I have a present for you all! One of the most useful tools we had for reporting at 538 was a dataset we maintained of every poll our research team came across, even if it didn't have horserace questions, starting after the 2020 election.
The last few months I've been seeing quite a big change in the generic ballot polls, so out of curiosity I went back to the last couple of cycles to see what kind of story the generic ballot might be telling in the early days of each presidency.
I'm not really upset that they wrote this article. It's a fine sort of model to consider. But the confidence with which they predict what is a clearly uncertain future is pretty gross to me. We would NEVER have used language like that at 538. (8/)
Beyond just Trump's job approval, I think this sort of thing should be a flashing warning sign to congressional Republicans. Although they're in a tough spot: go against Trump and maybe he takes you out, stay with him and maybe the voters do. Tricky.
We diligently logged each poll in every state, congressional district, major city or county, and nationally. We also included topic tags so that it would be easy to find whatever sort of data you'd be interested in.
But most striking to me here is the movement. In both 2017 and 2021, not much change in generic ballot over the first 100 days. This time though? A shift of over 6 points away from the party in power in the first few months.