Read on Twitter
(Thread: 1/6)
So, I hate to be critical of the NY Times, because I am a poll junkie and they're en route to providing a ton of it, for which I'm grateful.
But I'm having a ton of problems with this passage from the writeup this morning on the new Siena/NYT polling in Wisconsin:
So, I hate to be critical of the NY Times, because I am a poll junkie and they're en route to providing a ton of it, for which I'm grateful.
But I'm having a ton of problems with this passage from the writeup this morning on the new Siena/NYT polling in Wisconsin:
So, there are several problems with this observation.
But let's start with a basic one. As was confirmed yesterday by Nate Cohn, who is the big dog as it relates to NYT polling, this set of polls screened for likely voters, which their June polls did not. A key difference. (2/6)
But let's start with a basic one. As was confirmed yesterday by Nate Cohn, who is the big dog as it relates to NYT polling, this set of polls screened for likely voters, which their June polls did not. A key difference. (2/6)
A subsequent Cohn tweet this morning said the move from RV to LV cost Biden 1.3% on average (though the specific toll in Wisconsin was clear).
So, roughly a quarter of the drop off for Biden between June's poll and now could simply be the result of changing LV/RV screens. (3/6)
So, roughly a quarter of the drop off for Biden between June's poll and now could simply be the result of changing LV/RV screens. (3/6)
Inexplicably, the change in LV/RV screens is never mentioned. Instead, the shift is credited to "improvements" in the economy. The problem: their own poll refutes that as the cause for the shift.
In June, WI voters preferred Trump on the economy 50-43.
Today? 51-43. (4/6)
In June, WI voters preferred Trump on the economy 50-43.
Today? 51-43. (4/6)
So something shifted, but it wasn't voters' views on the economy.
What shifted? The sample did!
The June NYT sample was Trump +1 in 2016, which is fair, given that he won it by less than a point.
This sample? Trump +6. It should be pretty evident why that matters... (5/6)
What shifted? The sample did!
The June NYT sample was Trump +1 in 2016, which is fair, given that he won it by less than a point.
This sample? Trump +6. It should be pretty evident why that matters... (5/6)
Poll after poll shows that Trump 2016 voters are sticking with Trump. So of course, a Trumpier sample will yield a better result for him.
Instead, the "improving economy" is given the credit for the shift, even though the data shows voters have not budged on the economy. (6/6)
Instead, the "improving economy" is given the credit for the shift, even though the data shows voters have not budged on the economy. (6/6)
