🚨 NEW today is the @EchelonInsights#SplitTicketAtlas updated for 2022: Almost 100 maps showing unique voter coalitions by state, and answering who performed better where: Trump, or Republican gubernatorial and Senate candidates in 2022.
Governors' races are new to the most recent version of the #SplitTicketAtlas. A particularly telling slide: @BrianKempGA improved over Donald Trump virtually everywhere in GA, but not in places like Henry County, which have seen rapid demographic change in the last 2 years.
In Vermont, @PhilScott4VT had overperformances of as much as *101%* in margin over the Republican candidate on the ballot in 2022.
I initially thought this Idaho map had to be wrong, since it looks so much more random than other states. It wasn't: Just the effect of Ammon Bundy as a third-party candidate for Governor.
Nebraska is an interesting example of a state that seems split 50/50 regionally, but a lot of these orange Trump-first counties are very sparsely populated.
Lessons? Even if split ticket voting is on the decline, we saw in 2022 that it still matters. But when planning for a campaign you shouldn't assume over or underperformances will be spread out equally across the state.
🧵 A few quick notes for posterity to close the loop on those interminable polling error debates from before the election that have now seemingly been memory-holed. From our final MI/PA/WI surveys...
We were seeing much the same issues as @Nate_Cohn was with raw samples coming in much more pro-Dem across the board.
Without weighting on recall vote, MI & WI would have been Harris+ high single digits.
@Nate_Cohn .@Nate_Cohn highlighted the reasons not to do recall vote, which I'm sympathetic towards, but imo there's no way not to do it in states with no/minimal party data on the file, like MI & WI.
PA didn't change w/ recall vote because party reg data is so robust.
In an age of extreme partisan polarization, you’ll never get one candidate up by double digits in a way that tells the other one it’s hopeless. It’s always “just a standard polling error away!”
When average leads never get above 5 points it’s super-easy to gaslight people by pointing to one showing a 2 point race and then comparing it to this other one showing the same before the debate.
A 5 point gap today is the same thing and as hard to recover from as 15 points in the old days.
But leads within 5 are almost always “within the margin of error.”
Trump should skip doing some swing state rallies in order to attend every UFC event between now and November to remind young nonwhite men that there’s an election happening.
Yard signs can help in local races.
For every Karen we lose, we pick up two Julios and Jamaals
Why is Biden focusing more on his base than swing voters?
A polling experiment finds some answers.
At first blush, Biden appears to have a higher ceiling than Trump. 43% say there's ZERO chance they'd vote for him, 48% for Trump.
For months, we've been asking people to rate their likelihood of voting for Biden or Trump on a 0-10 scale. We then compare the answers to come up with an alternative to the traditional head-to-head ballot, assigning voters based on their probabilities.
This nuanced view of intensity shows why Biden had an upswing in the polls: voters totally committed to him (Biden 10, Trump 0), went from 22-23% of voters to 27-28% starting in March, matching Trump's hardcore group.
Before, there was probably doubt Biden would be the nominee.