Elise Kramer deconstructs a misleading statistic by Maureen Dowd that asserts: “A 2005 report by researchers at four British universities indicated that a high I.Q. hampers a woman’s chance to marry, while it is a plus for men. The prospect for marriage increased by 35 percent for guys for each 16-point increase in I.Q.; for women, there is a 40 percent drop for each 16-point rise.”
What did the researchers find? According to their regressions, the odds ratio of marrying was 1.35 for each one-standard-deviation (S.D.) increase in I.Q. in men, and 0.42 for each one-S.D. increase in I.Q. in women. In layman’s terms, this means that the marriage odds of a male with an I.Q. one S.D. above the mean compared to a man with a mean (100) I.Q. are 1.35 times higher — or 35 percent higher. It also means that Dowd read the statistics wrong: in fact, in the sample the researchers used, women were 58 percent less likely to marry with each S.D. increase, not 42 percent. Ouch. (Also, the researchers used an I.Q. scale with a S.D. of 15, not 16 as she states, but let’s deal with one thing at a time.) The researchers controlled for social class and height and found that the effect for men became non-significant — but the effect for women, though attenuated, remained significant.
Sobering statistics — but what do they tell us? Unfortunately, not much. This is the problem with social science research: although it’s possible to show very convincing correlations, when you’re researching with humans it’s usually impossible to determine causation or mechanism.
This means that the interpretation of findings requires quite a bit of speculation. Here’s an idea: maybe women with higher I.Q.s are less likely to get married because they’re less likely to want to get married. Maybe they are, on average, better off financially and better able to support themselves without a husband. Maybe they don’t have time for marriage and kids because women, unlike men, are often forced to choose between career and family. Or maybe more-intelligent women tend to have more progressive views and are less likely to subscribe to the traditional institution of marriage, instead cohabitating with their partners or avoiding long-term relationships altogether. Or maybe the sample used in this experiment, currently octogenarians, are representative of earlier times with earlier mores, when women who thought there was more to life than raising kids were prescribed medication.
My take: I am always amused when social scientists use IQ results to do analyses. An IQ test is a clinical tool used mainly to help diagnose the mentally retarded. Frankly, it’s hard to do cross-section or longitudinal analyses without having to deal with lots of secondary social issues. I don’t have experience with regression-analysis, but some things come to mind: does the significance of an IQ result change over a person’s lifetime? For a 10 year old, an IQ test may have some usefulness for predicting academic success, but for a 45 year old, what exactly does it predict? Also, I’d be curious as to how this so-called trend overlaps with the demographic trend of age’s impact upon marriage. Age makes it hard for women to find marriage partners anyway. How does this factor compare with IQ? The statistic could simply be saying that getting an advanced degree delays marriage, which in turn lowers the overall marriage rate. Other marriage trends may come into play as well. In the US, people are marrying later, perhaps in response to increasing divorce rates (Older marriages tend to result in lower rates of divorce). Could it be that the timespan of these study participants coincided with increasing divorce rates, making the delaying strategy a sensible one from the standpoint of marital stability?
Another criticism of the statistic by Caryl Rivers and Rosalind C. Barnett.