately for each of the education groups, and Fig. 3 demonstrates cumulative predicted probabilities. Results from Fig. 2 confirm the patterns of Table 4: the probabilities of marrying are higher, at every duration, for the college-educated, followed by those with some postsecondary schooling. Among high school graduates, the odds of marrying decrease slightly with each additional month they cohabit with their partners, and a similar pattern is shown among high school dropouts.
These monthly patterns cumulate to very different outcomes, as shown in Fig. 3. College-educated women are much more likely to transition from cohabitation to marriage than their lower-educated counterparts. After three years, college-educated cohabiting women are nearly twice as likely to have married as women with only a high school diploma. This finding further confirms that cohabitation serves a different purpose for highly educated women compared with their less-advantaged counterparts. College-educated women are much more likely to use cohabitation as a stepping stone to marriage.
Previous theory and research on the progression of sexual relationships into cohabitation or marriage, and from cohabitation into marriage, are limited. Most research on cohabitation begins observing respondents or couples only at the time of coresidence, overlooking the period preceding the entrance into shared living. However, the tempo of transitions into cohabitation may reflect various factors that influence subsequent trajectories. In this article, we used retrospective data from the 2006–2010 and 2011–2013 NSFG to examine the progression of women’s most recent sexual relationships into direct marriage or cohabitation, and then we assessed whether rapid or more tempered transitions into cohabitation were associated with subsequent transitions into marriage. The role of social class in differentiating the tempo of progression, into direct marriage or cohabitation, and from cohabitation to marriage, was a particular focus of our study.
The empirical results expose several patterns of interest to family demographers. First, they reveal that, on average, transitions into cohabiting unions occur rapidly after couples become sexually involved. The likelihood of entering cohabiting unions declines the longer women remain sexually involved. The rapid tempo to shared living exhibited among women from less-advantaged backgrounds is consistent with both qualitative (Sassler 2004; Sassler and Miller 2011) and quantitative research (Sassler et al. 2016) and suggests that economic reasons may play a larger role as an impetus into shared living than compatibility testing. Women who enter rapidly into cohabiting unions may view living together as a more intensive form of dating or as an alternative to being single, as well as a means to take advantage of factors (e.g., economies of scale, companionship) formerly touted as benefits of marriage.
Second, our analysis reveals important social class distinctions in the tempo of transitions into cohabitation, and from cohabitation into marriage. These findings shed some light on the processes contributing to diverging relationship trajectories. Rapidly formed cohabiting unions are less likely to transition into marriage than are cohabitations that are formed more slowly. But it is women from less-advantaged family backgrounds who are more likely to transition quickly from sexual involvement to cohabitation. For the most-educated women, in contrast, a protracted period of sexual involvement prior to cohabitation suggests that cohabitation is more often used as a precursor to marriage. Even as some social commentators urge couples to “slow down” in forming intimate relationships (Cherlin 2009), a better understanding of the factors associated with expedited or delayed transitions into cohabitation is required. The relationship between rapid entrance into cohabitation and subsequently lower levels of dedication or commitment proposed by Stanley et al. (2006) may be spurious. Negative selection of cohabitors with poorer economic traits accounts for lower levels of marital quality (Spencer and Beattie 2012) and higher levels of divorce (Lu et al. 2012). Nonetheless, such studies failed to account for timing into shared living, which previous research has found to matter with regard to relationship quality and satisfaction (Sassler et al. 2012). Additional exploration is required to ascertain the extent to which decisions to cohabit are motivated by economic difficulties, housing exigencies, or other factors such as unintended pregnancy, as some qualitative work has suggested (Sassler 2004; Sassler and Miller 2011). Doing so will require data better suited for studying transitions in the labor force or housing market.
Third, we found no evidence that those who rapidly moved in with their current partners progressed at an equally accelerated pace into marriage. Instead, our findings suggest that rapid transitions into cohabitation may reduce the odds of transitioning into marriage, although these results are only weakly significant. We could not, of course, directly address the marital intentions of women entering into cohabitations, but various studies have highlighted an important finding that ours further extended (cf. Lichter et al. 2006): among cohabiting women, the less-advantaged are far less likely to transition into marriage than their counterparts with more education and advantaged family backgrounds.
Our approach does, of course, have limitations. We were not able to ascertain how long couples were romantically involved prior to engaging in sexual activity, which may be another key predictor of relationship quality. Previous research suggests that the tempo of progression from romantic unions into sexual ones has remained relatively consistent, although the data sources used for these inferences are imperfect; such work suggests that one-third to two-fifths of American adults become sexually involved within the first month of romance (Busby et al. 2010; Cohen and Shotland 1996; Regnerus and Ueker 2011). Nonetheless, the earliest piece of the puzzle—when relationships started and how they progress to sexual involvement—is still missing. Furthermore, we are unable to determine whether the tempo to shared living is associated with poorer- (or higher-) quality relationships, given that measures of relationship quality are not obtained by the NSFG. Also lacking in the NSFG is information about employment or schooling transitions that may shape union transitions in important ways.
As a means of reducing recall bias, other studies of transitions from cohabitation into marriage have limited their time frame to shorter periods, such as cohabiting unions formed in the previous five years. Our longer period of coverage—10 years—overrepresents the outcomes of relationships that persisted for longer periods, or more “successful” unions. Because of data limitations, we miss many short-term sexual relationships (Sassler et al. 2016) as well as less-stable relationships formed within 10 years of the interview date that dissolved more than 12 months prior to the interview date. Even though we find no evidence of calendar year trends in transitions into cohabitation, our results suggest that more recently formed cohabitations are less likely to transition into marriage. This finding is consistent with evidence from other studies showing that contemporary cohabitations are taking longer, on average, to transition to marriage than those formed in earlier periods (Lichter et al. 2006; Manning et al. 2014). It also suggests that for some—particularly the less-educated—cohabitation may be transforming into an alternative to marriage. Other data sources that focus on cohorts may better enable us to untangle whether our associations are a function of time, opportunity, or exposure.
Clearly, our analysis represents a first step in learning about the tempo of sexual relationship progression into cohabitation and beyond. The underlying presumption of those concerned with the quality and stability of contemporary relationships is that building strong relationships requires time. The findings that more-advantaged women progressed more slowly into shared living in their most recent relationship and were also more likely to marry those partners seem to further substantiate the “go slow” (or slower) argument (Cherlin 2009). However, these findings beg the question of which factors enabled those women to transition more slowly into shared living. A better understanding of the mechanisms resulting in faster or slower relationship tempos is needed to shed additional light on the social class dynamics increasingly evident in the family building patterns of today’s youth.
An earlier version of this article was presented at the 2012 National Survey of Family Growth Research Conference and at the 2013 annual meeting of the Population Association of America.
The meaning of “dating” was left up to the respondent; no information was obtained about when the sexual relationship began. These were retrospective reports of relationships that began, on average, in the mid-1980s and early 1990s.
Compared with women, men reported shorter durations between dating and spending the night, and between spending the night and officially living together, but a longer duration from cohabitation until marriage. Couple-level disagreement over start times of different stages was common.
In the 2006–2010 NSFG, 90 % of women reported having at least one sexual partner in the past 12 months, as did 85 % of the women in the 2011–2013 NSFG.
The NSFG defines living together as having a sexual relationship while sharing the same usual residence. This specification results in a smaller proportion of cohabitors than found in some other nationally representative samples (e.g., Sassler and Joyner 2011).
Estimating the duration to cohabitation results in a number of women with negative durations between first sex and cohabitation because of inconsistencies in reported dates (n = 306). After consulting with researchers at the NSFG, we adjusted 84 cases for which the difference between the date of first sex and the move-in date was one month, assuming that these two events happened at approximately the same time. We adjusted another 64 cases where the dates of first sex and move-in resulted in a negative duration because when respondents were asked how old they were at each of these events, a positive duration was calculated between date of first sex and date of cohabitation/marriage. Finally, we adjusted two cases for which a negative duration resulted from imputing the season of first sex or move-in when respondents did not report a precise month. All told, we adjusted 150 cases following these NSFG guidelines.
Additional analyses using narrower windows (five or eight years, for example) revealed that the main results were quite similar, although given fewer transitions, often did not reach conventional levels of significance.
Using six-month intervals yielded the lowest Bayesian information criterion (BIC), when compared with a linear or quadratic function of duration, and therefore provided the best model fit of the data.
More than 90 % had entered their cohabiting union within 40 months; the remaining 5 % were sexually involved for more than 45 months before entering into shared living, with the longest duration between sexual involvement and coresidence being 119 months.