Thursday, April 01, 2010

Regression to the mean with GSS data

Do black and white adult children of parents who attended graduate school have the same average IQs? Why shouldn't they? They had the same advantages that come from growing up in a home of highly educated parents.

Using GSS data, I calculated mean IQs (based on a vocabulary test) for blacks and whites whose mothers or fathers completed at least 17 years of schooling (20 years is the highest possible score).

Mean IQ (N = 546)

Has highly educated mother
White 107.0
Black 99.5*

Has highly educated father
White 107.6
Black 99.7*

* significantly lower than white counterpart

Even though these folks have equally educated parents, there is a gap between them that is roughly half of a standard deviation--a fairly large difference.

Our results fit just about perfectly a regression to the mean equation where heritability for IQ is set at 0.5; the mean IQ for educated parents is set at 115; and the means for the white and black populations are set at 100 and 85, respectively.


  1. Wow, this is really good.

    Can you modify your results to incorporate an "affirmative action" effect whereby highly educated blacks obtain degrees in less demanding fields and thus require IQs lower than 115?

  2. Is the data 'thick' enough that you can break the number down for several other cases?
    For instance,
    Father with advanced degree, mother with an undergrad degree
    or by type of degree---perhaps as simple as Education vs all others?
    My experience on regression to the mean is that it really hammers 'outliers' (those being people that are a LOT smarter than the rest of their family) hard in terms of expectations for their kids. People from 'smart' families seem affected much less so.

  3. what is the 'n' of blacks in either group?

    are there enough to compare blacks and whites w/ educated mom+dad?

  4. Anonymous5:43 PM

    "what is the 'n' of blacks in either group?"

    It's like the 'g' of general intelligence, but when applied to blacks, the 'g' switches to 'n'. It's the 'n' factor, ya dig?

    When you're consider moving into a new neighborhood or city, you have to consider the 'n' factor.

    At least that's my theory. Or in this case, my theroy.

  5. Dear Ron,

    New article by me.

  6. Oops!




Meta-analysis of clinical trials: Eat walnuts

I am always looking for easy eating choices that are good for you. This new meta-analysis of 26 clinical trials looked to see if walnuts ma...