Sunday, November 5, 2023

About that New York Times/Siena Poll

 by Christopher Cooper

Two days before the 2023 elections that will decide  the next Governors of Mississippi and Kentucky, control of the Virginia and New Jersey legislatures, and the leaders of municipal governments in North Carolina and other states, the political world was gnashing teeth over....a poll about the 2024 Presidential election.

The New York Times/Siena poll of six battleground states (excluding North Carolina, but I digress...) showed Donald Trump with a commanding lead over Joe Biden in Arizona, Georgia, Nevada, Michigan, and Pennsylvania, and within the margin of error in Wisconsin. It also showed Trump garnering support from almost a quarter of African American voters. After reading the results, it seemed as if every Democrat in America commenced to panicking and trashing the New York Times while everyone with a closet full of MAGA hats sang the praises of the New York Times and expressed faith in public opinion polls. Strange times we're living in.

My takeaway: this poll reinforces that either Joe Biden or Donald Trump could win the 2024 Presidential election. That's it. Nothing more, and nothing less. Trump is not suddenly a favorite in 5 battleground states and he's certainly not up by 11 percentage points in Nevada. Nor is Biden favored in those states. 

So, why am I pouring a large bucket of cold water on the poll aftermath?

First, this is one poll. No shade on the Times for putting it out without the context of other polls (their job is to sell papers, clicks, ads, and subscriptions, after all--not to sing the praises of other people's polls), but as we try to understand what this means, we should always pair every individual poll with other polls to get a more complete picture. Case in point: remember that poll from the other week that showed RFK Jr. garnering 22 percent of the vote against Trump and Biden? (Voiceover: RFK jr. is not getting 22 percent of the vote).

There are also some head-scratchers once you dig into the crosstabs that suggest that maybe everything's not as it seems. The eagle eye of Nevada politics, Jon Ralston noted on Twitter/X that it stretches credulity that Trump would be ahead by 6 points in Clark County, Nevada--a county where Democrats hold an 8 percentage point lead in registration over the Republicans.  

The Times article also makes a lot of hay out of subgroup differences. In the most dramatic example, they note that "Black voters--long a bulwark for Democrats and for Mr. Biden--are now registering 22 percent support in these states for Mr. Trump, a level unseen in presidential politics for a Republican in modern times."

It would indeed be "a level unseen in presidential politics for a Republican in modern times." Which is why I'm not buying it. Subgroup analyses like these are notoriously unreliable (smaller sample sizes, among other things) and should always come with "buyer beware" in flashing lights. As pollster and public opinion expert Natalie Jackson noted on Twitter/X, if the pollsters planned to conduct subgroup analyses on black voters, they should have oversampled black voters. They didn't and as a result, we should take this analysis with not just a grain, but a huge shaker of salt.

Then there's this: the Times article stated, "In a remarkable sign of a gradual racial realignment between the two parties, the more diverse the swing states, the farther Mr. Biden was behind, and he led only in the whitest of the six." Empirically, I have no doubt they're correct. But my hunch is that they miss the causal mechanism here. These data do not suggest a "gradual racial realignment," but rather that in the more diverse states, whites (not blacks) are doubling down on Trump. That explanation (unlike the explanation that blacks in diverse states are moving to Trump in unprecedented levels), would be consistent with scores of studies on racial threat theory over the last 70 years.

And, of course, the write-up of the poll, while compelling, could benefit from some additional context. Again, writing on the time-wasting web site formerly known as Twitter, journalist Stephen Fowler noted that a "young black Georgia voter surveyed by NYT who's disappointed in Joe Biden has never actually voted in a Georgia election in the 8 years he's been registered..." Fowler adds, "Could he vote Trump/are there other like him? Sure! But that context could help." I agree.

Perhaps most importantly, the election is a year out. Most people (other than those who read this august blog, that is) are thinking about other things. College football is rolling along. Curb Your Enthusiasm is coming back for another season. And ski season is about to begin. A lot of what this (or any poll) is picking up this far out is just noise. The few people whose opinions are malleable just aren't paying attention to the presidential election yet. And there's a lot that can happen between now and the election. Heck, we'll probably have at least three more Speakers of the House by then, to say nothing of indictments. 

None of this means that this poll is bad or pointless--it's neither. This poll represents a notable and important attempt to understand public opinion in six states a year before an election. Its findings reinforce that neither Biden nor Trump is a shoe-in to win the 2024 Presidential election. That's it.


Dr. Christopher Cooper is the Madison Distinguished Professor of Political Science and Public Affairs at Western Carolina University, where he also directs the Haire Institute on Public Policy. He posts on Twitter/X, Blue Sky, and Threads (and, ignoring his better angels, probably whatever the next one is) at chriscooperwcu.