One question I get fairly frequently is what an economist is doing studying people’s sexual behavior, and other topics related to public health. There are lots of reasons, but a significant one is that the public health topics I work on (HIV, mostly) are interesting and important, and, moreover, that a lot of quantitative research that comes out of public health schools on these topics is bad. Like, really terribly bad. There’s definitely plenty of very good researchers in that field, but poor-quality empirical work in public health is common enough to be standard water-cooler fodder for certain groups of statistics nerds. The most common offense is for researchers to ignore the title of this blog, and to take two groups that differ in some specific way and attribute any differences in an outcome of interest to that difference. Want to conclude that alcohol makes you smart? Just compare drinkers and non-drinkers in terms of IQ. The former are, on average, smarter. But that’s a ridiculous conclusion; if anything, alcohol kills brain cells and makes you stupid. The problem is that, in the words of my advisor, a lot of other stuff might be going on. The predominant reason for becoming a teetotaler is religion, and higher levels of education tend to decrease religiosity as well as boosting measured IQ. If you don’t account for this, you will get the wrong answer. Unfortunately, this is a great way to make the front page of your university’s news wire/PR wing (every school has one, seriously) and hence the headlines of internet news sites. Chocolate prevents cancer? Fiber recuces your risk of bowel cancer? No, your study is terrible.
The latest offender of this kind comes to me via facebook, and alleges that exposure to semen improves women’s mood and health. The study’s methodology, according to this story from *The Sun*, was to use self-reported unprotected sex as “an indirect measure” of women’s exposure to semen, and look at health and scores on the Beck Depression Inventory. But condom use is related to all sorts of other things – the quality of one’s relationship, trust in one’s partner, STI and HIV status, and hormonal contraception use, to name just a few of the many potential omitted variables here.
Now, maybe that summary is just a poor description of the study, right? No; other than the titillating picture the *Sun* article is basically a paraphrase of the abstract. (Interestingly, this story was picked up just recently but the research is 10 years old). That link is gated, but I managed to dig up this copy of the PDF.* To the authors’ credit, they actually do control for some of the factors named above and find that condom use still correlates positively with depression in women. But that should go in the abstract! In any case they don’t adjust for trust or relationship quality. Here I can’t blame them – those are very hard things to measure on a survey. And they don’t consider the single most important factor: unprotected sex is more fun, feels better, and is more satisfying than sex with condoms. That is going to irrevocably confound any study that uses unprotected sex as a proxy for semen exposure to look at effects on depression – at least as long as it looks at women alone. If the hypothesis is just about semen, and we don’t think that vaginal lubricant has similar effects, you could just look for differences in these results across men and women. I submit that a study of men would find nearly-identical results.
Is it possible that semen improves women’s moods? Absolutely – given the relevant biology, it even seems likely that there’s some effect. But this study teaches us very little about whether there is an effect and how large it is likely to be.
*I did this, and wrote and posted this article, while sitting in my project’s minibus at Mpyupyu trading center, way out in rural Zomba district. Technology is awesome.