Last week CNN pulled a story about a study purporting to demonstrate a link between a woman’s ovulation and how she votes, explaining that it failed to meet the cable network’s editorial standards. The story was savaged online as “silly,” “stupid,” “sexist,” and “offensive.” Others were less nice. Most of the vitriol was directed at CNN and at the reporter, Elizabeth Landau, who pointed out on Twitter what should go without saying: She didn’t conduct the study.
The person who did conduct the study is Kristina Durante, along with two co-authors, Ashley Arsena and Vladas Griskevicius. Durante, an assistant professor of marketing at the University of Texas at San Antonio, whose doctorate is in social psychology, often writes about how ovulation affects women’s choices. Another forthcoming study of hers is titled “Ovulation Leads Women to Perceive Sexy Cads as Good Dads.”
The how-ovulation-affects-voting study has been dissected and ridiculed by science bloggers, including Adam Marcus of Retraction Watch who called it “poor science, poorly reported.” The blogger Scicurious wrote that the study “bases its hypotheses on assumptions about the menstrual cycle for which they have no proof and ignores some major confounds in the data.” Kate Clancy, a blogger and assistant professor of anthropology at the University of Illinois at Urbana-Champaign, wrote a takedown of the study, saying that it “requires assumptions about human physiology that are patently false.”
It was, in other words, not well received.
Durante sent out a statement defending her research as “relevant and timely” but acknowledged that it may “prove uncomfortable for some.” She didn’t respond in her statement to any of the specific criticisms, but over the weekend she agreed to answer some questions via e-mail about the study and the reaction to it. Here they are:
Q. Can you describe the feedback you’ve gotten since the CNN article about your study?
A. While much of the feedback has been overwhelmingly supportive, a substantial portion of feedback has been filled with anger and upset about the study and its findings.
Q. Did the article fairly summarize your findings?
A. The CNN article highlighted the basic findings of the study as is often the case with media coverage of scientific work. This unfortunately leaves room for misinterpretation.
Q. Your methodology has been criticized by a number of science bloggers. Among the criticisms is that you were asking women their opinions only once, so you don’t know whether, say, a specific single woman is more likely to vote for Barack Obama when she’s ovulating than when she’s not. Isn’t this a weakness of the paper?
And because your respondents were answering questions for an Internet survey, you had no way of testing these women’s hormone levels, so you couldn’t really know their fertility levels (or whether they were ovulating). Instead you asked about the dates of their recent menstrual cycles in an attempt to estimate when they were ovulating. Is that a reliable method?
A. The method used in the study is quite reliable. We looked at over 750 women (a sample size much larger than that used in other ovulatory cycle research) and compared the women who were most fertile—near ovulation—to those who were at a lower fertility point in their cycle. A large body of research looking at ovulatory effects on behavior has used this between-subjects methodology.
We also looked at political attitudes as a function of conception probability. This is a continuous measure of fertility that is used often in ovulatory research.
A large online panel was used as a starting point (as opposed to direct hormonal assay and measuring shifts within the same woman), because I wanted to obtain a large and diverse sample of women of different ages, relationship status, religion, socioeconomic status, etc. for this particular study. Results using estimation and direct hormonal assays generate very similar results.
Q. The blogger Scicurious notes that you write that “single women strongly prefer the more liberal candidate, while married women prefer the more conservative candidate,” yet the data from the survey show that a strong majority of both married and single women voiced support for Obama (though single women were more supportive). How do you account for this?
A. In general, regardless of fertility status, most of the women in our sample were leaning toward Obama. The differences only arose near ovulation.
Q. Do you stand by the paper?
A. 100%. This is theory-driven science. This paper reflects what my data showed in two studies. Now the door is open to build on this research line, replicate, and expand upon it.
The study is forthcoming in the peer-reviewed journal Psychological Science.
UPDATE: So several people -- including Kate Clancy, mentioned above -- think this post breezes too quickly past the lengthy, substantive criticisms of Durante’s study. I think they’re right. Though those posts are linked above, most readers probably won’t click through. So here are some excerpts:
They ask “could the hormones associated with ovulation account for some of the discrepancy between single versus married women?”
I’m not sure what that means. First off, they didn’t take any hormone levels. It was an internet survey. No hormone levels. Secondly...why would the hormones associated with ovulation account for DIFFERENCES between married and single women? Wouldn’t they, rather account for some of the similarities? Are they stating that estrogen and progesterone have entirely different effects on behavior due to whether or not there’s a committed relationship? They can suggest that and that’s fine, but there’s no data to back that up, and no studies have been done on this (and certainly none that are cited). And it certainly runs contrary to much of the behavioral literature on the topic.
6. Let me repeat. The authors here say that political views differ based on ovulatory phase...but they have no proof of ovulation. It was an internet survey at a single time point (what we call a cross-sectional study). They did not take hormone levels. The likelihood of ovulation actually occurring...is much lower than you’d think. This point, and the one I am about to list below, make me think these data are mostly artifact.
7. I’ll repeat again. The authors here say that political views differ based on ovulatory phase...but they do not examine the same women over time. This is a cross-sectional study, a single time point. They did not ASK any of these women whether their political views differed as a function of ovulatory phase because they did not ask them during different phases of the cycle.
There are two other problems with the ways Durante et al conceived this study. First, in their sample single women were younger, poorer, and had fewer children than women in committed relationships (and remember, they had somewhat loose definitions for both). Age, parent status and socioeconomic status all play roles in political preferences. When you have multiple differences between two groups, but you want to see if only one of those differences is meaningful (in this case, relationship status) there are very easy ways to control for those variables statistically.
Durante et al didn’t do this. So the differences found between the two groups could really be about income or age or being a mom, not relationship status. They didn’t actually test their hypothesis.
Finally, beliefs change over time for many people. But do they change for women over the course of a month, over and over again? And what about pregnant or lactating women? Can they even be trusted to do any important or interesting human behavior with their lack of a menstrual cycle? Or dear god, what about the adolescents? Their cycles are all over the place! It must be because we can’t predict their cycles that they don’t have the right to vote! And don’t even get me started about those postmenopausal grannies with hardly two molecules of estrogen to rub together.
Read both of their critiques (here and here). I think it’s hard to do that and make some of the comments people below have been making. Also, it’s worth reading the actual study. I linked to it above, but some people missed it. Here it is again. As for whether Durante completely answered the questions above, I think readers can judge that for themselves.