On Sons and Daughters

Over at the Freakonomics blog, I learn that

nurses, social workers and kindergarten teachers—those with “empathic” traits—also had more daughters than sons....scientists, mathematicians and engineers are more likely to have sons than daughters.

I wonder: Could this result be related to the finding of Andrew Oswald that

how parents vote is linked to the gender of their children. The more daughters there are in a household, the more likely the parents are to vote Labour or Liberal Democrat.

When I first learned the Oswald result, I concluded that having daughters must have a causal effect on political preferences. (This seems to be Oswald's interpretation as well.) That inference was based on the assumption that sons and daughters are distributed randomly. But perhaps that assumption is wrong. Maybe having daughters is instead a proxy for other personality characteristics that influence a person's politics.

Update: The study cited by the Freakonomics blog is, I just learned, not without controversy. See the critical comments of Columbia University statistics prof Andrew Gelman.