A reader's comment on polling: The errors of margins?

Yesterday, reader VoiceOfReason commented at some length on polling and the media. On rare, very rare, occasions a comment will get hung up in Typepad's spam filter, which I'm unaware of unless notified by the commenter. And when that happens, I then heroically rescue said comment captured by said filter, which I have on this occasion. Because VoiceOfReason's remarks on this post were of some length, I'm posting them here. —PM

***

It is suspicious at least that no matter what happens, the polls are almost always stuck within the margin of error. There seems to be little movement in spite of the debates, Trump’s convictions, his increasing dementia, his boring nonsense spewed at high volume in rallies where half the audience leaves, or anything else for that matter. One almost has to conclude that having a neck-and-neck race feeds the media’s craving for a horse race.

It also feeds into the strategies of the two parties. For the GOP, they want their voters (and the Democrats) to think they’re winning, so that if they lose they can scream about a stolen election (sound familiar?). For the Democrats, they want to be perceived as the underdogs when demographics and issue specific questions show that they’re not—so that our people will not grow complacent and take our foot off the gas.

If we start considering the constituents of the Trump vote, we have to subtract some pretty significant blocks of voters from Trump’s 2016 base. First, the Nicki Haley voters who said they will never vote for Trump. Have they all gone mad and returned to the Trumpian sewer? I doubt it. Second, that portion of Republicans who said they’d never vote for a convicted felon—the same question applies.

Third, the number of GOP women who are sufficiently pissed by Dobbs to vote for Harris, or at least not vote for Trump, especially in the five Trump states with Pro-Choice amendments on the ballot. Fourth, how many GOP voters died from COVID? Fifth, how many have died from other causes. And let’s not forget the Democratic ground game or Harris’ huge cash advantage.

No, I don’t believe the neck-and-neck thing. I suspect that the sample universes behind the polls are not being assembled to reflect what’s happening now, but rather rely on what happened in 2016. So I’m more optimistic than the polls.

 

By reader-commenter VoiceOfReason, 19 Sept. 2024

Comments

[VoiceOfReason]

Thanks, PM! I should probably have included the fact that the electorate is younger in 2024 than it was in 2016, or even 2020, and those younger voters seem to skew left. And I should also have mentioned the Swifties--how many of them are of voting age out of her more than 200 million followers? I'd imagine quite a solid block. And given Trump's recent outburst/confession/cri de coeur that he hates Taylor Swift, I doubt many of them will be leaning his way. Same with Randy Rainbow, who earned the distinguished Trumpian Badge of Hate overnight on his platform (whose stock, I see, hit another all-time low this morning).

[Anne J]

Watched a Politicon video on YouTube where WaPo's Jennifer Rubin was interviewing democratic pollster Simon Rosenberg. He said he believes that the race is not as close as the polls are saying. He also predicted in 2022 that the Dobbs decision will have a much larger impact on elections beyond the 2022 mid terms. Much more impact than the media thinks and much more than republicans want to think about.

And since that horrible decision, it has been documented that women have died due to these abortion bans. The state laws have only gotten worse. I believe a lot more women plus the men who care about them will be voting dem even with the gaping gender gap.

[VoiceOfReason]

Exactly, Anne. That's why I continue to think that even Ohio is in play. I remember how hard the GOP tried to keep the pro choice amendment off the ballot last year, and how badly they lost. I can't imagine that the women (and men) who voted for the amendment would suddenly turn around and vote for the GOP. Pro choice is on the ballot in ten states this November, and five of those are red states that went for Trump in 2016 and 2020. But the post-Dobbs history shows that when abortion is on the ballot, the GOP loses. As the Burnt Orange Menage would say, "Bigly!"

[PM]

Now you know how I feel, VoR, after posting something and hours later thinking, "Oh hell I forgot to mention that," whatever that was.

[VoiceOfReason]

Well, PM, the original comment was long as it was--long enough to earn a few silent "tl;dr" ratings from your audience.

[Tom Benjamin]

I agree with almost everything VoR said. I just posted something similar on Mr. Carpenter’s Nate Silver polling piece. I think the public polls are worthless.

I see a big Democratic victory on November 5th.

[Halster]

Polls have been irrelevant for several election cycles. The Pew Research Center says that phone poll response rates have dropped from 36% to 3% since 1997. These days pollsters tend to use survey panels, a group of people who have agreed to take surveys on a regular basis. Pew Research has their own panel of about 10,000 people called the American Trends Panel. If you don't have your own survey panel, there are plenty of third party companies that will rent you one for your specific needs. And now they try to convince us that online surveys are legitimate. Today's polls are based on junk science.