Assortative mating and IQ: Assortative mating for IQ spreads out the range of IQs, producing more high-end individuals. Estimates in the 1970s indicated that the American level of assortative mating accounts for more than 50 percent of the population with IQs over 130, and 80 percent of the people with IQs beyond 145 (Jensen. 1981. Straight Talk about Mental Tests, p. 90).

Has assortative mating increased since that time? The GSS gives us a good measure of a respondent's IQ (WORDSUM); unfortunately, the best we can do for the spouse is educational level. I looked at people in their 30s. All correlations are based on samples of at least 1,000:


Correlation between IQ and spouse's education

1970s .47
1980s .39
1990s .37
This decade .30

At first glance, this looks terrible. We see a big drop in assortative mating over the past four decades. But upon closer inspection, I suspect the shrinking correlation is due to the declining validity of education as a measure of intelligence. Here are the correlations between Wordsum and respondent's level of education for people in their 30s for each of the past four decades:


Correlation between IQ and one's own educational level

1970s .59
1980s .54
1990s .48
This decade .38

Wow. Schooling used to be indicative of intelligence. The validity coefficient for this decade is low: one's educational level just doesn't mean much anymore. Evidently, the decline in the correlation between the spouses' IQs we are observing is due to educational level becoming a lousier and lousier proxy for intelligence. Perhaps real assortative mating hasn't changed much.