On the Value of Just Observing

What descriptive research revealed about singles' sexual satisfaction

A big reason I got into singlehood research is that I felt like people were making pretty strong claims about things like how happy single people are, but that there wasn’t actually a lot of information (especially quantitative data) on which to base those claims. It seemed to me that this was a situation that called for a descriptive research approach, and descriptive research is largely the kind of work I’m drawn to doing.

Descriptive research, as I understand it, involves documenting what a phenomenon is rather than starting with a theoretical model about what to expect and testing hypotheses based on that. Descriptive researchers are just here to observe, not to judge. I don’t think it’s a coincidence that my hobby is photography, it’s a similar mindset.

Descriptive research and the replication crisis

A lot of people, especially post replication crisis, will say that descriptive research is valuable. But it’s still pretty common in reviews of our work for reviewers to downgrade the work because it’s not theoretical. Of course, those reviewers never make clear why that is a problem exactly, or how we are supposed to explain things that haven’t been described yet. It always makes me want to mash up these lines from one of my favourite poems:

He will not go behind his father’s saying,

And he likes having thought of it so well

He says again, ‘This research is not theoretical.’

The vibe around descriptive research reminds me of pre-replication crisis times when there was a lot of pressure to confirm your hypotheses in a study, which led to a lot of p hacking behaviour. I remember in one of my earliest publications around the turn of the century I tried to be explicit that we developed our story for the results after we had done the analyses, and the editor encouraged me to write it as if we had predicted it all along. That’s how things worked back then.

There’s something analogous to me about the current pressures to come up with theoretical explanations for descriptive results, no matter how post-hoc the theoretical story is. Why not just let the descriptions stand, and wait until we’ve described the phenomenon well enough to start a serious, cumulative process of theorizing?

Sometimes we don’t know enough to theorize properly

In general, I think descriptive research is a good idea when we have little information about the phenomenon we’re trying to explain. The first line of work my lab did when we really started to get serious about thinking of ourselves as a singlehood research lab is an example. When we started to pay attention to the singlehood literature, to the extent that there was one, we noticed that the overarching question seemed to be whether single people could meet their emotional needs without a romantic partner. It struck us as odd, and a little Victorian, that there wasn’t discussion of single people getting their sexual needs met.

Doing descriptive research absolutely does not mean ignoring open science principles

This was around the same time my former grad student Yoobin Park figured out that there were some big, pre-existing data sets lying around to answer questions like this. Yoobin is amazing, and among other lasting effects she’s had on my career, she changed the way my lab works by pointing us in the direction of these big data sets. She corralled three of these big data sets so we could test and re-test if there was a link between singles’ sexual satisfaction and their well-being.

And there’s an important point here in the fact that we tested this across three data sets. Doing descriptive work doesn’t mean downplaying the importance of open science principles. With this kind of work, you can’t confirm a priori hypotheses, but you can see whether any effect you find replicates consistently. Personally, I’ve always been more interested in whether an effect is reliable than if someone was clever enough to predict it ahead of time.

Is singles’ sexual satisfaction important over and above emotional support?

We particularly wanted to know if sexual satisfaction was related to singles’ well-being over and above the emotional support they felt from their family and friend relationships. In our analyses, family and friend relationships seemed to be related to singles’ life satisfaction and satisfaction with singlehood, but not to desire for a romantic partner. So emotional support seemed to be related to personal contentment, but wasn’t related to whether you wanted a romantic partner or not.

So what about singles’ satisfaction with their sex lives? Just like with satisfaction with family and friend relationships, singles who were happier with their sex lives reported higher life satisfaction and satisfaction with singlehood. But, unlike those emotional support variables, we also found that singles who were happier with their sex lives reported lower desire for a romantic partner.

Now, one important note here is that our ability to measure desire for a romantic partner wasn’t perfect. When you use pre-exiting data sets, it is very much a “you get what you get” situation. So in one data set we had a proxy measure (desire to marry) and in another data set there was no variable at all that mapped onto desire for a partner.

But in the third data set, singles were asked directly if they wanted a romantic partner not just once but year after year. This meant that we could do the analyses in a way where we could not just say that the kind of single person who is sexually satisfied wants a partner less (which could be because of something like that person’s personality), but also that in years where people are more sexually satisfied that person wants a partner less (which gets us closer to the idea that sexual satisfaction might cause lower desire for a partner). So it looked like sexual satisfaction could be different from emotional support in that being sexually satisfied might take the edge off desire to be partnered.

Our descriptive approach produced unexpected insights

We didn’t predict this ahead of time. In fact, Yoobin and I were both pretty skeptical that sexual satisfaction would predict any of our well-being outcomes above and beyond family and friend support at all. But post-hoc, it started to make sense. A single person can get a lot of things from their non-romantic relationships, but sexual gratification is typically not one of them. So if sexual gratification typically (not always, obviously) comes from romantic partners, but singles are getting their sexual needs met without a committed partner, then it makes sense that they would have one less reason to want committed partnership (why buy the cow et al.).

This was the finding that first planted the seed in my mind that we may need to think not just about singles’ overall life satisfaction, but to also think about how satisfied singles’ are in various domains of their lives. I’ve also started using this more broadly as a framework to help understand when someone will be disinterested in partnership. The more things that a single person has in their life that most people look to a romantic partnership to provide, the less we should expect that person to be interested in partnering.

Sexually satisfied singles say they want a romantic relationship less, but are more likely to end up in one

But there was one more catch. One thing we were able to do with these data sets is to look at which singles were more likely to get into romantic relationships over time. And lo and behold, despite these sexually satisfied singles telling us that they were less interested in partnership, more sexually satisfied singles were more likely to be partnered in future waves of data collection. We didn’t make too much of this finding in the paper because I’m not sure we totally bought it ourselves. But we replicated it again this year, so we’re now pretty certain it’s really a thing.

Whither causality?

Of course, the nature of this kind of descriptive research is that we can’t be sure of why sexually satisfied singles are more likely to partner (despite their protestations to the contrary). Realistically, the effect probably has multiple causes. When you believe in webs of reciprocal causality like I do, the push to provide the One Causal Explanation of your descriptive findings at the end of a paper like this doesn’t seem wise. Personally, I think our tools for describing phenomena are not bad, but our tools for really, seriously digging into causality are kind of weak (but getting better) at this point in history. This is another reason I like doing descriptive work; I trust it more and think it provides for more realistic conclusions than a lot of causal work does.

I guess we could have done this work by developing a theoretical model of singles’ sexuality and setting up a study to ostensibly confirm how clever we are. But, to me, the way we approached this work is way more interesting. And again, I just think it was more honest given that we didn’t really have many basic facts about singles’ sexual lives on which to base predictions. In the end, there were enough unexpected results that came out of the analyses that I think we would have missed some of the most important parts of the phenomenon if we tried to build the study around some preexisting theory. Sometimes it’s best to just sit back and observe.

So – fellow academics – the next time you’re reviewing a paper, and you find that you have typed words to the effect of, “atheoretical,” I guess I’d just ask you to pause and be clear about why that’s a problem. Maybe it is, and maybe it isn’t – but both you and the author whose work you’re reviewing would benefit from being clear and explicit about what tools really best suit the job you’re working on. Sometimes good fences do not, in fact, make good neighbours.