On Richard Lynn's European IQ Data

Richard Lynn gives IQ scores for several world nations here.

In 1981, Buj gave IQ scores for several European nations. Buj's study is particularly useful, since it used the same Culture Fair test over samples representing the same fraction of each country's population, in roughly the same time period and across different age groups. In other words, it accounts for many of the factors that make IQ scores different to compare against each other.

It should be noted that Buj's study is one of the studies used by Lynn in his tabulation of IQ scores. Hence, we expect some correlation to exist between the two just due to this fact.

So, it is extremely surprising to find out that the correlation of Lynn's reported IQ scores with those of Buj is only +0.51.

UPDATE #1

I wanted to see whether either Lynn or Buj's data have any relationship with per capita income.

The regression of Lynn's reported IQ scores with 2003 per capita PPP is barely significant (P=0.046) with a correlation of +0.44.

On the other hand, with Buj's IQ scores, no correlation exists (P=0.87).

It is perhaps too convenient that Lynn's IQ scores differ from Buj's precisely in the direction that supports Lynn's theory.

What is more surprising is that Buj's reported standard deviation of IQ is an even better predictor of per capita PPP with a -0.64 correlation (P<0.01). So it seems that IQ homogeneity predicts wealth more than mean IQ does.

UPDATE #2

It is obvious that countries with a communist past have lower per capita PPP than countries with a capitalist past. So, I encoded this with 1: capitalist, and 0: communist. Indeed, this variable predicts 76% of the variance in income in Europe.

Buj's standard deviation of IQ + Communism predict 84% of the variance in income in Europe, whereas Lynn's mean IQ + Communism predict 78% of the variance.

DATA

NOTE: I had placed this entry online for a short time and then pulled it to correct some errors. Make sure you've read the final version.

UPDATE #3:

Infant mortality is again a much better predictor of income (in Europe) than IQ is. Its correlation is -0.70, significant at the P<0.001 level.

Posted by Dienekes at November 30, 2003 04:24 PM | PermaLink