Thomas Edsall from the NYT writes that exit polls on general election day aren’t quite studied enough by social scientists:
“We are all circling around the absolute truth, but we are all using methodologies that have their problems,” he said. Asked if he thought Pew raised legitimate questions about the exit polling, especially on the shares of college and noncollege voters, Lenski said that all surveys “have their errors,” adding that there “are sampling errors, nonresponse errors.”
Brian Schaffner, a political scientist at the University of Massachusetts, Amherst, specializes in political polling. The major problem with exit polls, he wrote in an email, is that:
it is very difficult to weight an exit poll to the target population, especially on the night of the election. The reason is that we don’t know yet what the electorate’s demographics will be when the voting is happening. The best that exit pollsters can do is record the approximate age, race, and gender of people who refuse to take the survey (based on what the exit pollster thinks they look like), but this only allows them to weight on three variables, and these factors aren’t even measured very precisely.
The problems of the Edison exit polls are widely recognized in the political science community, but rarely discussed in the media. Robert Y. Shapiro, a political scientist at Columbia, noted that the biases in the exit polls before they are weighted for the actual vote are well known — they have overrepresented Democratic voters and also now also the better educated who are more likely to vote Democratic.
Christopher Achen, a political scientist at Princeton, wrote me:
The exit polls have traditionally overrepresented college grads, and even more dramatically postgrads, probably because less-educated people are less inclined to respond to the interviewers unless they are particularly enthusiastic about a candidate and want to say so.
He added that “when I read the Pew report, I thought it was very likely right.”
Ted Brader, a political scientist at the University of Michigan, also emailed:
the issue with the education distribution is that exit polls could provide, at best, a representative sample of those who turn out to vote, not a representative sample of the voting eligible US population.
This has to do with a recent survey by Pew Research, released last month.
This week, I have been examining the ANES (American National Elections Survey) data a little more this week, but a different segment regarding women voters. I think some questions were definitely overlooked here, especially when it came to working class women and their perceptions of Clinton. Here’s what I found about women voters who were surveyed post-election.
One section on Candidate Image items:
D02 Clinton : really cares
How well does the phrase ‘she really cares about people like you’ describe Clinton?
1 Extremely well 397
2 Very well 595
3 Moderately well 730
4 Slightly well 581
5 Not well at all 1329
Well, most people thought she cared, but not far from half didn’t think so.
There is a section on Women and Gender Issues From the ANES Codebook that also had results. I found these questions to be interesting:
L07 How Trump treats women
In general, how well do you think Donald Trump treats women?
1 Poorly 1950
2 Neither poorly nor well 906
3 Well 767
9 NA 27
L08 How Clinton treats women
In general, how well do you think Hillary Clinton treats women?
1 Poorly 438
2 Neither poorly nor well 1034
3 Well 2147
The highest block of those surveyed, btw, were from the South, not the Midwest, although the Midwest had a strong showing.
The NYT op-ed points out that Doug Jones went after the working class vote, which explains his win in the special election in Alabama.
I think about those catty remarks the baroness made in India a couple of weeks ago. I maintain that the baroness really needs to get to stage 7 of her grief. Acceptance. It will help the party move forward.
Bernie is way ahead of the party, the revolution is happening, and the kids from March for Our Lives are proof of the pudding.