I wanted to share Scott Rosenberg’s analysis of this NY Times Poll that has everyone melting down today:
It’s A Close, Competitive Election – Yes, the NYT released a poll today that has Trump ahead. Some initial thoughts:
Lots of other polls show the race even, competitive – Three national polls released this week (below) have the race even. 538’s Congressional Generic tracker is tied, 44%-44%. A new battleground state poll produced by top Democratic pollsters has Trump and Biden tied at 40%. Another battleground state poll that hasn’t been released yet that I was just briefed on has it 41%-39% Trump, essentially the same results. Senate polling is slightly better for us than for Republicans right now.
The Times poll has Trump leading among likely voters by 4 points, 48%-44%. This is a gain of 6 points for Trump since the December Times poll. None of these other polls have found a GOP surge of this magnitude or even a GOP lead. So the NYT results are not confirmed in lots of other recent polling which finds the race close and competitive, which is where I think it is now.
Biden 36% Trump 36% – AP/IPSOS
Biden 44% Trump 44% – Economist/YouGov
Biden 43% Trump 44% – Morning Consult (Biden’s gained 4 pts in recent weeks in this poll)
There are problems with the NYT poll – It has Trump winning both Hispanics and women – an impossibility. It’s likely voter electorate is +3 Republican – something we haven’t seen in a general election in actual voting in 20 years, and only once in the last 8 Presidential elections going all the way back to 1992. In the last 4 Presidential elections, Democrats have averaged 51%, Republicans 46%, and we gained ground in the 2018, 2020, 2022 and 2023 elections. After all these years of strong Democratic vote performance for the electorate to become +3 Republican this year is, um, unlikely. Overestimating their intensity and strength and underestimating ours was a central reason so many saw the red wave that never came in 2022.
The poll has Trump winning all his 2020 voters, and keeping his party unified – something that is not happening either in the actual voting or polling in the early states. It has Dean Phillips at 10% in the Dem primary against Biden. In Michigan this week he got 2.5%, and came in behind Marianne Williamson who dropped out of the race a month ago.
The poll’s initial likely voter sample was 29% Dem, 28% Republican and was “weighted” through a complex formula to become 32% R and 29% D. That’s a shift of 4 percentage points, something that would take an even race and make it +4 R, as this poll finds.
The poll has Biden winning Democrats 90%-7%, Trump winning Republicans 91%-6%, and Biden winning Independents 45%-41%. These results would normally produce a Biden lead but with the aggressive weighting and a very Republican sample, it produces a 4 point Trump lead.
The poll only interviewed 980 people, which is a relatively small sample for such an influential poll. Its margin of error is 3.5% for registered voters and almost 4% for likely voters. The sub-samples have margin of errors in double digits.
I want to thank my friends Tom Bonier and Joe Trippi for their help in putting together this quick analysis.
Now go have a nice evening.