What’s The Matter With Polling?

Those of us who have been yammering on and on for months about the problems with polling are now finally reading the mea culpas of pundits who lazily swallowed the data they were given without first sniffing to see if it was fresh. I’m not sure that polling is completely broken, but certainly political polling is.

Part of the problem is that with the widespread use of caller ID people don’t answer calls from unknown numbers, but technology can help with that. The biggest problem was much more basic than that:  the demographic make-up of the pool of respondents. Even though there was ample evidence that (1) women, young people, and people of color were registering to vote in significant numbers and (2) the Harvard Youth Poll broadcast a strong message that young people were going to turn out in large numbers, pollsters doggedly stuck their standard demographic splits, data be damned.

This was almost entirely fueled by the refusal of the pollsters and poll aggregators like 538 and RCP to value women, young people, and people of color as much as white men. That’s it. That’s the truth. Racism and sexism are baked into US culture, and pollsters and poll aggregators included it in their products.

This overvaluing of white males gave bad actors data with which to mislead the public. Because of 538’s and TCP’s uncritical acceptance of any polling based solely on the pollster’s reputation and not on who sponsored the poll, they often blindly accepted that the demographic splits really represented a likely set of voters, one of my First Draft hobby horses over the past few months.

Here’s an example:

So this looks bad, right? But it was sponsored by Evergreen Principles PAC, which is a right wing group whose head is tied to Mitch McConnell. And it contains no details about what the sample looked like in terms of gender or age:
Survey Methodology: MIG conducted this survey November 3 through 5, 2022. A total of 500 live phone and online
interviews were conducted among likely 2022 general election voters statewide in Washington State. The potential
sampling error is plus or minus four percent at the 95 percent confidence level.
We have no idea who that poll sampled. That’s a red flag, and there were scores of polls like this released in the final 2 weeks of the campaign. Those fake polls led to data like this:

The media and its attendant pundit class ran with these polls and tainted polling averages. Why would they do that? Because of their lazy need to perpetuate the tired “DEMS IN DISARRAY” narrative.  And also because it was a deliberate act of bad faith. It was an act of voter suppression. And it wasn’t random, either.

The bad actors at work in the US over the last decade have one goal:  to make sure you don’t have faith in anything so they can use those things–political figures, the government, polling–as propaganda to push their narrative. And very few people are taking this seriously enough. Someone paid for all of those polls. And they didn’t do it to save a woman’s right to choose. It’s just more gaslighting and the media should be looking at it very closely.


One thought on “What’s The Matter With Polling?

  1. I agree, the media should take a more skeptical view towards polling reports, particularly from polling outfits that have no history. Granting every pollster an a priori assumption of good faith can no longer be supported. But in the same way that someone paid for all those polls, someone made the editorial decision to do a story on those polls without looking too closely into the sample size or the methodology. And I’d bet a shiny nickel that the person who decided to run the Smiley-Murray dead even poll story isn’t someone who is routinely questioned about his decisions.

Comments are closed.