If Biden beats the polls in Alaska, what should we predict in New Jersey?
Thinking about a weird problem in 538’s election model
Building predictive models for elections is tough stuff. There are so many different factors to take into account, and making an adjustment for one of them can cause weird downstream effects in how you adjust for another.
That’s why it’s not a huge surprise, and not an especially poor reflection on 538, to find a pretty big error in their election model…