Premarital Cohabitation Is Still Associated With Greater Odds of Divorce

A new study published in the Journal of Marriage and Family finds that the “premarital cohabitation effect” lives on, despite what you’ve likely heard. The premarital cohabitation effect is the finding that those who live together prior to marriage are more likely, not less, to struggle in marriage. It has a long and storied history in family science.

Michael Rosenfeld and Katharina Roesler’s new findings suggest that there remains an increased risk for divorce for those living together prior to marriage, and that prior studies suggesting the effect has gone away had a bias toward short versus longer-term effects. They find that living together before marriage is associated with lower odds of divorce in the first year of marriage, but increases the odds of divorce in all other years tested, and this finding holds across decades of data.

Numerous Recent Studies Reported No Impact of Premarital Cohabitation

A number of relatively recent studies suggested that the premarital cohabitation effect has gone away among cohorts marrying in the last 10 or 15 years. Rosenfeld and Roesler pay particular attention to a report from the National Center for Health Statistics by Copen, Daniels, Vespa, and Mosher in 2012, which suggested there was no increased risk associated with premarital cohabitation in the most recent (at the time) cohort of the National Survey of Family Growth (NSFG; 2006 – 2010). Reinhold came to the same conclusion in 2010, and while not cited in the new study, Manning and Cohen reached the same conclusion in 2012, incorporating data from as late as the 2006 to 2008 cohort of the NSFG.¹ While all of these studies used the NSFG, Rosenfeld and Roesler had longer-term data for the most recent cohort they study (up to 2015). Contrary to these prior conclusions, they found that there remains a clear link between premarital cohabitation and increased odds of divorce regardless of the year or cohort studied. (In all these studies, the focus is on first marriages.)

The theoretical underpinning for all the prior papers noted above was the belief that, as living together became more normative, it would no longer be associated with negative outcomes in marriage. One reason often suggested is that there is no longer a stigma among friends and family about living together before marriage. Another reason, theoretically, is that those living together prior to marriage are no longer as select for higher risk as in the past because most people are doing it.

Based on a different line of reasoning, another prominent study also concluded that there was no longer an added risk for divorce associated with premarital cohabitation. However, in that study, Kuperberg (2014) concluded the risk was more about moving in together at a young age (before the middle 20s) than moving in together before marriage, per se. That’s one among many potentially important nuances in this complex literature.²

Recent Studies May Have Been Premature

Cohabitation is the gift that keeps on giving to family science, providing generations of scholars with the opportunity to say, “look here, wow, this is strange.” For starters, it’s counterintuitive that living together before marriage would not improve one’s odds for a successful marriage. And yet, whatever else is true, there is very scant evidence to support this belief in a positive effect (more on that in this piece; see also this.)

Enter Rosenfeld and Roesler. Their new paper is quite complex statistically, but their insight boils down to two things easily explained. First, they believe studies that suggested that the premarital cohabitation effect has disappeared simply did not have outcomes for divorce far enough out for those who had married in the recent cohorts that they examined. Second, they show that premarital cohabitation is associated with a lower risk for divorce, but only very early in marriage (in the first year); in contrast, the finding flips, with premarital cohabitation being associated with higher risks for divorce in years after that first year. That’s what earlier studies could not address. In particular, Rosenfeld and Roesler suggest that those who live together before marriage have an advantage in the first year because they are already used to all the changes that come with living together. Those who go straight into marriage without living together have a bigger immediate shock to negotiate after marriage, and as a result, have a short-term increased risk that’s greater than those already living together. But that’s short-term, and the risk remains long-term.

Here is a quote from the new paper (see pages 7-8):

Figure 2 shows that, for the years in which the NSFG has substantial numbers of marriages and breakups, there was no apparent trend over time in the raw or adjusted odds ratios of breakup for premarital cohabitation. Given the enormous changes over time in the prevalence of premarital cohabitation (see Figure 1), Figure 2 shows a surprising stability in the association between premarital cohabitation and marital dissolution over time.

Note: Used with permission of the National Council on Family Relations. 

Theories of Increased Risk

There are three dominant theories of causality in how living together before marriage could be associated with worse outcomes (on average) in marriage—explaining why the finding is just not what most people expect it should be. Rosenfeld and Roesler address the first two but did not say anything about the third.³

Selection. This theory is simply that there are many factors associated with who cohabits when and why, and with whom, and that those factors are also associated with how marriages will turn out regardless of cohabiting experience. For example, it’s well known that those who are more economically disadvantaged are more likely to: live together outside of marriage, live together with more than one partner, have a child with a cohabiting partner prior to marrying, and struggle in marriage. Other factors are religiousness, traditionality, and family history (parental divorce, etc.). The selection explanation is that those who cohabit in riskier ways (e.g., before marriage, before engagement, with more than one partner) were already at greater risk. In the strongest view of selection, living together does not add to the risk at all because it’s all already baked in. There is a lot of evidence for selection playing an important role in this literature, and scholars in this area note this and address it in various ways.

The Experience of Cohabiting Changes Things. In an older line of research that was brilliant but badly needs updating, Axinn and Barber (1997) showed that cohabiting changes attitudes about marriage and divorce, lowering esteem for marriage and increasing acceptance of divorce. This is consistent with scores of studies in psychology that show that attitudes will cohere to behavior. Earlier, Thornton, Axinn, and Hill (1992) showed that cohabiting led to people becoming less religious. Rosenfeld and Roesler include a lot on the theory of experience, but mostly use it to emphasize the short-term benefit of already experiencing living together when transitioning into marriage.

It’s counterintuitive that living together before marriage would not improve one’s odds for a successful marriage. Yet, whatever else is true, there is very scant evidence to support this belief in a positive effect. 

Inertia. We have argued since the early 2000s for another causal theory in this line of research. Drawing on theories of commitment, we suggested that what nearly everyone misses in understanding the risk associated with cohabitation is pretty simple: moving in together makes it harder to break up, net of everything else. The added risk is due to how cohabitation substantially increases constraints to remain together prior to a dedication to a future together maturing between two partners. Two key papers on this perspective are here and here.⁴

One primary prediction from the inertia hypothesis is that those who only started living together after being already committed to marriage (e.g., by engagement or actual marriage) should, on average, do better in marriage than those who may have prematurely made it harder to break up by living together before agreeing on marriage. The inertia hypothesis completely embraces selection, suggesting that relationships already at greater risk become harder to exit because of cohabitation. Various predictions from the inertia hypothesis have been supported in ten or more studies, seven of which include tests of the prediction about pre-cohabitation levels of commitment to marriage (aka plans for marriage prior to living together)—and this latter finding exists in at least six different samples across a range of outcomes.⁵

There is no particular reason to expect that the inertia risk will dissipate with increased acceptance of cohabitation because the mechanism is about the timing of the development of aspects of commitment, not about societal views and personal attitudes. For living together to lower risk in marriage, the benefit of learning something disqualifying about a partner has to exceed the costs of making it harder to break up that comes with sharing a single address. Hence, inertia is another possibility along with experience that could explain the persistence of a cohabitation effect, such as found by Rosenfeld and Roesler.⁶

Other Possibilities. Other factors that may be associated with differential outcomes include pacing (Sassler et al.), age at the time of moving in together (Kuperberg), and premarital fertility (Tach & Halpern-Meekin). All such theories of moderated outcomes suggest that the risks of living together before marriage are greater for some groups than others. Rosenfeld and Roesler are not really addressing this issue. However, they did find that the risks associated with premarital cohabitation were lower for African Americans. While that’s a subject far beyond our focus here, it does not surprise us. For most groups, cohabitation is no particular indicator of higher commitment. However, it may well signal higher levels of commitment among groups where marriage has declined a great deal, like African Americans.

Rosenfeld and Roesler also note that the risks of living together before marriage were even greater among those who had lived with more than just their mate prior to marriage. That finding is consistent with many other studies, including Teachman (2003).

It Lives

Research on premarital cohabitation has long been mired in arguments about causality, with the dominant view being that selection explains most, if not all, of the risk. However, many studies in the history of this field have controlled for putative selection variables and still found an additional risk. In fairness, it is not possible to control for all aspects of selection in such studies. Without randomly assigning people to walk different pathways before marriage, causality can never be proven. Arguments ensue—and since when does evidence cause us to stop arguing anyway when people are passionate about their view on something?

Rosenfeld and Roesler’s new study breathed life into a finding many concluded was dead.

Scott M. Stanley is a research professor at the University of Denver and a fellow of the Institute for Family Studies (@DecideOrSlide). Galena Rhoades is a research associate professor in the Psychology Department at the University of Denver.

---

1. We are mystified why the new paper does not cite or address the findings by Manning and Cohen. That study seems like it is the most recent major study directly addressing the question Rosenfeld and Roesler examine.

2. Scott wrote about the Kuperberg study at that time, taking far more issue with the media stories about it than the actual study, suggesting there are many ways people could misconstrue to whom those, and other findings of differential risk, applied. Those articles are here and here.

3. This omission does not seem as striking to us as the omission of Manning and Cohen’s paper, since their paper is already complex and they are intent on addressing one moderator of the cohabitation effect: how long after marriage the effect is measured. They do not address at all the growing literature on moderators of the cohabitation effect. Still, inertia is one of the major theories of increased risk, and only selection itself has more publications addressing it.

4. An accessible, word document version of the major theory paper can be found here. A full run-down of our theoretical and empirical work in this line is available here. That includes citations and links, mostly to accessible versions of the articles in the literature.

5. We have found evidence for inertia whether or not someone has cohabited only with their mate, and in numerous samples of people marrying after 2000 and later.

6. As an interesting side point on the subject of the inertia hypothesis, the commitment to marriage/timing effect exists in the NSFG. It was mentioned in passing in a working paper leading up to the 2010 publication by Reinhold, and it is mentioned prominently in the abstract (and paper) in Manning and Cohen’s 2012 publication.