In my first piece on Arielle Kuperberg’s study on cohabitation that got so much media attention, I focused on broad conceptual issues. In this piece, I am going to focus on more technical matters. While I remain impressed with aspects of Kuperberg’s study, I have concerns about some of the conclusions that can be drawn from the work. To recap, she showed that some of the risk of cohabiting is related to the age at which partners move in together, with those beginning to cohabit at a young age (just like those marrying at a young age) being at a higher risk for divorcing the partner with whom they cohabited prior to marriage. That’s an important finding, but I do not believe that it explains everything that is associated with risk in some patterns of relationship development that are associated with cohabitation before marriage.
To Whom Do the Findings Apply?
Kuperberg focused on a large sample that has many strengths for assessing outcomes related to divorce. At the end of the day, her analyses tell us about people who married their cohabiting partners, and whether or not that marriage was more or less likely to end based on a history of cohabiting together prior to marriage. By premarital, she means with a specific partner. Her main findings—and the media headlines—are not directly related to other patterns of risk, such as serial cohabitation or having an unplanned child together while cohabiting—all of which are “premarital” in that they are before one marries. Since cohabiting couples are decreasingly likely to eventually marry,i Kuperberg’s main findings really focus on the increasingly select group who marries, either with or without cohabiting first, without much else going on to complicate life before marriage. As others have noted (e.g., Laura Tach and Sarah Halpern-Meekin, 2009ii), unmarried and premarital cohabitation has become a heterogeneous phenomenon with many complex manifestations. I believe that part of the complexity lies in the fact that there are substantially interrelated pathways of risk.
Kuperberg’s main findings really focus on the increasingly select group of couples who marry, either with or without cohabiting first.
A colleague of mine likened Kuperberg’s main conclusion, that cohabitation before marriage is not risky regarding the odds of divorce, as similar to concluding that, among those who are super fit and exercise a lot, eating less healthy food has almost no consequence. I’m not a nutritionist, but I know people who are super fit who can eat about anything and they are not going to imperil their health to the same degree as others who are less fit. They burn it all up. I am not saying that cohabitation is junk food. I’m focusing on relative risk: this analogy is apt in that there is a lot of evidence that the pathways associated with some patterns of cohabitation are, indeed, riskier than other pathways. Some people are not at greater risk, and any individual’s risk level is related to a mix of variables that includes personal characteristics, background, socio-economic disadvantage, and, as I argue below, behaviors that increase risk.
We should not, therefore, be surprised to find that those who are on a lower-risk pathway (for example, as Kuperberg suggests, those who cohabit at age 23 years or later), are at lower risk of divorce. But there are other pathways of cohabiting prior to marriage that many people travel, such as cohabiting before reaching clarity about any commitment to the future, cohabiting with multiple partners, and having an unplanned child in a low-commitment cohabiting relationship. (Cohabitation leads to increased odds of unplanned births.) As a scientist oriented toward risk prevention, I don’t spend a lot of time worrying about people who are already on a lower-risk pathway. I’m worried about everyone else and what we convey to them publicly about what the literature shows.
What About Marital and Relationship Quality?
While the sample Kuperberg uses has massive strengths, since it is large and representative of the U.S., her analyses are limited to the outcome of divorce. That’s fair. I think divorce is the single most important outcome that tells you how a marriage goes. But divorce is not the same as marital quality or, more specifically, marital happiness—another important outcome.
In our research (with my colleagues Galena Rhoades and Howard Markman), we examine marital quality in addition to divorce. We predicted long ago that those who wait until either marriage or engagement to live together should be at lower risk in marriage than those who cohabit prior to achieving clarity about a commitment to the future. The premise here is based on what we call the inertia of cohabitationiii: it is harder to break up when cohabiting than it is to break up when dating but not sharing a single address. I’m not saying that cohabiting couples do not break up; they do all the time. But how can it not be harder to break up when sharing a single address than when you have your own place? I can imagine plenty of situations where people with different advantages and disadvantages are at greater or lesser risk from inertia, but I cannot fathom how the inertia of cohabiting is not greater than dating, on average, for just about everyone.
Cohabiting prior to marriage or making clear plans for marriage is associated with less happiness and more negativity in marriage.
Here’s the risk. Many couples move in together before having any clear and mutual commitment to the future. Therefore, some people make it harder to break up with their partner before they have settled the question of if they really are planning a future with this person. Is everyone who cohabits without clear commitment about the future at an elevated risk of divorce or an unhappy marriage? I do not think so. For example, I think there are plenty of people who begin to cohabit without having clarified any future path with their partner who do fine because they land with someone who is a good partner for them anyway—very often, with the same person they would have ended up with if they had been more cautious.
But, on average, the group with higher risk should be those who cohabit prior to attaining mutual clarity about commitment because therein lies a subgroup who is at increased risk for marrying someone they would not had married if they had never moved in together. And some of the couples who do not marry will, nevertheless, be together longer because of cohabitation, thereby increasing their odds of having an unplanned child together, which puts both their child and their own future marriages at risk. In other cases, people burn two or three years with someone they might have broken up with and moved on from after 1 year without cohabitation. There is an opportunity cost in that.
Based on this theory about inertia, we have tested and shown (in study after study) that cohabiting prior to marriage or making clear plans for marriage is associated with less happiness and more negativity in marriage. These ideas are consistent with the central advice Kuperberg gives in her paper: she suggests that people who move in young are at increased odds of making a poor choice in their partner, which, in turn, makes divorce more likely.
Norval Glenn, the famed sociologist of the family who passed on a few years ago, termed this inertia-related risk “premature entanglement.”iv His central focus was how premature entanglement shortens a person’s search for the best mate they may have otherwise obtained. If the average person thinks they can get around the fact that cohabitation makes it harder to break up, all other things being equal, they may be kidding themselves. For some, the increased risk may be marginal or negligible. For others, it is life-altering. My colleague Galena Rhoades and I like to recommend that people consider lower-cost ways to figure out whether a person is the right partner for marriage—ways that do not make it harder to break up even as you’re trying to figure out if a future makes sense.
After beginning to cohabit, couples’ negative communication rises sharply, and both relationship satisfaction and the perceived likelihood of marriage go down.
In one of our recent papers, Galena Rhoades explained a type of risk that is unrecognized by some couples until they experience it while living together.v In this paper, which included some of the most sophisticated analyses we’ve ever conducted on how couples change when they cohabit (controlling, powerfully, for selection by examining within-person changes), she noted that, for many couples, cohabitation combines two different developmental tasks in one period of time. First, consistent with what I just noted above about inertia, many (and likely most) cohabiting couples start living together before having clarified their plans for the future. So even as they live together, they must grapple with the big question about the future—which, when settled for a couple, provides immense benefits to relationship quality because a clear sense of a future together changes how people treat one another in the present. But for many couples beginning to cohabit, this is anything but settled. Second, moving in together involves changes in routines, roles, and expectations—just as in marriage. (Kuperberg likewise discusses the potential issues that arise when taking on new roles.)
When we recognize that many couples are experiencing two challenging developmental stages at once as they move in together, it is perhaps no wonder that we find that, after beginning to cohabit, couples’ negative communication rises sharply, and both relationship satisfaction and the perceived likelihood of marriage go down. Further, the type of constraints that make it harder to break up take a large jump and start to grow faster, and the type of commitment (dedication) most associated with having a high-quality relationship levels off.
My key point here is that relationship quality matters, and studies that focus on the risk for divorce tell an important, but incomplete story. A lot of dimensions of a relationship are impacted by cohabiting, with numerous implications for eventual marital quality (and divorce).
Kuperberg’s Study Does Not Examine Mechanisms of Risk
Kuperberg’s study is not designed to examine specific mechanisms of risk in how relationships unfold. Here is a partial list of what my colleagues and I think a lot about when it comes to mechanisms of risk. Not all of them are specific to cohabitation (references provided as examples).
- Cohabitation can create inertia causing couples to remain together prior to making a clear and mutual decision to be together in marriage.vi This is why we believe that cohabiting before a clear, settled commitment to marriage is more risky than waiting until either after marriage or after having mutual, public plans to marry.vii
- Sex too soon can lead to cohabitation too soon, perhaps leading to greater odds of remaining with a partner one would otherwise not have remained with, or remained with for as long.viii
- Some (maybe a lot) of the risk for divorce and lower marital happiness associated with premarital cohabitation is driven by non-marital births in the cohabiting population, with the risk being especially strong for premarital births with the marital partner (these associations are stronger for white women than black or Hispanic women).ix
- Early sexual connection may create relationships in which couples make key decisions about the future before other aspects of the relationship have fully developed.x
- Moving in together at a younger age (or marrying, for that matter) is associated with increased risk for divorce.xi
Everything I just listed reflects something about relationship transitions that can alter one’s future options. Different researchers will focus on different risks, but I think all of these risks are in a similar basket—including the variable Kuperberg proposes as mattering the most (young age at co-residence). They all can be seen as involving “sliding” through important, potentially life-altering transitions, rather than making deliberate, adequately informed decisions about them after settling other important matters for the individual or the couple.xii I think sliding through such transitions is now a defining feature of how we do romantic relationships before marriage in the U.S. In essence, this means people are routinely getting the most valuable information about the prospects of a relationship after they have already forfeited alternate options. Further, measuring any of these patterns that may reflect this generic risk factor can make it harder to detect others in social science because the overlapping variance is so great among all the things on this list.
Sliding through major transitions is now a defining feature of how we do romantic relationships before marriage in the U.S.
Kuperberg does not directly examine these various pathways of risk except for age at co-residence, though many of these things I just listed are intertwined with—or intensified in—cohabitation that occurs prior to marriage (with the future spouse or with other partners). In Kuperberg’s study, the variables most closely aligned with what I’d consider mechanism of risk are entered as “demographic” control variables. Not all her analyses use controls, but where she does, Kuperberg includes variables such as age at co-residence (or age at marriage, depending), education, race/ethnicity, family stability growing up, if one grew up religious or not, if one had previously cohabited with someone other than the mate (serial cohabitation), if the couple had moved in together while expecting a baby, and if there had been any birth prior to cohabiting (within the relationship or from a prior one).
To me, three of these things are not like the others: the last three. I would not classify them as mere demographic controls. I see them as behaviors associated with mechanisms of risk. Another colleague noted that these are really demographic events rather than demographic control variables. They are choices that affect one’s odds of achieving stable and lasting love in marriage or otherwise. The debate, of course, is in how much control a person really has, and I fully accept that some people have a lot less ability to control some of these things in their lives than others. There is certainly a lot of selection involved, but unless you have a highly deterministic view of human behavior (and many do), it seems wiser not to control for things that are highly interrelated with your predictor (cohabitation) as you examine a specific outcome like divorce. Of course, if you do use a lot of controls that involve complex patterns of risk, then the conclusion you want others to draw from your findings should be tightly specified so that the average person does not draw the wrong conclusion.
In essence, Kuperberg was able to show a reduction in the risk of divorce after cohabitation by controlling for age at co-residence instead of age at marriage. However, even in this aspect of her study, she shows that, for every age, those who cohabited before marriage were still more likely to divorce (at least, this is how I would interpret Figures 3 and 4 in her paper). She further reduces the association between premarital cohabitation and divorce by introducing the “demographic controls” noted above, but as I just explained, I think this amounts to controlling for the types of risk deeply intertwined with cohabitation in a study that will be understood as suggesting cohabitation does not matter. Not all of her analyses use the extensive list of control variables, however, and I think her finding that age of co-residence matters for marital outcomes makes a lot of sense.
There are some people who are not in any way at greater risk for divorce or lower marital happiness because they cohabited before marriage. But there is a rich set of interrelated risk behaviors—the ones I listed above—that reflect a more complex story about cohabitation than what the average person could have taken away from the media coverage of Kuperberg’s study. I think we need to keep trying our best, as social scientists, to make sure people can accurately see how various romantic patterns can bend the whole curve of their future possibilities in life. Where there is a complex story, we should try hard to tell it, despite the limitations of our sound-bite world.
i. Vespa, J. (2014). Historical trends in the marital intentions of one-time and serial cohabitors. Journal of Marriage and Family, 76, 207-217.
ii. Tach, L., & Halpern-Meekin, S. (2009). How Does Premarital Cohabitation Affect Trajectories of Marital Quality? Journal of Marriage & Family, 71(2), 298-317.
iii. Stanley, S. M., Rhoades, G. K., & Markman, H. J. (2006). Sliding vs. Deciding: Inertia and the premarital cohabitation effect. Family Relations, 55, 499–509.
iv. Glenn, N. D. (2002). A plea for greater concern about the quality of marital matching. In A. J. Hawkins, L. D. Wardle, and D. O. Coolidge (Eds.), Revitalizing the institution of marriage for the twenty-first century: An agenda for strengthening marriage (pp. 45-58). Westport, CT: Praeger.
v. Rhoades, G. K., Stanley, S. M., & Markman, H. J. (2012). The impact of the transition to cohabitation on relationship functioning: Cross-sectional and longitudinal findings. Journal of Family Psychology, 26(3), 348–358.
vi. Stanley, S. M., Rhoades, G. K., & Markman, H. J. (2006). Sliding vs. Deciding: Inertia and the premarital cohabitation effect. Family Relations, 55, 499–509.
vii. Rhoades, G. K., Stanley, S. M., & Markman, H. J. (2009). The pre-engagement cohabitation effect: A replication and extension of previous findings. Journal of Family Psychology, 23, 107-111.; Stanley, S. M., Rhoades, G. K., Amato, P. R., Markman, H. J., & Johnson, C. A. (2010). The timing of cohabitation and engagement: Impact on first and second marriages. Journal of Marriage and Family, 72, 906-918.
viii. Sassler, S., Addo, F. R., & Lichter, D. T. (2012). The tempo of sexual activity and later relationship quality. Journal of Marriage and Family, 74, 708 – 725.
ix. Tach, L., & Halpern-Meekin, S. (2009). How Does Premarital Cohabitation Affect Trajectories of Marital Quality? Journal of Marriage & Family, 71(2), 298-317.
x. Busby, D. M., Carroll, J. S., & Willoughby, B. J. (2010). Compatibility or Restraint? The Effects of Sexual Timing on Marriage Relationships. Journal of Family Psychology, 24(6), 766-774.
xi. Kuperberg, A. (2014). Age at coresidence, premarital cohabitation, and marriage dissolution: 1985-2009. Journal of Marriage & Family, 76(2), 352-369.; regarding age at marriage, see Raley, R. K., & Bumpass, L. (2003). The topography of the divorce plateau: Levels and trends in union stability in the United States after 1980. Demographic Research, 8, 245-260.; Teachman, J. D. (2002). Stability across cohorts in divorce risk factors. Demography, 39, 331–351.
xii. If you want to read more about this risk model, there is a chapter available for download. See page 28 and following, where it says “Our work on transition and risk.” Stanley, S. M., & Rhoades, G. K. (2009). Marriages at risk: Relationship formation and opportunities for relationship education. In H. Benson and S. Callan (Eds.), What works in relationship education: Lessons from academics and service deliverers in the United States and Europe (pp. 21 - 44). Doha, Qatar: Doha International Institute for Family Studies and Development. For the original finding that couples more often slide into cohabitation than deliberate about what it all means, see: Manning, W. D., & Smock, P. J. (2005). Measuring and modeling cohabitation: New perspectives from qualitative data. Journal of Marriage and Family, 67, 989 - 1002. For our initial paper that details how we see inertia and sliding mixing together to create added risk for some people, see: Stanley, S. M., Rhoades, G. K., & Markman, H. J. (2006). Sliding vs. Deciding: Inertia and the premarital cohabitation effect. Family Relations, 55, 499–509.