This presents a good enough opportunity to point out what has caused me some low-level cognitive dissonance for years now–there isn’t much evidence in the GSS for assortitative mating, at least when it comes to vocabulary (and thus presumably IQ).
First, the following graph shows the mean Wordsum score among whites born in the US who were between the ages of 30 and 65 when they participated in the survey, by the decade they were born in. That’s a mouthful, but
here comes an even bigger one it is so to avoid racial confounding, language fluency issues, and to allow respondents to have had at least three decades to build their vocabularies without pulling in people who have been beset by the cruel cognitive decline senescence brings:
The early boomers look pretty good, but there isn’t much variation across cohorts. There’s little to glean here with regards to the Flynn effect. That’s not to suggest the effect is oversold–vocabulary is one of the areas of intelligence least effected by the, uh, effect.
What does stand out, though, is the reduction in the size of a standard deviation in scores by cohort over time:
The size of a standard deviation in scores has declined by one-quarter in little more than a generation. That is, there has been a bunching of Wordsum scores towards middling performance over time, with both especially high and especially low scores declining as a proportion of all results. Among those born in the 1940s, 1-in-7 either completely bombed or perfectly aced the test. Among those born in the 1980s, just 1-in-25 either bombed or aced it:
This is the opposite of what would be expected to happen if assortitative mating was increasingly leading to cognitively gifted men pairing up with cognitively gifted women to have cognitively gifted children and dull men procreating with dull women to have dull children.
Parenthetically, because the quiz is multiple choice with five possible answers per item, scores of two or fewer correct answers are expected from participants ignorant of any of the correct answers who merely guessed randomly on all of them.
GSS variables used: ETHNIC(2,7-15,17-19,21,23-27,32,35,36), WORDSUM, COHORT(1900-1929)(1930-1939)(1940-1949)(1950-1959)(1960-1969)(1970-1979), BORN(1), AGE(30-65)