Tuesday, July 25, 2006

The "Flynn effect" observed in GSS data: Now we can't get too excited about measuring IQ with 10 vocabulary questions, but what the heck--this is a blog, not Intelligence. The GSS has been asking vocab questions since the early 70s, so I averaged the number correct out of 10 for blacks and whites. Here they are:

70s 4.59
80s 4.69
90s 4.97
00s 5.15

70s 6.17
80s 6.14
90s 6.33
00s 6.39

So we see increases for both groups, but it's larger for blacks. While whites increased 4% over the period, black averages rose 13%. One bit of evidence that this measure is not just nonsense is that we see the consistent, large gap between blacks and whites, like a guzillion IQ studies have shown. And there has been some narrowing of the gap over the period.

Now it's possible that the words got easier--I don't know if they used the same words every year, and it's too late for me to look the damn thing up. I will say that if the test got easier, then why did whites only improve between the 80s and 90s while blacks improved every decade? And why did blacks improve more? I thought as people age, they might gradually pick up more words, and while it's true that mean white age rose 2.6 years over the period, blacks dropped .4 of a year. And whites aged gradually from the early 70s until now, while their vocabs only improved between the 80s and 90s. Comments are appreciated, and maybe I'll try to see what explains the increase later.


Tino_G said...

There probably is something to it, but note that vocabulary is likely to be low G loaded, and more a function of education than IQ. No doubnt Black education has increased over time.

Jason Malloy said...

Now we can't too excited about measuring IQ with 10 vocabulary questions

The WORDSUM measure in the GSS is a valid measure of IQ, that correlates very highly with other IQ tests (.70+). I find the narrowing here a data point with importance.

Jason Malloy said...

but note that vocabulary is likely to be low G loaded

This is absolutely false. Vocabulary is actually one of the highest g loaded tests.

Anonymous said...

Let me rephraze that. Changes due to more education are likely to have little do to with changes in G. You can without a doubt increase vocabulary with education, however that will not change G.

Half Sigma said...

For white people, nearly all of the increase comes from respondents o50 or older, and it's even more pronounced for respondents over the age of 60.

The percent of white respondents under the age of 50 scoring a perfect 10 seems to be declining.