The value added by election forecasters is mainly to aggregate polls. If we compare the raw polling average to Nate Silver’s Bulletin and SplitTicket, we see that the raw polling average I posted on November 4th performs as well as forecasts that incorporate many more variables.

Before the 2024 election, I put out a simple forecast that accounted for the within-state, across time polling errors that have persisted since at least 2000.

As it turns out, simple adjusted polling average outperforms the two best election forecasters from 2024.

x.com/danielfirooz...

I also make use of a top percent admission policy at UC campuses called eligibility in the local context (ELC). Essentially, students in the top 4% of their HS class had an admissions advantage. They end up less likely to register to vote as Republicans when they reach the age of roughly 30.

I use compulsory schooling laws (CSL) that bind on date of birth to identify the impact of high school education in Florida and California on partisanship. People born before the cutoff start school younger and must remain in HS longer before they can legally drop out. For them, GOP affiliation ⬇️.

Does educational attainment impact partisanship? Yes.

In my working paper “Education and Partisanship”, I use multiple RD designs and admin data on millions of voters to show that education reduces Republican affiliation across generations, settings, institutions, and demographics.

Results below!

Yup!

Had to explain to my partner that the reason why I was excited after checking the mail today was because we were randomly selected into the ACS.