Tag Archives: election polls

There Are Polls…And The Polls

I get just as worried as anyone else over those polls that our media outlets obsess over–the ones that find more Americans voting for Mr. Mental Case than for President Biden. But I know I shouldn’t.

Over the past several years, in election cycle after election cycle, even opinion polls produced by highly credible sources have proved to be wrong, and not by just a little. There are a lot of reasons for the misses–ever since cell phones replaced land lines, efforts by those conducting the surveys to compensate, to adjust in order to ensure they are surveying representative groups of voters, have proven inadequate.

A recent post to Daily Kos reminded me why we should take all polls–with the exception of actual voting at the election polls–with a heaping measure of salt.

The post began by reminding readers of Congress’ abysmal polling–and the fact that, despite overwhelming disapproval, incumbent representatives get re-elected more than 90% of the time. (We used to refer to this as the “I hate Congress but love my own Congressperson” problem.)

It is difficult to square these high reelection rates with Americans’ overall low opinion of Congress. Gallup tracking polls of Americans’ opinions of Congress over the past twenty years reveal that rarely do more than 25 percent of Americans approve of Congress and frequently their approval is down in the 14-20 percent range. In one recent Gallup survey, as few as 9 percent of people approved of the way Congress was handling its job.

The post– written by someone who is himself a political pollster–went on to explain:

There is a lot of bad data out there, mostly because of the fact that it is almost impossible to collect it.  So when I tell you there is a lot of bad polling, almost all of it, it does not mean I think the polling is intentionally bad, or done with a purpose of propaganda.  It just means that Republicans are much, much easier to reach than Democrats.  I have spoken about this before.  It is one of the biggest reasons I saw the 2022 election going differently than most.

What the pollsters did not figure out, however, is that this is not a problem that can be fixed with weighting.  As I have said, a young person that talks on the phone, is more conservative than one that likes to text.  Generally, this cuts across all demographics.  Republicans are more likely to participate…

As Michael Podhorzer has demonstrated, pollsters influence outcomes by letting their own biases and intuitions tilt poll results by deciding who to include in the sample. A month before its late-October 2022 poll showing a four-point deficit, the Times/Siena showed Dems up by two points. In the October story, the Times/Siena showed Biden losing ground among independents and women. But as Podhorzer writes, “What the paper didn’t disclose was this: Independent voters hadn’t changed their minds; the New York Times changed its mind about which Independents would vote.”

Perhaps the most significant insight in this particular post is this: support doesn’t change. Turnout does. Political preferences tend to be stable; in today’s highly polarized political environment, self-identified Republicans and Democrats are highly unlikely to go to the polls and cast a vote for a candidate of the other party.

The motivation to go to the polls–the motivation to cast that vote–is another matter.

The number one thing to remember about polling, is that after about 2000, for the most part, bases solidified and voters do not often change their minds.  Partisanship has hardened.  All elections are about turnout.  While independents may be in a sour mood, they are also displaying nihilistic tendencies, and my research suggests the most likely scenario is one that makes 2024 a base election.

Permit me to repeat the most important insight in that paragraph: All elections are about turnout. 

Democrats did extremely well this November, because they turned out. Thanks to the Dobbs decision and the unremitting MAGA attacks on democracy and the rule of law, Democrats have been motivated to turn out in much higher numbers than usual. That motivation is likely to persist into 2024.

If there is a moral to this story, it’s this: don’t try to “get through” to your bigoted uncle at Thanksgiving. Don’t try to explain to your co-worker why his vote for the GOP is really a vote for cutting Social Security. Instead, make sure your college-age children, your live-and-let-live bowling buddy, and your pro-choice aunt are registered and that they have rides to the polls.

You know–those polls that (contrary to the Big Lie) are actually accurate and meaningful.