I’m considering moving into an apartment with my girlfriend next semester. Probably the biggest risk here is us breaking up and then being unable to bear the sight of each other, leaving one of us with an inconveniently expensive lease.

I feel like that’s a fairly unlikely outcome–I don’t see reasons that I’d want to break up with her and she says she feels similarly towards me–but I’m also willing to bet that people in their early twenties are systematically overconfident about the stability of their romantic relationships.

What I need is some actuarial tables! At its most basic, I’d love to know the distribution of relationship lengths which started when the participants were 18 and 19. From that distribution, I could calculate the probability that a relationship which lasted 19 months would last another 5.

We’re are a fairly atypical couple, and it would be great if the actuarial tables could get enough detail to take some of that into account. I suspect that me living in San Francisco for 7 months is evidence for our relationship lasting longer in person, if we survive the first month after I get back. I also suspect that this distribution varies greatly based on class and background, so it would be nice to have that level of detail.

Luckily, psychology researchers have an easy time conscripting university students to fill out questionnairres for them, thanks to compulsary research participation. So this is a reasonably well researched little subfield. I’ve done a very hasty review of the literature.

## Breakups in the general population

The best resource on nonmarital romantic relationship disillusion of the general population seems to be [1], a meta-analysis. It lists a variety of factors in relationship stability. Many of them are obvious–love, commitment, lack of other options–but most of them are pretty subjective. The one I found most useful is that longer relationships are less likely to break up at a given time. In general, I think this paper wasn’t enormously useful for my situation because I bet that the dynamics of relationships are quite different between university students and the general population.

The strongest factor in relationship stability from that paper was apparently “positive illusions”, which are described in most detail in [3]. I don’t know if I buy this. They’ve got a bunch of evidence that a high opinion of your partner’s dedication correlates with having a relationship which lasts longer. But that’s exactly what you’d expect!

The positive illusions papers try to get around this by saying that they’re talking about people who are unrealistically overconfident. I’m skeptical of that too. If we take these studies at face value, we’re asking people to predict the chance their relationship will survive for whatever length of time, and then seeing what proportion of them actually last that long. This sounds like a lot like a test of people’s probability calibration, though, and we already know that people are totally lousy at that. So I don’t know if we have sufficient extra evidence to actually conclude that positive illusions are a causal factor in relationship success.

## Breakups in university students

But as I said, I’m more interested in research particularly aimed at relationships among university students. A good study of university student dating was done in 1985 by Simpson ([2]). He found that relationship stability was predicted by: relationship satisfaction, closeness, prior length of the relationship, availability of an alternative partner, quality of alternative partners, whether the relationship was exclusive, and a conservative attitude to casual sex. (Incidentally, he also found that everything which makes relationships last longer also makes breakups worse.)

We certainly have relationship satisfaction, closeness, and prior length of the relationship on our side. It’s not obvious how to apply the other variables, because of the nature of my relationship. (For example, relationship exclusivity was probably just a proxy for how seriously people took the relationships, so seems inapplicable here.) In terms of alternative partners, I’m quite sure there’s no-one I’d rather be dating, given how closely we fit on a variety of issues where it’s rare for me to find a match.

By far the most informative variable here is prior length of the relationship, given how long the relationship has lasted. Based on that alone, I have pretty strong confidence in its continued health.

[1] “Predicting nonmarital romantic relationship dissolution: A meta-analytic synthesis.” Benjamin Le, Natalie L. Dove, Christopher R. Agnew, Miriam S. Korn, and Amelia A. Mutso http://onlinelibrary.wiley.com/doi/10.1111/j.1475-6811.2010.01285.x/abstract

[2] “The Dissolution of Romantic Relationships: Factors Involved in Relationship Stability and Emotional Distress.” Jeffry A. Simpson http://www.unc.edu/courses/2006spring/spcl/091p/016/DissolutionOfRelationships.pdf

[3] “A Leap of Faith? Positive Illusions in Romantic Relationships.” Sandra L. Murray and John G. Holmes http://psp.sagepub.com.virtual.anu.edu.au/content/23/6/586.full.pdf+html