Read on Twitter
New @monkeycageblog piece this morning on problems with the Exit Polls.
Some are in a rush to explain what happened, but readily apparent issues with the Exit Polls should make folks skeptical about using them to do that storytelling. [thread] https://www.washingtonpost.com/politics/2020/11/10/dont-trust-exit-polls-this-explains-why/
Some are in a rush to explain what happened, but readily apparent issues with the Exit Polls should make folks skeptical about using them to do that storytelling. [thread] https://www.washingtonpost.com/politics/2020/11/10/dont-trust-exit-polls-this-explains-why/
Short version:
-Historically, the exit polls have had race/edu composition issues.
-It happened again. The Exit Polls are under-reporting the share of the electorate that is white non-college.
-This has effects that can ripple throughout the survey, influencing the vote margins.
-Historically, the exit polls have had race/edu composition issues.
-It happened again. The Exit Polls are under-reporting the share of the electorate that is white non-college.
-This has effects that can ripple throughout the survey, influencing the vote margins.
In '16, both States of Change (modeling ACS+CPS) and Pew's voter validation study said that 74% of voters were white, including 44% white noncollege. Assuming relative turnout stays the same and you get 72% and 41% in '20. Will likely be a little different, but a good baseline.
At present, the Exit polls are reporting that 67% of voters were white in '20, including 35% white noncollege.
That might not seem like a big gap, but it really is. It suggests a radically different set of implied turnout rates operating under the hood.
That might not seem like a big gap, but it really is. It suggests a radically different set of implied turnout rates operating under the hood.
Given reasonably well estimated eligible voter #'s and their compositions along with projected total turnout and the stated composition of those voters, you can work your way backward to derive the implied turnout of different groups.
The Exit Poll's current composition implies turnout rates that are radically at odds with other research. Specifically, it implies that white turnout was barely higher than Latino turnout and lower than Asian/other turnout.
White: 66%
Black: 67%
Latino: 63%
Asian and others: 74%
White: 66%
Black: 67%
Latino: 63%
Asian and others: 74%
This is not something that is supported by other research. For example, the CPS has consistently shown white turnout to be well above the turnout rates of Latinos, Asians, and those belonging to other racial and ethnic groups.
If anything, research by Ansolabehere, @blfraga, @b_schaffner suggests that the CPS is *over-reporting* the turnout of non-white populations. It makes these implied turnout rates - and the compositions they're derived from - all the more unbelievable.
https://www.bernardfraga.com/uploads/2/2/3/4/22341374/cps_a_f_s_042920.pdf
https://www.bernardfraga.com/uploads/2/2/3/4/22341374/cps_a_f_s_042920.pdf
Again, all this matters because it can affect the vote margins. While everyone says "wait for the final weights" these compositional issues make it possible that the final weighting could make the vote margins *worse*.
If you're under-reporting white noncollege, you're missing out on a chunk of the population that is Republican-leaning. The final weighting might compensate for the compositional issue by (all else equal) pushing the vote margins of more Democratic groups rightward. Not ideal.
Just to be clear: I'm agnostic about the substance of arguments people are making with the Exits. Maybe X group shifted Y points for Z reason.
What I'm saying is, let's form narratives about what happened using data that has less obvious issues.
What I'm saying is, let's form narratives about what happened using data that has less obvious issues.
