“When wise people explain Biden’s bad poll numbers, they cite objective facts in the real world: inflation, crime, etc. Yet Obama had almost an opposite set of facts (no inflation in 2011!) — and got almost exactly the same poll numbers.
“My hypothesis: Obama/Biden poll numbers in 2011 and 2023 tell us less about the world – more about the inherent problems of managing the Democratic coalition, which is always more fractious and fissile than the (smaller) Republican coalition.”
What do we make of the new polling that came out over the weekend? It sounds bad for Biden – and Democrats by extension (although the polls were focused on the presidential election). Is it really that bad for the president? I am sure this is your favorite answer to these kinds of questions, but it is yes and no.
Here’s a snapshot of the polling that dropped this morning. Siena College (New York Times), which is an A+ rated pollster, surveyed six battleground states and found Trump beating Biden in all but one. Even worse were findings that young people and Black and Hispanic voters were supporting Trump in large enough numbers to cost Biden the election (assuming these states decide the overall outcome, which is a fair, but by no means certain, supposition). YouGov (CBS News) (B+) also released a poll, which was focused on national trends and used a nationwide survey. Trump is beating Biden here too, but it’s closer and we don’t see the same trends among young folks that we see in the Siena poll. The racial trends are still problematic.
If you want to read more about the polls themselves, Nate Cohn has a good analysis of Siena and CBS News has an article on the YouGov poll here.
Should we be worried about the findings in these polls? Sure. I have written previously about the data we have that young voters and Black and Hispanic voters are softening for Democrats. It’s real and it may not be about them becoming more conservative as it is about them feeling like Democrats are no putting progressive policies in place. But also, no, we don’t need to be worried. Here’s some reasons why we should not be too worried about the polling – and this polling specifically – on November 6, 2023.
No poll has any predictive value one year out from an election. The result could turn out to be the same in the end, but all any poll ever tells us is that this is what it might look like if the election were held today.
For instance, in November 2011, Nate Silver looked at how Obama was polling and concluded he probably would not win reelection. He won convincingly a year later. It was never clear whether respondents were down on the president or whether it was about fractious Democratic coalition which came home to Obama when it was time to (Frum thinks the latter).
The Siena poll is of registered voters, not likely voters. That’s not that big of a deal, in my opinion, especially this far out. It’s a much bigger deal when the election is close. There are different methods for determining a “likely voter” and Siena tries one option: asking the respondents how likely they are to vote. In this case, 89% said they were very likely or almost certain to vote. However, this question was not used to screen respondents and the results are for the entire sample of registered voters. (To be clear: there was not one poll; there were six polls – the numbers I use here are for all six polls combined.)
The YouGov poll is of adults. We don’t know how many are registered voters, but they did ask the likely voter questions. In this case, 90% said they were likely voters – even more than in the Siena poll (albeit not much more). Here’s the problem with asking all adults how likely they are to vote: a lot say yes even when they are not registered, and probably will not register, when the election arrives.
Siena used live telephone interviews as its data collection method. Sure, they combine landlines and mobile phones but that is something that fixed a problem ten years ago. Younger voters increasingly did not have landlines ten years ago, so pollsters were missing a lot of people. Today, the issue is less what type of phone one has but whether they even use it as a traditional phone. Younger voters more and more tend to not answer the phone when they do not know who is calling. And it’s not just youngsters; I certainly don’t answer the phone when an unknown contact calls and I know a few older folks who do the same. A better method today is using things like online panels (as YouGov did) and text-to-web inquiries. Siena may have missed a big segment of young voters, and they can’t see it in the data, so it looks like a good sample.
There is a more important reason to not worry about the polling data: Since 2020, off-year and special elections have seen a huge shift in support to Democratic candidates and issues (ballot measures). Tomorrow, there are elections in New Jersey, Ohio, Mississippi, and Virginia that are of interest to us.
Republicans are trying to take over the Virginia legislature and abortion rights are a big part of the campaign (they are trying to downplay it, of course). Ohio has a constitutional amendment on the ballot that codifies a right to an abortion. The New Jersey races might be a test of how much damage the corruption of Sen. Bob Menendez (D) has done to Democratic chances next year. And while we expect that Mississippi Governor Tate Reeves (R) will be reelected, the race may be very close – and a Democratic victory is not out of the question.
While GOP victories in New Jersey could make for a problematic story, the other states are the ones to watch. If Democrats retain the Senate and win the House in Virginia, if the Ohio abortion measure passes, and if Democrat Brandon Pressley runs a close race in Mississippi, that will tell us a lot more about the electorate than these polls will.
Election results mean more than polling. Let’s see what happens tomorrow before we get too worried.