Heritability of IQ

There’s two surprising things about this table:



















The first is the zero correlation Prof. Caplan points out between IQ of adopted 16 year olds and their adoptive parents (the left column). The second is the low correlation between the IQs of natural born children and their parents (the right column).

6 thoughts on “Heritability of IQ”

  1. Not a statistician, but 285 seems like a smallish sample. Are there other similar studies that corroborate these findings?

  2. That’s actually sounds like an ok sample to me, but I’m not much of a psychometrician.

    Check out the paper for discussion of other papers…

  3. WordPress ate my earlier comment.

    For a study with twice the sample size of adoptees (which also links adoptees to both their adoptive and biological parents) see this paper:

    They’re looking at educational attainment but if the question you’re interested is “does a child’s environment affect them in any meaningful way?” then this paper is relevant. And it answers with a yes.

    So even if environment doesn’t affect “inherent human capital,” it does affect total human capital which is what we care about anyways. This is the paper that should be getting the press.

  4. If one believes educational attainment is an important socioeconomic indicator AND the relationship between that proxy and actual socioeconomic status is constant across family types, then that paper, Jason, should be getting press. Assuming those things are true (and I’m more likely to believe the first is true rather than the second), then the paper shows post-birth effects (environment) are about half as important than pre-birth (genes) effects. Like in the IQ study cited by Prof. Caplan, the effect size even for biological parents is small in this study. If both biological parents get a college degree (vs. both not), then the kid is only expected to complete less than one year of college. Not a huge effect.

    In any case, its the second assumption that bugs me. It could be that smarter, and likely better off, adoptive parents are more likely to send their relatively stupid adopted kids to school who do poorly there (which, in turn, means their earnings outcome isn’t much effected beyond the sheepskin effect). So for these adopted kids we see higher educational attainment but not higher economic status.

    I would find earnings regressions more compelling, but the paper you cite reports mush for earnings on education of parents. As the authors say, this is because one year’s earnings is a poor indicator of socioeconomic status. I agree. Isn’t the Swedish data a panal? Couldn’t they just calculate lifetime earnings instead of using earnings from one year?

Comments are closed.