Byungkyu Lee and Dalton Conley write:

Recently, the sex of child has been widely used as a natural experiment and shown to induce the change of the allegedly stable political predisposition, however, prior results have been contradictory: in the U.K., researchers found that having daughters leads to parents favoring left-wing political parties and to holding more liberal views on family/gender roles, whereas in the U.S. scholars found that daughters were associated with more Republican (rightist) party identification and more conservative views on teen sexuality.

Yes, we discussed this last year.

In this new paper, Lee and Conley analyze more data and summarize:

In analysis of 36 countries, we obtain null effects of the sex of the first child on party identification as well as on political ideology.

What makes this paper particularly interesting is that one of the results they’re disagreeing with comes from . . . an earlier paper by Dalton Conley and Emily Rauscher.

I admire that Conley is not afraid to shoot down his earlier work. As he and Lee note in their new paper, these are very small effects being studies in a noisy and biased context, so lots of things can turn out statistically significant in particular datasets without telling us anything useful about the general population of voters.

Just one thing: Near the end, Lee and Conley write of “the paradigm suggesting that political identity formed early in childhood (largely through parental socialization and/or genetic inheritance) and was subsequently immovable.”

This is certainly not the paradigm of how we think about things in political science! We have evidence that political attitudes are strongly influenced by what happens to you between the ages of 15 and 30. See this recent paper with Yair Ghitza, for example.