In Reply to: The margin of error in an average of polls posted by mh on August 07, 2026 at 16:40:24
who is being sampled correctly models the electorate.
Polls are generally very good in elections that do not have Trump on the ballot, although the last Harris–Trump polls were not particularly bad, despite, off the top of my head, having about a 2% bias toward Harris.
I have no evidence for this yet, but I strongly suspect we are seeing another Trump-induced change in the composition of the electorate in favor of the Dems.
Since Democrats have to run the table in all of the close races, the mathematics are against us. The only reason Kalshi has Democrats taking the Senate at barely below a coin flip is that it appears to be assuming the polls are all wrong in the Democrats’ favor.
They believe that the polls do not reflect who will actually show up to vote.