1 Comment
Jan 26·edited Jan 26

Congratulations to Echelon!

And I'll further note that if 538 were to reduce their weightings a bit on N, Echelon with just ten polls would move up (I'm not a polling expert to say the least - if polling were to be a strongly stationary process, you'd expect MSE to go with sqrt(N) in the limit). See 538's own "Skill and luck play a role in pollster ratings".

Who is the #2 partisan/campaign poll (I assume Data For Progress, a good example of 538 probably overweighting N, is #1)? Hart?

Expand full comment