Saturday, May 05, 2007

Surpassing your father: Reading the other day about how Robert Frost's son killed himself, it made me wonder if, in general, sons of more successful fathers are less happy. Using General Social Survey data, I calculated the educational difference between father and son. Knowing that absolute educational level is correlated with happiness (as well as having an interest in educated people), I limited the range by only looking at sons with at least a bachelor's degree. I also figured that small differences might not matter much, so I compared two groups: sons with 4 years more education than dad versus sons with 4 years less than their father's:

Percent of sons happy with 4 more years of education than father

Very happy 34.6
Pretty happy 57.9
Not too happy 7.4

Percent of sons happy with 4 fewer years of education than father

Very happy 26.5
Pretty happy 57.4
Not too happy 16.2

While the difference is not huge, sons that did not go as far as their fathers are not as happy. This may be due to the expectation that, with all of their advantages, sons will surpass their fathers. If they don't, it must because they failed to apply themselves.

This is an example of how people are harmed by a belief in environmental determinism and by not understanding statistics. When I tell my very best students that their kids may not do as well as them due to regression to the mean, they look at me like I'm crazy and evil. If you think about it, if you were Robert Frost's son, why in the world would you think that you could match or surpass such a freak of nature? Our culture would be much healthier if we acknowledged and made peace with our natural limitations. It's okay to be a regular guy. Thank God for your life and health.

1 comment:

Tim said...

I finally figured out why we see the "regression to the mean" effect for IQ. When parents with high IQ have kids, the IQ of the kids tends to be lower than the parents, and to move towards the average IQ for the parent's genetic heritage.

Take a gene that gives higher IQ. A lot of these cause problems when they are homozygotes (e.g. when both copies are the same), because they overexpress some protein or cause the total loss of the protein encoded by the typical alleles. If one of the parents has one of these genes, then the kids have a 50% chance of having it or not having it. So half the time, a child will not get the benefit of this gene on IQ and will have lower IQ. If both parents have one copy of the gene, then 1/4 the time a child will not have it, and 1/4 will have it from both parents which can cause problems (or could be even better than just having one copy).

Because the genetic component of IQ is due to the effects of a number of genes, a child with much-smarter-than-average parents will typically be smarter than average, but not as smart as his/her parents.

The caveat here is whether the parent's IQs come from the same set of genes. If the parents both have high IQ but their IQ comes from different sets of genes, then a child could get lucky and get the best genes from both, and have the potential for very high IQ. If the overlap in the parent's genes that contribute to high IQ is small enough, then the children have an excellent chance to be as smart as or smarter than his/her parents.

You may say "Well, doh!" at this point :-)