NYT/Siena Polls of 6 Battleground States

[Wiz in Wis]

1) I think these polls are all believable, and people shouldn't freak out about minor differences between states. Sampling error and unexpected shifts will happen. Maybe AZ and FL are shifting more D than we think, and maybe PA is less D than we think, we don't know.

2) Looking into crosstabs is a terrible idea. Those are even smaller margins of error. Don't look at one weird result and be like "The whole poll is trash!"

3) I do wonder if these polls are really "good news" for Trump. He's lower in these states, even against Warren, than he was in 2020. Undecided at his point aren't likely to break for him, unlike 2016.

[Wiz in Wis]

Also how can Warren be doing worse in Michigan then Wisconsin. Trends wise that makes no sense

In the methods write up, Nate Cohn notes that the Michigan poll had a lower than expected completion rate (500 completes vs goal of 650) and higher variability due to post-weighting, making the Michigan numbers a bit more wonky than for the other states.

Not that a difference of a few points is actually a big deal in a poll. That will just happen.