The tempo of relationship progression among low-income couples

Abstract

This paper examines the factors associated with the tempo of low-income couples’ relationship progression into sexual involvement and coresidence. Data come from a recently-collected survey, the Marital and Relationship Survey (MARS) that obtained information from low- to moderate-income married and cohabiting couples. Over one-fifth of male and female respondents reported becoming sexually involved with their current partner within the first week of dating. Entrance into shared living was also quite rapid; about one-third of respondents moved in with their partner within 6 months. Furthermore, about two-thirds of married respondents initially cohabited with their partners. Indicators of family disadvantage accelerated entrance into sexual involvement and coresidence; these effects are more pronounced for women than men. Our results also suggest that the pace of relationship progression, into sexual involvement as well as shared living, has accelerated among unions formed more recently.

1. Introduction

“It is not time or opportunity that is to determine intimacy – it is disposition alone. Seven years would be insufficient to make some people acquainted with each other, and seven days are more than enough for others.” – Sense & Sensibility, by Jane Austen (1811).

Concern with the state of American’s relationships has risen in the past few decades, fueled by marital delay, high rates of divorce and non-marital childbearing, and perceptions that the partnering behavior of contemporary young adults has changed irreversibly. Apprehension regarding how rapidly relationships become sexual, for instance, can be seen in the growing attention paid to “hook ups” (Boggle, 2007; Stepp, 2007; Whitehead and Popenoe, 2000). The federal government has even gotten involved, supporting relationships skills courses with funding provided by the Department of Health and Human Services’ Healthy Marriage Initiative, which had a budget of up to $150 million per year following the passage of the Deficit Reduction Act of 2005 (U.S. Department of Health and Human Services, 2006).

Researchers have begun to highlight the need to know more about how contemporary relationships unfold (Cherlin, 2009; Sassler, 2010). Some scholars have posited that Americans become sexually involved too early within new relationships, and as a result become prematurely entangled (Glenn, 2002). Others suggest that couples who move rapidly into cohabiting relationships also frequently “slide” into marriage without adequate commitment to partners (Rhoades et al., 2006; Stanley et al., 2006). The research on the tempo of relationship progression, however, is relatively scarce. What little exists is often based on small samples, frequently of cohabitors or college students (e.g., Boggle, 2007; Sassler, 2004; Stanley et al., 2006; Whitehead and Popenoe, 2000). We therefore know relatively little about how rapidly relationships actually progress, whether that varies by population group, or if it has changed over time. This research gap is particularly meaningful because many publicly funded programs designed to strengthen low-income couples’ relationships lack a fundamental understanding of how couples come to be living together (Ooms and Wilson, 2004) – where they met, how long they dated before becoming sexually intimate, and what factors expedite or delay the progression of established couple relationships.

The period between the start of a romantic relationship and its transition to a sexual or coresidential union is an important one. Longer courtships enable partners to gather information about each other and assess compatibility, communication skills, and life goals. A better understanding of the challenges facing the ability of less advantaged couples to build strong and healthy partnerships, then, requires more knowledge about the early stages of their relationships. Our paper addresses this gap. We examine first sexual involvement within relationships and the entrance into shared living as particular “events” within romantic unions, and explore the factors associated with relationship tempo. Data are from a recently-collected survey, the Marital and Relationship Survey (MARS) that obtained information from low-income married or cohabiting individuals living with minor children (for a description, see Lichter and Carmalt, 2009).

2. Research on the tempo of relationship progression

The belief that modern relationships evolve differently from those of the past is quite prevalent in contemporary American society (e.g., Blow, 2008; Boggle, 2007; Stepp, 2007; Whitehead and Popenoe, 2000). Scholars who study relationship formation and dissolution have contributed to this perception, some by highlighting the rapid metabolism of American’s relationships (Schoen and Standish, 2001), and others by encouraging Americans to slow down the pace at which they form new unions (Cherlin, 2009; Glenn, 2002). To date, however, there has been very little research documenting the tempo of relationship progression into sexual involvement or shared living.

This absence is rather surprising, given the extensive body of research by social psychologists studying the association between relationship progression and commitment (e.g., Kline et al., 2004; Lloyd and Cate, 1985; Stanley et al., 2006; Surra, 1987). A synthesis of such work reveals that relationships that evolve as a result of spending time together, growing levels of behavioral interdependence, the development and deepening of trust, and increases in commitment and dedication result in higher quality matches than unions where commitment levels change rapidly, often as a response to external happenings such as the loss of housing, job shifts, or pregnancy (Kline et al., 2004; Reed, 2006; Stanley et al., 2004; Surra, 1987; Surra and Hughes, 1997). Early studies of dating college students, for example, found that couples who waited longer to become sexually intimate reported higher levels of love, closeness, and knowledge of their partner (Peplau et al., 1977). More recently, scholars have hypothesized that rapid progression into sexual intimacy and shared living results in lower levels of commitment - to partners and to marriage (e.g., Rhoades et al., 2009; Stanley et al., 2006). Those concerned with premarital relationships have suggested that rapid sexual involvement leads to premature entanglements, making relationships harder to exit and reducing the likelihood that lower quality matches are winnowed out (Glenn, 2002). Other research reports that many individuals “slide” into shared living, often without having discussed plans for the future (Manning and Smock, 2005; Sassler, 2004; Stanley et al., 2006). The evidence suggests that rapid entrance into shared living may result in miscommunication and lower levels of relational commitment (Kline et al., 2004; Rhoades et al., 2006, 2009; Stanley et al., 2006).

Published studies documenting the length of time between the start of a relationship and sexual involvement are scarce. This paucity of information is due largely to limitations in survey data; information on when or how romantic relationships begin is rarely gathered. Despite the dramatic increase in cohabitation (Kennedy and Bumpass, 2008), what little evidence exists with regards to cohabiting relationships prior to entrance into shared living is drawn from qualitative studies. In an examination of 25 cohabitors residing in New York City, Sassler (2004) reported that over half had moved in with their partner within 6 months of the relationship’s start. Other qualitative studies also suggest that cohabitors slide rapidly into shared living (Manning and Smock, 2005; Stanley et al., 2006).

There is a similar lack of data regarding how long couples date if they enter directly into marriage without first living together. Such a population is increasingly selective, as the majority of all marriages today are preceded by cohabitation (Kennedy and Bumpass, 2008).² Those who marry directly may enter into shared living more slowly than do those who cohabit; the evidence indicates that they are also more economically advantaged, less likely to have experience family disruption while growing up, and more religious (Sassler and Goldscheider, 2004). Variation in timing to shared living, then, and factors associated with what one learns about a partner prior to moving in together, may further distinguish those who cohabit from those who marry directly.

2.1. Theoretical perspectives on relationship tempo

The theoretical perspectives most appropriate for examining relationship tempo are similar to those used to examine the relationship between childhood living arrangements and subsequent behaviors in young adulthood. Studies of early marriage, non-marital parenting, and divorce are often guided by one or more of the following conceptual frameworks: socialization, the socioeconomic status of the family, and the life course perspective. An understanding of each can shed light on the factors affecting the progression of contemporary relationships.

The socialization perspective posits that the childrearing environment experienced by children has an impact on their subsequent behaviors; parents impart desired values and goals to their offspring, while children observe their parents’ actions. The type of family in which children live, accordingly, sets the stage for subsequent behaviors in adulthood. Those born to young mothers, for example, may view early childbearing as normative (Edin and Kafalas, 2005), whereas young adults with highly educated parents may expect (and be expected) to attend and complete college prior to becoming parents (Lareau, 2003). Children from single-parent or stepparent families may also receive less encouragement to achieve academically than those in intact married-parent families (Astone and McLanahan, 1991; McLanahan and Sandefur, 1994). Family structure experienced by children while growing up has, in particular, been considered a primary venue of socialization with regards to sexual behavior and union formation. Children who experience the disruption of parents’ marriages or who grow up with unmarried parents are frequently exposed to the new romantic partnerships parents form (Goldscheider and Sassler, 2006; Graefe and Lichter, 2007; Sassler et al., 2009a), and express more liberal attitudes about sexuality, cohabitation, and divorce than those whose parents remained married throughout their childhood (Clarkberg et al., 1995; Kapinus, 2004). Drawing from the socialization perspective leads to the expectation that those with younger, less educated mothers, and who experienced family disruption would experience more rapid transitions to sexual involvement and shared living than those whose mothers were more mature or better educated when they were born, or who grew up in stable married-parent households.

The social class perspective suggests that those growing up in less advantaged families will experience more expedited entrance into sexual involvement and coresidence, and would be more likely to cohabit than marry, as a result of economic exigency, or a shortage of resources that might enable other pursuits (such as schooling). Those raised in more advantaged backgrounds are often encouraged to focus on the acquisition of skills necessary for middle class life, which often means discouraging romantic unions at young ages (Lareau, 2003). Mothers who gave birth at young ages, in contrast, are more likely to have come from less economically advantaged families and to have dropped out of school (Cooksey et al., 2002; Pearson et al., 2006). Such mothers are also more likely to experience marital disruption (Martin, 2006). Of course, family social class can change. A great deal of evidence indicates that experiencing parental divorce or union disruption while growing up adversely affects the socioeconomic status of children (Amato, 2000; Avellar and Smock, 2005), frequently resulting in residential moves and school changes. Young adults who experience parental divorce leave the home at younger ages (Aquilino, 1991; Cherlin et al., 1995; Goldscheider and Goldscheider, 1998; Teachman, 2003), are more likely to turn to romantic partners (rather than parents) for emotional and economic support (Cavanagh et al., 2008; Cooksey et al., 2002; Pearson et al., 2006), and form cohabiting unions earlier than do the more economically privileged (Clarkberg, 1999; Sassler and Goldscheider, 2004).

While family of origin characteristics play important roles in establishing behaviors in adolescence and into young adulthood, the life course perspective suggests the need to pay attention to how processes are sequenced and interrelated (Elder, 1988). Bearing a child as a teenager, for example, may weigh heavily on subsequent patterns of union formation in adulthood (Lichter et al., 2010), while experiencing marital disruption affects the type of relationships formed subsequently (Goldscheider and Sassler, 2006; Teachman, 2008). A focus on a life course perspective would emphasize behaviors in adulthood as key mechanisms shaping subsequent relationship processes. Respondents who are parents or were previously partnered, for example, might be expected to pursue new relationships differently than those without such experiences. Children have long been seen as a deterrent to repartnering (Goldscheider et al., 2009), though parenting status exerts different effects for men and women (Carlson and Furstenberg, 2006; Goldscheider and Sassler, 2006; Teachman, 2008). The evidence with regard to life course experiences suggests that parents will proceed more slowly into new relationships. Divorced adults and those who have cohabited in the past, in contrast, should enter new relationships more rapidly than those who have never before lived with a partner, and the literature suggests that these unions are more likely to be cohabitations (Goldscheider and Sassler, 2006; Lichter and Qian, 2008).

Adjudicating between these perspectives is challenging, as there can be more than one theoretical explanation for observed behaviors. There is substantial research evidence that children who have experienced parental divorce have an earlier sexual debut (Cooksey et al., 2002; Pearson et al., 2006), form cohabiting unions earlier and more frequently (Clarkberg, 1999; Sassler and Goldscheider, 2004) and are more likely to experience early marriage and early divorce than their counterparts who grow up in married-parent families (Goldscheider and Goldscheider, 1993; Teachman, 2002, 2003). But whether these behaviors are the result of socialization, or ensue from experiencing economic strain and therefore accelerated transitions to adulthood, is difficult to determine. Studies of adult relationships also generally pay scant attention to how prior relationships influence subsequent ones (Sassler, 2010).

Furthermore, extant theorizing pays inadequate attention to whether and how socialization, social class, and life course progression differentially affect the behaviors of women and men (see also Ryan et al., 2009). Recent studies of adolescents report that boys experience more adverse effects than girls when they do not live in married-parent households (Cavanagh et al., 2008), though much of the research on those growing up in the 1970s and 1980s found that family structural changes exerted stronger effects for daughters than for sons (Glenn and Kramer, 1987; Teachman, 2002; Thornton, 1991).³ Studies focusing on union types increasingly prevalent in the late 20th century have found that women who experienced family disruption during childhood were more likely to form early cohabitations and marriages than their male counterparts (Ryan et al., 2009). Gender norms may differentiate the factors shaping men’s and women’s partnering behaviors; there are vastly different expectations regarding how men and women are expected to act within relationships (Baumeister and Vohs, 2004; West and Zimmerman, 1987). Men with few resources may be hindered in pursuing relationships to a greater extent than similar women might be, as men have long been expected not only to initiate dates but also to pay for them (Laner and Ventrone, 1998; Ross and Davis, 1996). Alternately, exchange perspectives have long held that women rely on non-monetary assets – attractiveness, for example, as well as sexual allure – in the partner market to a greater extent than men (Baumeister and Vohs, 2004). Furthermore, women who have experienced family disruption may require the social and economic support of romantic partners more than do girls from intact families, or males (regardless of their family structure experiences) given gender disparity in earnings. The cumulative evidence suggests that different factors should influence the tempo of relationship progression among men and women.

3. Data and methods

3.1. Data

Data used for this study come from the Marital and Relationship Survey (MARS) (see Lichter and Carmalt, 2009), a web-based survey administered by Knowledge Networks (KN) to a nationally representative sample compiled using listed and unlisted telephone numbers. Knowledge Networks relies on probability samples of members who are part of a web-enabled panel that covers the online and offline population of the United States. Unlike other internet or web-based surveys, which recruit current web-users who are willing to participate in online surveys, KN provides on-going household panelists with an Internet appliance, Internet access, Web TV, and a cash payment in return for completing the survey. Panelists then receive unique log-in information for accessing surveys online, and are sent emails three to four times a month inviting them to participate in research. Because Internet accessibility was provided, the use of an Internet survey did not exclude members of disadvantaged backgrounds, who are the least likely to own a computer or have access to the internet (Fairlie, 2004).

The MARS questionnaire was administered to low- to moderate-income married or cohabiting individuals during March and April of 2006, and took approximately 35–40 minutes to complete. The sample was restricted to couples with minor children present in the home, with household incomes of $50,000 or less, and where the woman was less than 45 years old. This is the population of greatest interest to those forming programs supported by the Healthy Marriage Initiative. The MARS response rate for the target population was 80.3% and item non-response was low (less than 4%). The sample is comprised of 563 women and 532 men.⁴

3.2. Dependent variables

Two measures of relationship progression serve as our dependent variables. The first involves the amount of time from the beginning of the relationship to first sexual involvement with the current partner. Respondents were asked: “How long did you and your [spouse/partner] date prior to having sex for the first time?” and were given six options to choose from: less than a week; more than a week but less than a month; 1 or 2 months; 3–6 months; more than 6 months but less than 1 year; and 1 year or more. For respondents who indicated that they are married and waited until marriage to have sex, the missing value is replaced with the category that corresponded to the time from the start of the relationship to marriage.⁵

The second measure of relationship tempo is the amount of time from the beginning of the relationship to when individuals reported moving in with their partner. Respondents that were currently cohabiting or who had cohabited prior to marriage were asked: “When did you and your current partner start dating?” and “When did you and your current partner start living together?” The answers, given in terms of month and year, were used to calculate the number of months from the onset of dating to movement into shared living with the current partner. For those who did not live with their current partner before marriage we calculate the period between when they began dating and their marriage anniversary. Those missing the date of either living together or marriage (n = 56) were set to the mean value of their union status group (e.g., married directly). Because small numbers of respondents reported being romantically involved for very long periods of time (over a decade, for example) prior to entering into shared living, we top-coded the duration to shared living at 48 months, in order to reduce bias introduced by the cases with extremely long durations (n = 37 for women; n = 36 for men).

Our third dependent variable – the type of coresidential union first entered – utilizes respondents’ answers to the question of whether they had lived with their partner prior to marriage, and if they were currently cohabiting. Current cohabitors were grouped with those who entered into shared living with their partner prior to marriage. Those who married without first living with their spouse serve as the reference group.

3.3. Key independent variables

The MARS data includes information about a range of family background characteristics, personal attributes, and measures of relationship history that are utilized to construct independent variables. Family background characteristics include the age of the respondent’s mother when she bore the respondent, maternal levels of educational attainment, and whether the mother was married to the respondents’ father at the birth and remained in an intact union with him. For maternal age at birth, we designate those who began childbearing at age 21 or earlier as young mothers. Maternal educational attainment was disaggregated into those who had not completed their high school degree, those who received their high school diploma, and those who pursued some post-secondary schooling. Our indicator of family structure as a child distinguishes between children whose parents were married at theirbirth and who remained married throughout their childhood, and those thatexperienced an alternative family form, either because parents subsequently divorced or were not married when they were born.⁶

Individual and relationship history variables include race/ethnicity, age at the start of the current relationship, whether respondents indicated they were a parent or had a prior coresidential union at the beginning of their current relationship, and where and when respondents met. Race and ethnic origin were based on self-identified survey responses, and consisted of non-Hispanic White, non-Hispanic Black or African American (Black), Hispanic/Latino, and small numbers of others (“Other, non-Hispanic, and “2 or more races, non-Hispanic”).⁷ We assess the context of the relationship by examining the respondents’ age at the start of the union, via the questions on when the participant and their current partner began dating and their birth date. We attempt to isolate the effects of relationships that began during high school from those formed later in life, by establishing age 19 or less as “young” romances. We also measure whether respondents had a child prior to the formation of the current relationship, and if they had previously lived with a partner (either marital or cohabiting).

A unique aspect of the MARS data is its inclusion of a question on where respondents met their current romantic partner. Although there were initially 11 categories (including other), they were grouped into six possibilities: (1) through family and friends (the omitted category); (2) at work or a work-related function; (3) at a place of worship; (4) at a club or a bar; (5) At an online dating or internet chat room, through a personal ad in a magazine or newspaper, at a singles’ group, or through a special interest group (e.g., sports club or volunteer activities), or when walking, shopping, or other public places; and (6) at school.⁸ We also construct categories for how long the relationship has endured. Finally, an indicator of whether the couple initially cohabited or married directly is constructed.

Descriptive results for the independent variables used in the analyses are presented in Table 1. About a third of the women respondents and close to 40% of the men were born to young mothers, and over a quarter had mothers who had not completed high school. Less than 60% of the men and women in our sample grew up in married-parent families. Women were significantly more likely than men to have begun their current relationship when they were aged 19 or younger, and to have already been parents at the start of the relationship. Men were somewhat more likely to have already experienced a prior coresidential union, though this difference is not statistically significant.

Table 1.

Means and standard errors of variables in analyses.

Independent variables	All women		All men
	Coefficient	SE	Coefficient	SE
Family background characteristics
Mother’s age at respondent’s birth
Young mother (≤21)	0.331	0.025	0.394	0.031
Mother’s educational attainment
Less than high school	0.257	0.023	0.263	0.028
High school	0.426	0.026	0.518	0.032
More than high school	0.317	0.024	0.219	0.026
Family structure as child
Intact married-parent family	0.571	0.026	0.590	0.031
Individual/relationship measures
Race
Non-Hispanic White	0.888	0.017	0.873	0.021
Black1	0.046	0.011	0.060	0.015
Hispanic	0.066	0.013	0.068	0.016
Age at relationship start: Young (≤19)	0.344	0.025	0.155	0.023
Child from previous relationship	0.290	0.024	0.239	0.027
Previous coresidential relationship	0.388	0.026	0.434	0.031
Where met partner
Family or friends	0.372	0.025	0.351	0.030
At work or work-related function	0.172	0.020	0.195	0.025
At school	0.128	0.018	0.108	0.020
Place of worship	0.025	0.008	0.020	0.009
Club or bar	0.074	0.014	0.096	0.019
Online dating/Internet chat room/personal ad/singles group/special interest group/walking, shopping, other public places/other	0.161	0.019	0.163	0.023
When met (time period)
Within the past 4 years	0.243	0.022	0.247	0.027
5–7 years ago	0.156	0.019	0.155	0.023
7–10 years ago	0.197	0.021	0.171	0.024
10–15 years ago	0.197	0.021	0.219	0.026
15 or more years ago	0.208	0.021	0.207	0.026
Current union status
Married directly	0.284	0.024	0.311	0.029
Cohabited prior to marriage	0.587	0.026	0.578	0.031
Currently cohabiting	0.128	0.018	0.112	0.020
N	563		532

Note. Underlining denotes significant difference between women and men at the .05 level.

The largest proportion of men and women in our sample reported meeting their current love interest through a known intermediary such as a family member or friend (mentioned by over a third of respondents), while the second most popular meeting place was at work. Over 10% of our respondents also mentioned having met their partner at school, while between seven and 10% of respondents met at a club or a bar. Less than 3% of men and women met their current partner at a place of worship. As for when respondents met and how long these relationships have endured, nearly one-quarter of these unions were relatively recent, having been formed within the past 4 years. Divorce rates peak during the fourth year (for both first marriages and remarriages) (Goldstein, 1999), resulting in our uneven distribution of couples across the marital life course; less than 16% of women and men had been together for 5–7 years. Among the women respondents, about one-fifth have been together between seven to 10 years, 10–15 years, or 15 or more years, respectively. Our sample, then, contains sizable proportions of more recent relationships (nearly 40% are less than 7 years in duration) as well as unions that have endured for 10 or more years. As for couples’ union status, 12.8% of the women and 11.2% of the individual men were currently cohabiting at the time of the survey. Of the remainder, who were married, the vast majority had initially cohabited.

3.4. Analytic strategy

We estimate multivariate OLS regression models to determine the relationship between progression tempo and other individual partner characteristics. Models are run separately by sex, as predictors of relationship tempo may differ for men and women.⁹ Because the measure of duration to first sexual involvement is categorical rather than continuous, we initially ran models using ordered logit analysis, in addition to OLS regression. The results were quite similar across models; coefficients were nearly identical in terms of their direction, size, and significance. For ease of interpretation we therefore report the OLS results. The second analysis, of duration to coresidence, also utilizes OLS regression. For both of these analyses, negative coefficients indicate a faster progression into either sexual involvement or coresidence, relative to the reference group, with positive coefficients indicating a more delayed tempo. The third analysis, which explores the type of union first entered (cohabitation or marriage) uses logistic regression to model the probability of initially forming a cohabiting union relative to marrying directly. Unweighted logit coefficients show changes in the log odds of experiencing a particular union type; we present odds ratios for easier interpretation. Numbers lower than 1 indicate a reduced odds of cohabiting relative to direct marriage. All regression models were estimated using multiple imputed data created from the imputation chained equations (ICE) program for STATA (Royston, 2006) to maintain maximum sample sizes for all variables utilized in estimation.

4. Results

4.1. Descriptive results

Our descriptive results reveal that relationships progressed quickly into sexual involvement. Over one-fifth of both women and men reported they became sexually involved with their current partner within the first week of meeting (Table 2, Panel A). On the other end of the spectrum, only about one in six respondents indicated that they deferred sexual involvement until they had been dating for 6 months or longer. Entrance into shared living takes longer (Table 2, Panel B), but for a sizable proportion still occurs rapidly. About a third of both men and women moved in with their partner within 6 months of their relationship’s start. Over half had entered into shared living within a year. Nonetheless, over one-quarter of women and men indicated that they it took 2 or more years before they began living with their partner.

Table 2.

Distribution of duration to sexual involvement and coresidence, by sex.

	All women		All men
	Coefficient	SE*	Coefficient	SE*
(A) Duration to sexual involvement
Less than a week	0.216	0.022	0.243	0.027
1 week to 1 month	0.189	0.020	0.207	0.026
1–2 months	0.246	0.023	0.215	0.026
3–6 months	0.186	0.020	0.191	0.025
6 months to a year	0.101	0.016	0.092	0.018
1 year or more	0.063	0.013	0.052	0.014
(B) Duration to coresidence
Less than 6 months	0.363	0.025	0.327	0.030
6–11 months	0.180	0.020	0.199	0.025
12–23 months	0.202	0.021	0.211	0.026
24 months or more	0.254	0.023	0.263	0.028
Mean (in months)	15.209	0.788	15.208	0.925
N	563		532

^*Standard errors.

These aggregate figures mask considerable disparity, in both entrance into sexual involvement and shared living, by union type. A closer look at the tempo of relationship progression reveals sharp differences between women who were currently cohabiting, those who lived with their partner prior to getting married, and those who married directly (Fig. 1). Those who married directly, for example, became sexually involved at a far more moderate pace than did their counterparts who cohabited prior to marriage or were currently cohabiting; over one-third reported being romantically involved for over a year, with another 19.4% reporting deferring sexual intimacy for over 6 months. Among those who initially cohabited, in contrast, over a quarter became sexually involved within the first week. The most rapid transition to sexual involvement occurred for those who were currently cohabiting. Similar tempo differences can be seen upon examining the duration to shared living. Nearly half of both groups of cohabitors reported entering into shared living within 6 months of the relationship’s start (see Fig. 2), consistent with Sassler (2004) qualitative study of the relationship progression of cohabitors. In contrast, only 10.6% of those who married directly began living with their partner within that time period. In fact, only about one-quarter of current cohabitors, and a third of those who cohabited prior to marriage, deferred entrance into shared living for over a year, in contrast to over 70% of those who married directly.

Fig. 1.

Women’s reports of duration to sexual involvement, by union status.

Fig. 2.

Women’s reports of duration to coresidence, by union status.

Where couples met their partner also varied widely by union type (Fig. 3). Cohabiting women and those who had cohabited prior to marriage more often reported meeting their partner through family or friends, or at work, than those who married directly. Respondents who married directly are more than twice as likely as those who cohabited before marriage to report meeting at school (21.8% versus 9.3%) with another 15.6% said they met at a place of worship. Current cohabitors were also far more likely than the married, even those who first cohabited, to have met at a club or bar, or at some singles venue or public setting. These descriptive results suggest that those who initially form cohabiting unions may differ in important ways from those who wed directly, with potential affects for subsequent relationship quality and stability.

Fig. 3.

Women’s reports of where met current partner, by union status.

4.2. Multivariate results

4.2.1. Duration to sexual involvement

We turn now to our multivariate results, to identify factors associated with relationship progression and the type of union initially formed. Results of OLS regression of duration to sexual involvement and coresidence with current partners are presented in Table 3. In the reduced model on entrance into sexual involvement (Model A), family background characteristics exert different effects for women and men, suggesting the importance of sex-specific analyses. For men, being born to older mothers is associated with slower entrance into sexual involvement, compared to those whose mothers were young at theirbirth; no such delaying effect is observed for women. There is also evidence of gender variance in the impact of maternal educational attainment. Whereas men whose mothers are high school drop-outs experience an expedited transition to sexual involvement relative to men whose mothers completed high school (B = −.33, p ≤ 10), for women having a mother with some post-secondary schooling is associated with a significantly slower entrance into sexual involvement. Women whose mothers did not complete high school (a sizable proportion of this sample) do not differ significantly from their counterparts whose mothers had a high school degree. As hypothesized, those who experienced family instability while growing up become sexually involved significantly more quickly than did those whose parents remained married throughout their childhood. This effect is evident for both women and men.

Table 3.

OLS regressions on relationship progression for women and men.

Independent variables	Duration to sexual involvementa				Duration to coresidence				Cohabit (versus Marriage)b
	Model A		Model B		Model C		Model D		Model E
	Women	Men	Women	Men	Women	Men	Women	Men	Women	Men
Family background characteristics
Maternal age at first birth
Older (≥22)	−0.02	0.41*	−0.07	0.24	−0.42	1.58	−0.71	1.52	0.729	0.994
Mother’s educational level (High school diploma)
Less than high school	−0.01	−0.33+	0.08	−0.23	−0.92	−3.09*	−0.81	−2.05	1.185	1.921*
More than high school	0.38*	0.18	0.30+	0.01	2.42+	−1.07	1.51	−1.50	0.639	0.818
Family structure as child (intact married-parent family)
Parents divorced/never married	−0.55**	−0.38*	−0.35*	−0.21	−4.20**	−2.25+	−2.44+	−1.35	2.675***	1.692+
Individual and relationship attributes
Race (non-Hispanic White)
Black			0.29	0.12	0.00	5.02*	−0.57	5.38*	0.627	1.433
Hispanic			0.36	0.42	−1.34	1.38	−2.09	1.42	0.733	1.172
Respondent ≤19 at relationship start			−.018	−0.11	5.35***	7.66***	6.21***	8.85***	1.461	2.132*
Child from previous relationship			−0.24	−0.01	−2.43	−1.54	−1.94	−1.66	1.436	0.895
Previous coresidential relationship			−0.65**	−0.97***	−0.19	−2.84*	2.15	−0.29	2.430**	2.594***
Where met partner (family/friends)
At work			0.15	0.13	3.52*	4.58**	3.34*	5.05**	0.892	1.582
At school			1.17***	0.92***	10.88***	9.13***	8.53***	6.83***	0.602	0.443*
At place of worship			1.76***	1.78***	5.64*	3.99	0.30	−0.99	0.044***	0.051***
At a club or bar			−0.22	−0.05	−3.30	−1.33	−2.60	−0.48	1.303	2.437
Online dating/singles group/personals/public place			0.33+	0.20	1.72	2.76+	1.91	3.01+	1.558	1.379
When met (within past four years)
5–7 years ago			0.30	0.05	−1.83	−3.38+	−3.11+	−4.47+	0.431*	0.468+
7–10 years ago			0.47*	0.15	−0.31	1.71¥	−1.00	1.19¥	0.797	0.819
10–15 years ago			0.42+	0.24	1.10	1.32¥	−1.006	−0.465¥	0.307**	0.3298**
15 or more years ago			0.64**	0.42+	0.55	2.04¥	−1.84	0.14¥	0.288**	0.326**
Entered cohabiting union (married directly)			NA	NA			−10.47***	−9.57***	NA	NA
Duration to sexual involvement (less than 1 week)
1 week to 1 month							NA	NA	0.324*	0.186**
1–2 months							NA	NA	0.389*	0.132****
3–6 months							NA	NA	0.234**	0.093***
6 months to a year							NA	NA	0.097***	0.098**
1 year or more							NA	NA	0.031***	0.023***
Constant	2.50***	2.28	2.03***	2.20***	13.27***	12.78***	20.16***	18.35***	2.354***	2.419***
log_likelihood	−1079.0	−1026.6	−1013.1	−959.9	−2253.2	−2113.8	−2222.1	−2087.7	−238.5	−232.2
R-squared	4.0%	4.7%	24.1%	25.7%	20.1%	21.1%	28.5%	28.5%	36.7%	35.1%
N	563	532	563	532	563	532	563	532	563	532

^*p ≤ .05.

^**p ≤ .01.

^***p ≤ .001.

⁺p ≤ .10.

^aMeasured in categories.

^bOdds ratios.

^¥Significant difference from 5 to 7 years ago (p ≤ .05).

Incorporating measures of individual and relationship attributes has a sizable impact on the model fit (Model B). The inclusion of these measures reduces the impact of most family background variables to non-significance. Once relationship trajectories are established, they may assume greater weight than indicators of family social class in determining the tempo of subsequent relationships. Only growing up in an alternative family attains conventional levels of significance for women, speeding their entrance into sexual involvement.

Among the relationship attributes examined, those with previous coresidential unions entered into sexual involvement with their current partner at a significantly more rapid tempo than did those with no prior coresidential relationships. Where respondents met their partners also emerged as a salient predictor of relationship tempo. Those who met at school or a place of worship became sexually involved at a much slower pace than did respondents who were introduced to their current partner via family or friends.¹⁰ We find no significant difference in the tempo to sexual involvement among those who met at a club or bar, and those meeting at work, from each other (results not shown) or relative to those meeting through friends and family. Of note is that women who met their partner through a singles group, internet dating site, or in a public place progressed into sexual involvement more slowly than those introduced by family and friends, though this effect is only weakly significant (p ≤ .10).

Finally, the inclusion of measures of when individuals first became romantically involved provides support for popular perceptions that current relationships progress more rapidly than they used to. Compared to women whose relationships began within the past 4 years, those whose relationships were initiated seven to 10 years before experienced significantly slower entrance into sexual involvement, as did those whose romances began 10–15 years ago, and women who had been involved with their current partner for over 15 years. Among men there is less evidence of an accelerated progression into sexual involvement, though the coefficients are in the right direction. Changes in tempo to sexual involvement may be associated with shifts in the types of unions initially formed, as those who entered into cohabiting unions, even if they subsequently transitioned to marriage, became sexually involved far more rapidly than did those who married directly (results not shown).

4.2.2. Duration to coresidence

Results of the examination of factors associated with the tempo of progression into coresidence with current partners are presented in Models C and D of Table 3. Accounting for individual and relationship attributes has little effect on the impact of family background characteristics for progression into shared living; we therefore do not present models limited to family background characteristics. Results from Model C suggest that maternal education shapes coresidence differently for sons and daughters. Men with the least educated mothers enter into shared living over 3 months faster than do those whose mothers have completed high school, while women with more educated mothers progress more slowly than their counterparts whose mothers’ highest level of schooling was a high school diploma. Our results are also consistent with published literature indicating that entrance into coresidential living occurs more rapidly for those who have experienced alternative family structures while growing up (Teachman, 2003); women whose parents divorced or were never married began living with their current romantic partner over 4 months sooner than did those from intact, married-parent families. The effect for men was smaller and only weakly significant.

Many of the individual and relationship attribute measures are associated with the timing of entry into shared living. Black men entered into shared living at a significantly slower pace than their white counterparts, consistent with other studies (Raley, 1996). Respondents that were under 20 at the start of their relationship took longer to move in with their partners than those who met at older ages – over 5 months later for women, and nearly 8 months later for men. Men with prior coresidential experience, in contrast, moved in with their new partner considerably more quickly than their counterparts who had not previously lived with a partner.

Our findings also suggest respondents may exercise greater caution in advancing relationships formed in particular meeting locales. Men and women who met at work, for example, took on average 3–5 months longer to enter into shared living than did those who met through family and friends. Delays are especially long among those who met their current partner in a school setting, with women taking nearly 11 months longer to move in than did women who were introduced through family and friends, and male students taking over 9 months longer. The impact of meeting at church is only significant for women, deterring entrance into shared living by nearly 6 months.

As for when respondents formed their relationship, the results do not indicate that relationships formed more recently progressed into shared living any more rapidly than those entered into in the more distant past. Though men who had began their relationships within the past 4 years also did not enter shared living any more rapidly than men in relationships of older vintage, those who began their relationships 5–7 years before did begin living with their partner significantly more quickly than did men in the next three categories (7–10 years, 10–15 years, and 15 or more years before).

Our final model for the analysis of entrance into coresidence (Model D) includes an indicator of the type of living arrangement formed – cohabitation versus marriage. Incorporating this one measure increases the amount of explained variance considerably, though some indicators, such as those for family background characteristics, are no longer significant at conventional levels. Both women and men who initially entered into cohabiting unions began living with partners much more quickly in their relationship than did those who married directly – 10 months earlier, on average.

4.2.3. Type of union entered

Our final analysis explores the factors associated with entrance into cohabiting unions relative to marital ones. We focus on the initial union entered, even though the majority of those who entered cohabiting unions subsequently wed that partner. Results from Model E in Table 3 indicate that persons with less advantaged family backgrounds are at greater relative odds of forming a cohabiting union over a marriage. Men whose mothers were high school drop-outs, for example, were nearly twice as likely as men whose mother had completed high school to initially form a cohabiting union. Both women and men who grew up in alternative families were far more likely to cohabit than to marry directly. These effects remain even after controlling for measures of relationship attributes.

As expected, men and women with prior coresidential experience had odds over two times as great of entering into informal (cohabiting) unions as did their counterparts with no prior union experience. In addition to being associated with slower entrance into sexual involvement and shared living, meeting a partner at school or a place of worship also greatly reduced the likelihood of entering into a cohabiting union over a marital one, at least for men. As for how relationship length and union type are associated, the most recently formed relationships are substantially more likely to be cohabiting unions that those formed longer ago. Among couples that had met 10–15 years ago, for example, women and men were less than one-third as likely to enter a cohabiting over a marital union as were couples that had been together for less than 5 years. Finally, the evidence indicates that the longer the wait to sexual involvement, the lower the odds that one entered into a cohabiting union versus a marriage. Both women and men who deferred sexual involvement beyond the first week of romantic involvement are less than one-third as likely to have entered cohabiting unions as those who became sexually intimate within the first week of meeting.

5. Discussion and conclusions

The need clearly exists to pay more scholarly attention to how rapidly American relationships unfold (Cherlin, 2009). Substantial proportions of women and men in low- to moderate-income families moved quite quickly into sexual involvement and shared living. Over one-fifth of the women and men in this sample reported becoming sexually intimate within the first week of meeting their partner, and about a third had moved in together within 6 months. Furthermore, those who became sexually involved the most rapidly demonstrated the greatest odds of initially entering into cohabiting unions. Consistent with both the socialization and social class perspectives, spending time during childhood in alternative families was associated with a more rapid entrance into sexual involvement, coresidence, and cohabitation. We cannot determine whether this behavior was modeled on what was learned in the family home, or it if resulted from a desire to escape the family environment, to achieve adulthood, and the lesser ability of economically disadvantaged parents to subsidize other living arrangements. Our findings suggest that the processes of relationship progression may be yet another mechanism contributing to widening social inequality, as today’s children are increasingly born to unmarried parents whose unions are often quite fragile (Carlson et al., 2004; Graefe and Lichter, 2007).

Our results also provide evidence that the pace at which contemporary relationships are formed has accelerated. Among the women in our sample, more recently formed relationships progressed into sexual involvement at a quicker pace than did unions that had begun 7 or more years before, whereas men in more recent unions reported entering into shared living more rapidly than their counterparts who had begun their relationship longer ago. These estimates may be conservative ones, if rapid relationship progression is associated with union dissolution. The couples whose relationships began less than 7 years ago may also not be comparable to couples that are celebrating their tenth or higher anniversary. Longer term couples are selective; they have weathered the tests of time and remained together. Our sample probably over-represents those couples with better quality relationships, as well as more stable ones of longer duration. It does contain a disproportionate share of couples that cohabited prior to marriage (about two-thirds of our married sample), and the results also reveal a significantly more rapid entrance into shared living among those who cohabited before the wedding. Whether premarital sexual entanglements (Glenn, 2002) explain the negative relationship between premarital cohabitation and subsequent marital quality and stability is unclear (Rhoades et al., 2009).

These findings add weight to the idea that economic disadvantage plays a particularly salient role in women’s relationship and residential decision-making. Such actions may then exert important effects on subsequent life trajectories, consistent with life course theory. As already noted, women from more disadvantaged backgrounds experienced more expedited entrance into sexual relationships and shared living than did other women. But forming cohabiting unions expeditiously increases women’s likelihood of unplanned pregnancy (Fu et al., 1999; Sassler et al., 2009b), which may result in poorer quality matches and relationship churning (Cherlin, 2009; Lichter and Qian, 2008). While markers of disadvantage appear to expedite relationship progression, our study also reveals how relationship context can help delay the advancement of sexual and coresidential unions. Those who met at school or a house of worship, for example, took more time to become sexually involved, and also entered into shared living more slowly, as did respondents who met their partner in their teens. Additional research is necessary to determine if relationship quality differs among those meeting in different venues, and whether that is associated with tempo or due to differential selection into particular meeting locales (such as school or singles groups or bars) (Paik, 2010). This is particularly important as new relationship forms and ways of meeting become more prevalent and accepted.

Our study is not without limitations. The results are based on a relatively small sample of low- to moderate-income respondents with minor children. Furthermore, the data are cross-sectional rather than prospective, and include only relationships that have endured. Our key dependent variables, duration to sexual involvement and coresidence, are based on retrospective recall, and results from the sub-group of couples reveals that partners are not always in agreement regarding these durations. We cannot ascertain the extent to which the women and men in our sample interpreted the meaning of sexual involvement differently; nor do we adjudicate between their responses and determine which was more accurate. An additional challenge is our inability to include many personal attributes, as relationships were often formed prior to the completion of schooling or attainment of stable employment; such attributes may be associated both with the tempo of relationship progression and the stability of unions formed. While longitudinal studies can be used to explore the association between individual attributes, union formation and stability, however, such studies lack information on the duration from relationship start to sexual involvement or coresidence found in the MARS data.

The social psychological literature suggests that rapid, event-driven relationships are less stable, more volatile, and of lower quality. Social scientists have lately been advocating that individuals “take their time” in forming new attachments (cf., Cherlin, 2009). Better understanding the factors that contribute to relationship progression and stability is particularly important for low-income couples, who are at the greatest risk of union disruption. In particular, additional research focused on the very earliest stages of romantic relationships, that explores why relationships progress at the rates they do, and whether relationship advancement differs by stage in the life course, social class, or race is necessary. The qualitative evidence suggests that housing need, accidental pregnancies, and economic uncertainly expedite entrance into shared living, often among very recently formed relationships (Reed, 2006; Sassler, 2004). Future data collections, particularly longitudinal surveys, must begin to include questions that ask about the very early stages of relationships, including where respondents meet partners, when dating relationships become sexually intimate, if and when they become sexually exclusive, as well as whether partners had discussed future plans (for marriage, children, etc.) or had gotten engaged prior to entering into cohabiting unions. To return to the quote from Jane Austen’s Sense and Sensibility, for a sizable proportion of the respondents in our sample, 7 days was enough to determine that couples were sexually attracted. The jury is still out regarding whether those who took longer to ascertain whether a romantic relationship would become a sexual one ended up with greater levels of intimacy or better quality, more supportive and stable relationships.