Introduction

Marriage as an institution has been transformed by rising rates of pre-marital cohabitation. At the same time, the conceptual boundaries between cohabitation and marriage have become increasingly ambiguous (Kiernan 2004; Brown and Manning 2009), especially as a context for childbearing and childrearing. Indeed, more than 1-in-4 first births in the United States today are to cohabiting women (Copen et al. 2013) and growing shares of all first unions are now seemingly motivated or prompted by a non-marital conception, often within cohabiting unions (Gibson-Davis and Rackin 2014; Hayford et al. 2014). In this paper, we ask a straightforward but heretofore unanswered question: Do cohabiting unions and marriages last if coresidence begins as the result of a pregnancy? Whether unions formed after a conception (heretofore called post-conception unions) lead to stable and lasting unions is far from clear. Most previous studies have examined only whether divorce rates are elevated by a pre-marital cohabitation, but typically overlook factors that precipitated the cohabitation, for example, an unintended pregnancy.

This paper has a singular goal: to document comparative patterns of union stability among women whose first births occurred in a coresidential union—either cohabitation or marriage—based on whether the coresidence began before or after the conception. Unlike most previous research on the stability and dissolution of cohabiting unions (e.g., Guzzo 2014; Manlove et al. 2012; Manning and Cohen 2012), we identify and track the relationship trajectories of women who became pregnant before moving in their partners. These include both marital unions that began with a non-marital conception, either as singles or cohabiting couples (i.e., shotgun marriages) and cohabiting unions that began with a non-coresidential conception (i.e., shotgun cohabitation).Footnote 1 This is a timely topic. Lichter et al. (2014) recently reported that nearly 1-in-5 cohabiting unions began only after conception. Non-marital pregnancy is now a significant precursor to cohabitation before childbirth (18 %), exceeding transitions to marriage (5 %) by a factor of over three. Gibson-Davis and Rackin (2014) found that the percentage of post-conception cohabitations has supplanted post-conception marriages over the past two decades. Yet, we know virtually nothing about the trajectories—the subsequent stability—of these (seemingly) pregnancy-induced unions. Do they lead to stable unions?

Here we examine the trajectories of all unions following a first birth. We distinguish between coresidential couples whose first birth (1) took place in a cohabiting union formed after the conception (i.e., a shotgun cohabitation); (2) took place in a cohabiting union formed prior to the conception; (3) took place in a marital union formed after the conception (i.e., shotgun marriage); and (4) took place in a marital union formed prior to the conception. Following Rackin and Gibson-Davis (2012) and Lichter et al. (2014), we label these, respectively, post-conception cohabitation, pre-conception cohabitation, post-conception marriage, and pre-conception marriage. Using data from the National Survey of Family Growth (2006–2010 and 2011–2013), we first examine marriage and dissolution rates among those cohabiting at the time of first birth, comparing rates between those who were cohabiting before conception and those who began living together only after the conception. We then examine dissolution rates among those married or cohabiting at the time of the first birth, focusing on differences across the four union types: post-conception cohabitations, pre-conception cohabitations, post-conception marriages, and pre-conception marriages. As we argue here, post-conception unions, especially post-conception cohabiting unions, are likely to face elevated rates of union dissolution by virtue of possessing predisposing characteristics (i.e., selection), the lack of preparation and matching among partners, the larger shares exposed to the destabilizing effects of unintended pregnancy, and a shorter period of time available for relationship adjustment without the presence of children.

Transitions to Stable Unions

Fertility among cohabiting couples is becoming an increasingly important context for non-marital childbearing in the United States. Nearly 60 percent of all non-marital births in the United States now occur in cohabiting unions (Lichter et al. 2014); these births account for roughly 22 percent of all first births today, up from 12.4 percent in 2002 (Copen et al. 2013; Martinez et al. 2012). In Europe, even higher percentages of all first births and all first non-marital births now occur within cohabiting unions (Perelli-Harris et al. 2012).Footnote 2 To be sure, whether cohabitation undermines traditional marriage is a question fraught with controversy. The answer depends, at least in part, on whether cohabitation leads to a stable union, one that provides a healthy familial context for emotional and financial security for America’s children (Manning et al. 2014; Sawhill 2014).

Several behavioral mechanisms can be linked to elevated risks of dissolution among newly formed unions, including cohabiting unions, established between conception and birth. Elevated risks of dissolution among post-conception unions arguably reflect (1) the selectivity of post-conception unions vis-à-vis other unions; (2) a greater likelihood of such unions being hastily formed in response to a perceived emergency (in this case, pregnancy) and therefore mismatched; (3) the stress-producing and destabilizing effects of an unplanned pregnancy; and (4) the absence of a period of child-free adjustment in a new cohabiting union or marriage. We consider each of these behavioral mechanisms in turn below.

First, compared with marital unions, cohabiting unions—including post-conception cohabiting unions—may simply be selective of dissolution-prone couples (Manning and Cohen 2012). Selection on predisposing demographic characteristics and marital attitudes and values among cohabitors seemingly account for some but not all of their higher rates of separation and divorce (e.g., Lu et al. 2012). A few studies have shown that the higher separation rates among those who cohabit prior to marriage are entirely due to selection and that once observed and unobserved characteristics are controlled, pre-marital cohabitation reduces the risk of marital separation (Kulu and Boyle 2010; Lillard et al. 1995). One previous study has even shown that couples who enter cohabiting or marital unions at younger ages are more likely to experience divorce than couples who form coresidential unions at later ages (Kuperberg 2014). Of course, many other variables may also select on both post-conception union formation and divorce, including race and ethnicity and socioeconomic status (Edin et al. 2004). Whether selection factors account for the trajectories of stability and dissolution of post-conception unions is unclear.

A second argument—a causal one—is that pregnancy and parenthood disrupt the positive trajectories of ongoing non-marital relationships. A pregnancy may prompt couples to reevaluate their relationships, i.e., whether to stay together or not, or even whether to accelerate the relationship by moving in together or getting married (Lichter et al. 2014). One implication is that hastily formed cohabiting and marital relationships—those precipitated by pregnancies—may be especially prone to disruption. Compared with their married counterparts, newly cohabiting parents lack “enforceable trust” that comes with being legally married (Cherlin 2004; Kuperberg 2012). Partners also may be ill-prepared for a cohabiting relationship or they may be poorly matched, which increases the likelihood of dissolution (Zhang and van Hook 2009). Some couples may rapidly enter into a coresidential relationship rather than make a reasoned, forward-looking decision about the long-term success of their relationships—either in cohabitation or marriage (Sassler 2004; Stanley et al. 2006). In fact, Reed’s (2006) study found that many newly cohabiting parents moved in together only after pregnancy; although many couples expressed some uncertainty about the relationship, they also thought moving in together was best for the child. A straightforward hypothesis is that pre-union pregnancy undermines the long-term stability of the relationship—a pattern that will be reflected in higher disruption rates, both in cohabitation and in marriage (if these cohabiting couples marry). Previous studies of shotgun marriages, although now more uncommon than in the past (England et al. 2012), clearly indicate that marriages that resulted from a pregnancy (i.e., those occurring within 7–9 months of the marriage) have exceptionally high rates of subsequent divorce (for discussion, see Furstenberg 1976). Cohabiting unions resulting from pregnancies are likely to face a similar fate, even if they progress to marriage.

Third, the likelihood of dissolution—in either a cohabiting or marital union—for post-conception unions may depend on whether the pregnancy was intended or not. Most births by single or cohabiting women are unintended (Musick 2002; Sweeney 2010). An intended pregnancy followed immediately by cohabitation (or marriage) suggests that these couples are making simultaneous decisions, i.e., they make the decision to move in together or marry at the same time they decide to have children or to start a family. In this case, a post-conception marriage implies that decisions about cohabitation, childbearing, and marriage are highly interrelated; they are not made in isolation of each other (Steele et al. 2005; Wu and Musick 2008). For these couples, an intended pregnancy connotes commitment to their future together; these couples are thus more likely to proceed to marriage, and these marriages are expected to experience lower divorce rates than the marriages that began with an unintended pregnancy. Guzzo and Hayford (2014), for example, found that intended pregnancies reduced the risk of dissolution, even when compared with cohabiting women who avoided pregnancy and a birth. Although their study did not identify post-conception unions, the results nevertheless clearly buttress our hypothesis that links between a post-conception cohabitation and dissolution will vary by whether the pregnancy was intended or unintended.

Finally, the preceding discussion has emphasized predisposing explanations—those underscoring the period before entry into cohabitation (i.e., selection, preparation and matching, and fertility intentions)—that influence or change the trajectory of the relationship and subsequent patterns of union dissolution. The entry into cohabitation (or marriage) is a time of adjustment. It is a time when couples learn to accommodate each other on a day-to-day basis, make joint purchasing decisions, and meld family and friendship networks (Doss et al. 2009). A fourth explanation of why post-conception unions may be less stable emphasizes that “learning to live together” is cut short by the arrival of a baby shortly after the union is formed. A new baby requires some adjustments in the relationship, which represent potential sources of conflict (Crohan 1996). For example, a new family member dilutes the time and attention partners are able to devote to each other, raises new concerns about work–family balance (i.e., decisions about employment and childcare), and creates additional financial pressures (i.e., another person to feed, clothe, and house) (Edin and Nelson 2013). New stressors raise the prospect of discord and dissolution. Couples who wait until after they marry to have children arguably are less exposed to potential conflict (and dissolution) than couples who have children almost immediately after moving in together. Even for cohabiting couples, waiting to have children until after they have moved in together is expected to “pay off” in lower separation rates than post-conception cohabiting or marital unions.

Current Study

This paper tracks the transitions into and out of unions among couples who have had a first birth together. As such, it contributes to a large and complicated literature devoted to understanding the risk factors associated with union dissolution. Here, unlike previous studies, we highlight the union trajectories of new parents, comparing married and cohabiting couples who began living together only after the conception to those who were coresidential prior to the conception. Indeed, a large literature links cohabitation to marriage and subsequent dissolution (see Manning and Cohen 2012), but rarely considers (1) whether newly cohabiting parents have elevated risks of dissolution vis-à-vis newly married parents; or (2) whether the coresidential union (either cohabitation or marriage) began as the result of a pregnancy. To our knowledge, no previous study has examined the link between post-conception cohabiting and marital unions and subsequent stability.

We have several specific empirical goals. First, we provide baseline estimates of the prevalence of post-conception unions, i.e., cohabiting and marital unions that began only after pregnancy, among different population subgroups (e.g., by family background and race). Second, we ask whether post-conception cohabiting unions last, which we answer by presenting life table survival rates as well as competing risks event history models of transitions into marriage and separation. If cohabiting unions end, do they end in dissolution or marriage? Are there differences based on whether the cohabitation began before or after the conception? Finally, we are interested in the relative differences in separation rates among all cohabiting and marital unions following a first birth. We address this question using life tables as well as discrete-time event history models, estimating relative separation risks following a first birth for the four union types described above: post-conception cohabitations, pre-conception cohabitations, post-conception marriages, and pre-conception marriages. These analyses deal with issues of selection on observable characteristics, but also address the question of whether union stability and transitions are affected by pregnancy intentions and other characteristics that often distinguish cohabiting couples from marital couples (Manlove et al. 2012; Manning 2004). Based on previous studies of cohabitation outcomes (Guzzo 2014; Lichter et al. 2006), we expect that some groups, including those with higher socioeconomic status, stable family backgrounds, and Catholics, will be more likely to transition to marriage rather than cohabitation or dissolution. Post-conception cohabiting unions resulting from an unintended pregnancy are also expected to have the highest dissolution rates, both before and after the birth of the child.

Data and Methods

National Survey of Family Growth, 2006–2010 and 2011–2013

The data for this paper were obtained from the 2006–2010 and 2011–2013 National Survey of Family Growth. These cycles contain a nationally representative household sample of females aged 15–44 years, and provide detailed information on women’s relationship and reproductive histories. For our purposes, we identify all first births that occurred within coresidential unions—either in cohabiting or marital unions. Nationally, first births account for 45 % of all births. About 75 % of all first births occur within cohabiting and marital unions.

By convention, pregnancies that end in miscarriage or abortion are not considered.Footnote 3 We then link the unions that resulted in a first birth with information that predated the union, including when a conception occurred and the start date of the cohabitation or marriage. In doing so, we are able to identify these unions as (1) post-conception cohabiting unions, (2) pre-conception cohabiting unions, (3) post-conception marital unions, and (4) pre-conception marital unions based on when the conception occurred relative to the coresidential start date. This typology is identical to Lichter et al. (2014) and Gibson-Davis and Rackin (2014). We focus here on first births rather than higher-order births because first births typically mark a significant transition in the family lifecycle.

For each respondent in our sample, the NSFG collects information on the start and end dates of all coresidential unions, allowing us to identify transitions between singlehood, cohabitation, and marriage during the period between conception and birth and subsequent union transitions. We use these start and end dates to construct a person-month file to identify whether the respondent was cohabiting or married to her partner at the time of conception. This information is used to construct the four union types described above. For cohabiting unions, we then model the risk of entering marriage or separating as a function of months since the first birth, reflecting our interest in modeling union stability among couples who have had a first birth (although results were quite similar if we modeled the risk of entering marriage or separating as a function of months since cohabitation or conception). For all unions, we then model the risk of separating as a function of months since the first birth, evaluating how separation risks differ across the four union types at first birth described above.

Measurement and Analysis Strategy

Our analyses focus on the relationship outcomes of the aforementioned sample of marital and cohabiting unions. Our interests are twofold. First, we are interested in whether cohabiting unions transition into marriage or dissolve based on whether the cohabitation began before or after the conception. Second, among all unions, we are interested in the likelihood that these unions end (in separation) based on whether they began before or after the conception. We answer these questions first with life table estimates and then conditional on demographic and family background characteristics using discrete-time competing risks models and discrete-time event history models. In particular, using multiple-decrement life tables, we estimate the year-to-year probabilities of marriage and dissolution for both pre-conception and post-conception cohabiting unions, censoring observations at the interview date, marriage date, or separation date, whichever comes first. Then, for both cohabiting and married couples (pre- and post-conception), we provide life table estimates of breaking up (i.e., dissolution defined by separation or divorce). These represent baseline estimates of 5-year survival rates.

Next, we estimate discrete-time competing risks models of relationship transitions among post-conception and pre-conception cohabiting couples, predicting the likelihood of a subsequent union transition (i.e., to marriage or dissolution) compared to staying in the cohabiting relationship. For these competing risks models, we censor observations at the marriage date, separation date, or interview date, whichever comes first. For example, cohabiting couples who marry and then subsequently separate are considered married in this analysis. Finally, we estimate discrete-time event history models predicting the likelihood of separating among all pre-conception and post-conception married and cohabiting unions (with first births).Footnote 4 These models estimate the probabilities of breaking up by months since first birth controlling for union type at first birth, censoring only at the separation date or the interview date. In both analyses, we model the probability of marrying or breaking up as a quadratic function of months since the date of birth, as this structure provided the best model fit compared to a linear function or a Cox model. The quadratic function also accounts for potential non-linearities in the effect of duration since birth on the risk of marrying or separating.

A limitation of the NSFG is that few independent variables are measured at the date of conception (or earlier); most variables are measured at the survey date. Consequently, we are able to include only a small number of strictly exogenous variables in our models of relationship transitions. We include a set of age dummies (<20, 20–24 [reference category], 25–29, and 30 or older) at conception to distinguish the age profile of union transitions. Based on previous studies, we expect older women to be more likely than younger women to enter into post-conception unions (Lichter et al. 2014).Footnote 5 Race/ethnicity is represented with Hispanic, non-Hispanic black, and “other” indicator variables; non-Hispanic whites serve as the reference group. Based on well-known differentials in non-marital fertility rates, we expect that minorities will be less likely to transition to a union between conception and birth. The NSFG collects information on high school graduation dates and college graduation dates, allowing us to control for educational attainment at the time of conception. We expect an educational gradient in our results; more highly educated women are more likely to enter partnered fertility (through pre-conception unions) relative to their less-educated peers. Finally, given the increase in the share of first births to cohabiting couples over the last decade, we allow the likelihood of marriage or dissolution to differ by year of birth, distinguishing between births before 2000 [reference category], 2000–2002, 2003–2005, 2006–2008, and 2009 or later. This set of time dummies accounts for the upward trend (and potential non-linearities) in post-conception unions (see Gibson-Davis and Rackin 2014). Significant period effects, net of other variables in the model, indicate that trends are not simply located in the changing social and demographic composition of mothers (in unions).

Our models also consider family socioeconomic background. Economic status while growing up is measured by whether the respondent’s mother or mother figure had ever attended college. Growing up in an intact family is defined as living with both biological parents at age 14, and not experiencing the separation of biological parents before age 18, until the interview, or until they began living on their own (if before age 18). We also control for reported Catholic upbringing (i.e., “In what religion were you raised, if any?”). Finally, we consider whether the births were intended or unintended. We distinguished between unwanted pregnancies (i.e., at the time of pregnancy, the respondent did not want any children) and mistimed pregnancies (i.e., they came earlier than intended).Footnote 6 We expect that trajectories of stability and dissolution will vary by the intendedness of the birth, and that differences in intentions across types of unions contribute to differences in union stability.

Descriptive statistics for all of the independent and dependent variables are provided in Table 1. All values are weighted by the person weights provided by the NSFG, while we present unweighted counts of the number of women in each category. We also show variation in the distributions of sample characteristics for each category in our fourfold union typology. Not surprisingly, women in post-conception unions are more disadvantaged than average. They are overrepresented among those who did not live with both parents at age 14, had less-educated parents, and were less educated themselves, were minority, were younger at the time of conception, and had unintended births.

Table 1 Mean characteristics of sample by union type at first birth, women with first births in a cohabiting or marital union
Full size table

Findings

Prevalence of Post-Conception Unions

How common are post-conception unions? Table 2 shows the prevalence of post-conception cohabitations and marriages for the sample as a whole, as well as for subgroups, weighted by the person weights provided by the NSFG. Among all first births within unions, we report that 18 % were formed after conception. That is, nearly 1-in-5 first births in unions were conceived prior to the transition to either cohabitation or marriage at first birth (last column, Table 2). Pregnancy clearly is linked to union transitions among a significant share of unions. Of these post-conception unions, equal shares comprise couples transitioning to cohabitation (9 %) and to marriage (9 %). Still, a much larger share of cohabiting relationships in our sample (25 %) began with a post-conception transition than was the case among married couples (14 %). These comparative estimates are entirely consistent with previous studies showing that shotgun cohabitations have now supplanted shotgun marriages (Gibson-Davis and Rackin 2014; Lichter et al. 2014).

Table 2 Prevalence of post-conception unions by subgroup
Full size table

Union transitions precipitated by a pregnancy are most likely to occur among young adults, blacks and Hispanics, and the less-educated. Twenty-two percent of first births to married or cohabiting blacks and Hispanics began with a post-conception transition into either cohabitation or marriage. Among these groups, transitions to cohabitation were more common than transitions to marriage, while the opposite is true among white women. And, among those aged 20 or younger at the time of conception, a disproportionate share of post-conception unions involved cohabiting couples rather than marriages (22 % of first births to teens involved a post-conception transition into cohabitation, while an additional 13 % involved post-conception marriages). Among older women, the majority of post-conception unions were marriages (even though these marriages represented only a small fraction of unions with first births, i.e., 5 %). Women who were younger than 20 at the time of conception were also more than four times as likely to have a post-conception union than women who were 30 or older at the time of conception (35 % compared to 8 %, respectively).

Survival Rates of Cohabiting Unions Following Birth

After a first birth, do cohabiting relationships persist, dissolve, or transition into marriage? Table 3 provides answers by showing year-to-year probabilities of survival from multiple-decrement life table analyses. We distinguish between post-conception and pre-conception cohabiting unions in addition to providing life table estimates for both groups combined.

Table 3 Multiple-decrement life table estimates of cohabitation outcomes
Full size table

Only 25 % of post-conception cohabiting unions survive as cohabiting unions for at least 5 years after the birth. This compares with 35 % among pre-conception cohabitations, i.e., those that were already cohabiting when they conceived and bore a first child. Clearly, some share of these pre-conception cohabiting unions represents an alternative to marriage. Whereas 56 % of these pre-conception cohabitations last at least 5 years (1–0.44), either as cohabiting or marital couples, 53 % of post-conception cohabitations last this long. This is a small difference. The overall message is that the chances of surviving 5 years is only slightly better than 50–50 among couples that have a first birth while cohabiting whether they started to cohabit before or after conception.

The estimates in Table 3 also show that post-conception cohabiting unions end more rapidly than pre-conception cohabiting unions. Only 70 % survive the first year after the birth. Compared with those already living together at the time of conception, post-conception cohabiting couples are both more likely to marry (13 vs. 10 %) and separate (16 vs. 12 %) in the first year following childbirth. Of course, whether these marriages last is far from certain. It is clear, however, that the pregnancy and childbirth are highly associated with subsequent union transitions. For a significant share of post-conception cohabiting couples, pregnancy and childbirth are linked with the rapid succession (over a 2-year period) from singlehood to cohabitation to marriage.

As shown in Table 4, cohabiting unions—regardless of whether they began before or after the conception—had much higher rates of disruption than marriage. Among all cohabiting unions with first births, one-half ended within 5 years of the birth. These overall dissolution rates among cohabiting couples, unlike the case in Table 3, include the dissolutions of cohabiting couples who subsequently married after the birth of the child. The difference in separation rates between cohabiting couples and married couples is striking. Only about 17 % of all couples whose first birth occurred within marriage ended their union within 5 years of birth. Following first birth, the likelihood of dissolution is 3 times greater among cohabiting couples than married couples (51 vs. 17 %).

Table 4 Life table estimates of share of sample separating by years since first birth
Full size table

The estimates in Table 4 nevertheless indicate that the higher dissolution rates among cohabitors are not due entirely or mostly to the inclusion of post-conception unions. Although post-conception cohabitations have the highest 5-year cumulative dissolution rates (53 %), the difference in the risk of dissolution is only around 6 % higher than the dissolution rate among pre-conception cohabitors (50 %). Post-conception marriages also are less stable than pre-conception marriages, with a 5-year separation rate that is more than two times as high as couples who waited until after marriage to have children (31 compared to 15 %, respectively). Clearly, a post-conception union is a risk factor for dissolution, but the risk pales in comparison to the much larger differences between cohabiting couples and married couples overall. This claim is illustrated in Fig. 1. This figure clearly highlights the large differentials in dissolution between pre- and post-conception marriages, but comparatively small differentials between pre- and post-conception cohabitors.

Fig. 1
figure 1

Life table estimates of share of couples breaking up within 5 years of first birth, by union type at first birth

Full size image

Modeling Union Transitions among Cohabiting Unions

Some cohabiting couples marry after the birth of their first child. Others break up, or break up after they first marry. In this section, we fit discrete-time competing risks models predicting the likelihood of entering a marriage or breaking up, relative to remaining in a cohabiting union. The estimates in Table 5 indicate that a cohabitation formed after a pregnancy (post-conception cohabitation) is no more likely to transition into marriage than a pre-conception cohabitation. There is also no evidence that post-conception cohabiting unions differ from pre-conception cohabiting unions in the risk of dissolution, once other variables are included in the models. The observed differentials reported in multiple-decrement life table analysis are apparently due to differences in risk factors (i.e., selection).Footnote 7

Table 5 Multinomial logistic regressions predicting likelihood of remaining cohabiting, entering marriage, or breaking up, by months since birth of first child
Full size table

These analyses also reveal that observable characteristics do little to predict transitions into marriage following a first birth within cohabitation. This could be due, in part, to the fact that few cohabiting unions transition to marriage following a non-marital birth. We do find evidence that couples whose first birth occurred after 2005 were half as likely to marry compared to couples whose first birth occurred before 2000. Patterns of union transition among cohabiting couples have undergone a large shift away from marriage over the past decade or so (see Furstenberg 2014). This is not due to the rise in shotgun cohabitation.

Finally, the results in Table 5 address the question of whether an unwanted or mistimed pregnancy is responsible for the lower marriage rates of cohabiting couples. Our estimates indicate that fertility intentions are associated with breaking up, but not with getting married. That is, among cohabiting women who became pregnant “earlier than expected,” the likelihood of relationship dissolution (or break up) was 29 % higher than among women whose first birth was planned, although this finding was only marginally significant (p value = 0.095). Planning status appears to be strongly linked to commitment, at least as measured by lower dissolution rates. Moreover, if the pregnancy was unwanted, the likelihood of dissolution was even higher—80 % higher (OR = 1.84). Intended fertility is clearly associated with the long-term stability of cohabiting unions. These results support previous work by Guzzo and Hayford (2012, 2014), who argue that an unintended birth disrupts fragile cohabiting relationships and unsettles the family system.Footnote 8

In Table 6, we fit a logistic regression predicting the likelihood of separating following a first birth for both marital and cohabiting unions. We include indicators for each union trajectory: pre-conception cohabitation, post-conception cohabitation, and post-conception marriage, with the pre-conception married couples serving as the reference category. Odds ratios on these indicators represent the relative risk of separating in each month following the birth compared to couples who were married at the time of conception and birth. First and foremost, the results here confirm the unusually high rates of dissolution among all cohabiting couples. The odds ratio was 2.22 among post-conception cohabitations, which means that couples who transitioned to a cohabiting union between pregnancy and child birth are at 2.22 higher risk of separating than married couples who had a first pregnancy and birth after getting married.Footnote 9 Clearly, a post-conception union represents a significant risk factor for dissolution, even after taking into account family background, SES, pregnancy intentions, and racial and ethnic background.

Table 6 Logistic regressions predicting likelihood of breaking up by months since birth of first child, first births in cohabiting or marital unions
Full size table

The risk of dissolution, however, is slightly higher among pre-conception cohabitations (OR = 2.34)—those couples that were already cohabiting at the time of conception. Whether this reflects selection into post-conception marriage is unclear. One argument, supported by the data here, is that those cohabiting couples who became pregnant and then got married rather than remained cohabiting may be the most committed to the long-term success of their relationships. Indeed, post-conception marriages had dissolution rates only about one-half as large as those of pre-conception cohabitations. These results beg an obvious question: whether these pre-conception cohabiting unions would have had more long-term stability had they married. If so, some observers suggest that pregnancy among fragile families (poor unmarried cohabiting couples) may be a “magic moment” for policy intervention (Gibson-Davis 2014).

The results in Table 6 provide some additional lessons about “risk factors” for union dissolution. Couples are less likely to dissolve if they grew up in intact families. Unions were also more likely to last if women were older at conception, and the pregnancy was intended rather than mistimed or unwanted. These predictors cannot, however, account for the much higher risks of dissolution among cohabiting than married couples and among those whose coresidential unions began only after the conception occurred (see Musick and Michelmore 2015). The growing consensus among family scholars is that so-called “cohabitation effects” on dissolution are declining or has been eliminated altogether by declines in selection (i.e., who cohabits) (Kuperberg 2014). But our results suggest, at a minimum, that such blanket conclusions are best avoided, especially among those who formed new families (i.e., those with a first birth). At least among cohabiting couples who have children, the long-term stability remains in doubt, especially when compared with the low dissolution rates among women who became pregnant and had first births after getting married.

Sensitivity Analyses

We also conducted some additional robustness checks. For example, we replicated these analyses (Table 6), but limited the sample to post-conception and pre-conception cohabiting unions that subsequently transitioned to marriage. Perhaps paradoxically, these results indicate greater marital stability among post-conception than pre-conception cohabiting unions. Post-conception unions were significantly less likely—57 % percent less likely—than pre-conception unions to dissolve (OR = 0.431) when clocked from date of birth and 58 % less likely to dissolve when clocked from date of marriage (OR = 0.420). Again, as before, these results seem to point to the poor relationship quality of cohabiting couples who conceived and bore children as cohabitors. These cohabiting couples evidently were unable or unwilling to make a commitment to marriage when they became pregnant, and they faced a greater risk of divorce if they married later (after the birth). These results seemingly speak to the negative selectivity for stable marriages among this group of cohabitors.

Additionally, we re-ran the analyses, generating parallel results to those reported in Tables 5 and 6, but using, in separate models, age at conception and age at first union (continuous, in months) instead of the categorical indicator of age used in our analysis. We found that these substitutions had little if any effect on our results or conclusions. We also replicated these models—with and without including age at first union—to evaluate how much the coefficient of post-conception cohabitation was affected by this different specification. We found no evidence that age at first union is responsible for the estimated effects of cohabitation on dissolution (cf., Kuperberg 2014). At least for cohabiting and married couples having a first birth, the higher separation rate among cohabiting couples does not appear to be due to age at conception or entry into union.

We also replicated the analyses in Table 6 separating the post-conception marriage category into two groups: those who were single at the time of conception and those who were cohabiting at the time of conception. We found few differences between these two groups. Finally, we also experimented with using the date of the coresidential union as the starting point of the analysis rather than the date of the first birth. Once again, we found that results were quite similar to those presented here. We did find that all three union types (pre-conception cohabitation, post-conception cohabitation, and post-conception marriage) were at elevated risks of separating compared to pre-conception marriages, likely reflecting the fact that those who married prior to the conception were coresidential for a much longer period of time than any of the other three union types. Those who were married at the time of conception were living together for, on average, 52 months prior to the birth of their first child. In contrast, those who were cohabiting at the time of conception were living together for 25 months prior to the birth of the first child, while those who began cohabiting only after the conception were living together for just 5 months at the time of the birth.

Discussion and Conclusion

Post-conception cohabiting unions are on the rise, supplanting the longstanding but diminishing significance of the so-called post-conception or shotgun marriages. Many more unmarried couples today are moving in together after knowledge of a pregnancy, often an unintended one (Gibson-Davis and Rackin 2014; Lichter et al. 2014; Manning 2001). Whether these cohabiting unions persist, lead to marriage, or ultimately break up is unclear. Our primary objective, using data from the 2006–2010 and 2011–2013 National Survey of Family Growth, was to compare the long-term stability of post-conception cohabiting unions with other coresidential unions (i.e., pre-conception cohabiting and marital unions, and shotgun marriages). Our theoretical and empirical approach bridges a large literature on union instability and other new research on pregnancy and childbearing among cohabiting unions, which is often associated with relationship “churning” over the family life course (Cherlin 2010; Furstenberg 2014; Lichter and Qian 2008).

Our analyses focused on the union outcomes of couples whose first birth occurred in coresidential cohabiting or marital unions. But unlike previous studies, we highlighted the fact that many first births in unions resulted from a conception that occurred before coresidential cohabitation or marriage. Indeed, our analyses revealed that roughly 1-in-5 first births within unions in the United States were associated with rapid transitions either into cohabitation or marriage. For cohabiting couples in particular, 1-in-4 first births were the result of a pregnancy in a non-coresidential union. And, as we showed in this study, these couples had disproportionately high rates of dissolution—nearly 50 % dissolved within 5 years. Only a small minority of these couples married (i.e., less than one-third). Moving in together following a pregnancy—especially an unintended one—does not seem to be a recipe for union stability. These couples are much more likely to break up than to marry.

Interestingly enough, the dissolution rates among post-conception cohabiting unions were only slightly higher than those of pre-conception cohabiting unions. Rackin and Gibson-Davis (2012) found similar results when they tracked the cohabiting unions of adolescent and young adults in the late 1990s. Our hypothesis that a post-conception cohabiting union cuts short the time for child-free adjustment or leads to a poor match between partners received little support. Indeed, in our multivariate analyses, any differences in dissolution between post- and pre-conception unions were relatively small and statistically insignificant. The high rates of dissolution among pre-conception unions occurred despite the fact that these new parents had some time to adjust to each other in an ongoing coresidential union. Instead, our results may reflect another kind of selection: The most committed cohabiting couples may decide to marry after finding themselves pregnant. This suggests that the least prepared or committed to their relationships may stay together in fragile cohabiting relationships.

In fact, our results revealed slightly lower rates of dissolution among post-conception marriages than pre-conception cohabitations, after controls were introduced (cf., Rackin and Gibson-Davis 2012). The stability of marriages—regardless of its pathway (i.e., a non-marital conception or not)—was clearly greater than that of cohabiting unions with a first-born child. For pre-conception cohabiting unions, one obvious counterfactual would have been to marry before the birth of the child. Since we have not measured commitment or love or other characteristics, we cannot discern whether this large differential in union stability reflects selection into post-conception marriage or something about the “enforceable trust” or commitment associated with having a marriage certificate and the legal responsibilities it entails. From a policy standpoint, this raises the usual but controversial question about whether government (e.g., healthy marriage initiatives) or non-profit or faith-based organizations should endeavor to incentivize marriage for the sake of the children (Gibson-Davis 2014; Su et al. 2015). Regardless, our results clearly shed new light on the implication of childbearing among cohabiting couples at a time when nearly 60 % of all non-marital births occur in cohabitation (Lichter et al. 2014).

While efforts to promote marriage are often controversial, few scholars downplay the potentially deleterious implications of high rates of unplanned pregnancy among single and cohabiting couples, or discount the likelihood of negative developmental trajectories for unwanted children. Although unintended pregnancies reduce union stability, our study also indicated that these higher rates of unintended pregnancy were not solely responsible for the much higher rates of dissolution among cohabiting than married couples. Pregnancies—unintended or otherwise—also did not account for differences in dissolution between post- and pre-conception cohabiting unions. At a minimum, our results nevertheless underscore the need to improve access to family planning among unmarried and cohabiting couples (Guzman et al. 2010). They also highlight the need to better understand contraceptive behavior and family planning decisions among cohabiting couples, who contribute disproportionately to unintended pregnancies and union instability (Sassler and Miller 2014; Sweeney 2010; Sweeney and Raley 2014). Qualitative studies suggest that open and honest discussions of childbearing and contraception among cohabiting couples, particularly those who are economically disadvantaged, are often limited (Reed 2006; Sassler and Miller 2014).

Our results are not without some limitations. The empirical results and substantive conclusions pertain to the stability of unions in which a first birth occurred, and are not necessarily representative of all unions but selective of (or conditioned on) those that proceeded to a conception and live birth.Footnote 10 The cross-sectional design of the NSFG has the advantage of including retrospective data on union histories and fertility. But it is also limited by the small number of variables available at the time of conception. This means that our analyses provide a benchmark rather than the final answer to many potentially important questions about the origins of alternative trajectories to a first birth. We also cannot draw unequivocal conclusions about the causality or “causal effects.” And, lastly, our results for the United States and for highly aggregated pan-ethnic minority populations (e.g., Hispanics) are useful but also risk masking substantial heterogeneity in partnering and parenting across racial and ethnic groups, national origin groups, and immigrant populations. As the United States moves toward a majority–minority society (Lichter 2013), national averages and estimates, even those based on nationally representative survey data, may increasingly misrepresent subgroup variation.

In an age of increasing fertility among cohabiting unions, however, our study nevertheless highlights the importance of union stability in the lives of increasing shares of American children. The results—at least for this sample of new parents—suggest that it may be premature to hail “the end” of the longstanding strong positive association between cohabitation and dissolution. For those cohabiting couples who have children, our analyses indicate that the risk of dissolution remains very high in comparison to couples who wait until after marriage to have children. Obviously, we cannot dismiss selection as a primary explanation for our finding, but neither can we dismiss causal arguments that emphasize the fragility of cohabiting relationships and the destabilizing effects of childbearing, especially if it is unplanned or unintended (Guzzo and Hayford 2012). Now, perhaps more than ever, it is important to better understand the changing relationship between cohabitation and fertility, especially as the boundaries between cohabitation and marriage blur and are reshaped by shifting patterns of contraceptive use and efficacy, growing income inequality, and increasing racial and ethnic diversity (Hayford et al. 2014; Sweeney and Raley 2014). As we have shown in this study, the American family system is rapidly being transformed by changing patterns of fertility among cohabiting couples.