Sunday, December 08, 2019

What is the strongest correlation I've ever estimated?

Everyone who does social research knows that correlations about people are typically weak. You suspect that IQ predicts criminality, but then the data tell you the association is a mere .2. This is the rule rather than the exception. People are complicated. You can't reduce their behavior to a single factor. Plus, measurements are far from precise.

But on rare occasion, connections can be strong. The correlation between your education and your spouse's is .6 or .7.  The link between number of delinquent friends and one's own involvement in delinquency--about .6. Pretty good.

The largest individual-level correlate I've ever calculated (macrolevel correlations tend to be bigger) deals with sexual attraction: Using data I collected myself on 330 people, the point biserial correlation between being male and level of attraction to females is .82. For women being attracted to men, it's .84. Those are huge numbers.

Another way of describing it is in terms of standard deviations: the gap between male and female attraction toward females is 3.6 standard deviations. The difference between the two sexes on liking males is just as big--3.6 sds. You've probably heard that the black-white IQ gap is big. It is, at ONE standard deviation. The attraction gap is enormous.

Now you might be saying, duh, we would expect men to like women and women to like men.

Well, you might expect that, but then again you don't have a PhD in Gender Studies. Many years after the idiot Kinsey claimed that sexual attraction is a continuum, not categorical, researchers are still making similar claims.

With a continuum, one assumes that males will range from being strongly attracted to females to being strongly attracted to males, but most men will be somewhere in the middle with various levels of attraction for both sexes. The same for women. Most of them will have a mix of attractions. If this were the case, being male would not be a strong predictor of level of attraction to females. But common sense and my data predict the outcome very well: what do you know, men like women! And women like men!

I'm afraid a great deal of social research makes us dumber, not smarter. I'd love to see a great deal of it scrapped.

UPDATE: The latest sex theory I've come across claims that our ancestors were indiscriminate copulators: They would hump anything that moves. So males and females being attracted to each other was not the original system. With its de-privileging heterosexual sex, I predict the theory will be wildly popular.


  1. Anonymous11:34 AM

    Saying it's a continuum doesn't imply the average is in the middle though.

  2. If Kinsey wanted to help people see attraction most accurately, he wouldn't have chosen a continuum. He would have presented a heterosexual model plus noise.

  3. "3. The Marriage Gap: As I first reported in VDARE.COM last December, the single best correlation with Bush's share of the vote by state that anybody has yet found is: the average years married by white women between age 18 and 44: an astonishing r-squared = 83 percent.

    (This has to be one of the highest r-squareds for a single factor ever seen in political science.)"


GSS data: Trump-voting women are happier and here's why

Feminism is a form of liberation that should make you happy, right? Wrong. Here is the relationship among white women between being happy an...