10/27/2024
Thanks David Siemers 🙃 our friendly neighborhood political scientist! A must read for those interested in the Trump/Harris election . Yup, I put Trumps name in just so people would read this. 😉
Political scientists call reporting on polls during elections "horserace coverage." The media likes to focus on horserace coverage for a number of reasons: 1. it's easy...generally others conduct the polls, and all you have to do is report the numbers; 2. they avoid charges of bias by reporting on something that seemingly requires no judgment from them--insulating themselves from criticism in reporting on a subject that is otherwise fraught; 3. it doesn't require much writing skill or knowledge about politics--or even knowledge about public opinion polling, for that matter; 4. the subject is interesting and consequential--people want to know who is going to win!!
1, 2, and 3 aren't great reasons to emphasize this kind of coverage over other coverage. 4 probably is, but in the current year, the answer just simply isn't there. And the amount of space that this kind of story gets crowds out discussion of issue positions, candidates' approaches/personalities, and transition/new administration plans, all of which are harder to write about and more difficult for the public to read, meaning that they don't "put eyes on screens," or whatever the equivalent to "sells newspapers" is these days.
As someone who has watched politics a long time, I find it strange to note that polling and horserace coverage has increased exponentially in the last 45 years at the same time that its usefulness has declined. This is not for a lack of making strides in polling methods. The good polls of today are conducted with methodologies that are far better than the best polls were even 15 or 20 years ago. But...1. so many people are conducting polls that there are a lot of junk polls along with good ones; 2. response rates have plummeted and are now well below 10%, making it harder to cobble together a representative sample; and 3. there aren't fixed, eternal truths in social science, as people are a moving phenomenon--what worked or was the case in the past may not work or be the case anymore as human beings respond in different ways to new conditions and new incentives.
This leads me to discuss one innovation in poll methods that has been introduced by the best pollsters in 2024--asking respondents how they voted in 2020. They are asking this because along with controlling for race and gender and party affiliation and socioeconomic status and religion and all the other things that might have some influence on a vote, they want to build a sample that accurately reflects the full population being polled. If Wisconsin was 50% Biden and 49% Trump last time around, a responsible pollster will build a sample of Wisconsinites that looks like that (noting that current 18-21 year olds can't be part of that analysis because they didn't vote in 2020).
The purpose of doing this is said to be to avoid undercounting Trump voters, a greater percentage of whom may have shied away from answering pollsters questions in 2016 and 2020, and who might have been undercounted by 1 or 2% in those years. This seems to have happened, but it is just a guess whether Trump voters are less likely to answer a pollsters questions in 2024 than prospective Harris voters. Trump fans have to hope that the shy Trump voter phenomenon is as big or bigger in 2024 than in the past two elections. Harris fans, meanwhile, must hope that Harris voters are now sufficiently turned off by presidential politics that they are as likely as Trump voters to avoid being polled.
One or the other of the two phenomena stated above is likely true. That being the case, it's quite likely that most of the battleground states will tip in the same direction. If this occurs, one of the two major party candidates will win 5, 6 or all 7 of the battleground states (PA, MI, WI, GA, NC, NV, AZ), and whoever does that wins the presidency.
May you live in interesting times!
