The Latino shift to Republicans is real, but it's (mostly) not about whether someone has a degree | #203 – July 24, 2022

Thoughts on "education polarization" and a new study claiming Hispanics did not move right at all from 2016 to 2020

I like to think that things happen for a set of reasons — that causes have effects and effects have causes. Lots of other people think this too. Do not jump down the stairs (or you will break your ankle). Do not bite the hand that feeds you (or you will not get fed). If I run really fast I will get tired.

But while humans are pretty good at accurately identifying causality in the real world (I know that if I break my ankle by jumping down the stairs then it is partly my fault for doing a stupid thing, and also partly gravity’s for pulling me down), it is not so easy to figure this out with data. That is why your next ice cream cone probably won’t murder you.

There are several aspects of political data analysis that muddy the waters of causality. I was reminded of this by an ongoing debate over why Donald Trump did better among Latino voters in 2020 than he did in 2016. Because we cannot ask every American why they voted, we have to guess their reasons for doing so by looking at relation…