Aug.19

electoral-vote.com

Dem 51

   

GOP 49

Vance: The Polls Are Fake

Poll after poll shows Kamala Harris surging and Donald Trump wilting. Look at our electoral-vote graphs to see the change in the past month. Sen. J.D. Vance (R-OH) went on Fox yesterday to discuss the situation. His explanation: The polls are fake. Oh. How does that look back on Planet Reality? Here are the 10 most recent head-to-head national polls from 538:

Harris is ahead in nine of them and the average over all 10 polls is Harris +3.1. The conventional wisdom is that a Democrat has to lead by 3 points nationally to overcome the Republicans' built-in advantage in the electoral college, with small red states overweighted.

How did Vance explain all these polls? He said: "Consistently, what you've seen in 2016 and 2020 is that the media uses fake polls to drive down Republican turnout and to create dissension and conflict with Republican voters." So he is effectively saying that nine different pollsters are all making up numbers and only Fox's poll (Beacon + Shaw) is right? Massive conspiracy? Guess so, in Vanceland.

In any event, Vance has learned well from his mentor. When faced with difficult facts, just lie and make up your own facts. He also contradicted himself within a minute. He said that Harris' surge was a "sugar high." That sort of implicitly admits that Harris is surging then, no? Either the polls are fake and Harris is not surging or the polls are real and her surge is a sugar high. It's hard to have it both ways.

What else did Vance say? This: "If you talk to insiders in the Kamala Harris campaign, they are very worried about where they are because the American people just don't buy the idea that Kamala Harris, who has been vice president for 3½ years, is somehow going to tackle the inflation crisis in a way tomorrow that she hasn't in the past 1,300 days." Umm. It is well known that presidents have very little impact on the economy. We always wondered why. Now we have the answer: It's vice presidents who control the economy. Who knew?

When host Shannon Bream showed Vance a WaPo-ABC News/Ipsos poll with Harris ahead 6 points nationally, he said Harris' numbers were "stagnating" and the poll was "wildly inaccurate." Of course he said this. What else could he say and not be fired? Is it going to convince the swing voters, especially those who may watch Fox, and so heard the interview, but also get their news from other sources? We have our doubts.

And it could get worse. Candidates often get a polling bump from their conventions, so Harris could go higher. That could energize her base even more. Then there will be the debate on Sept. 10 and Trump's sentencing on Sept. 18, unless he gets that postponed. Just as Biden needed the debate to change the momentum, now Trump needs something to change it.

And as long as we are on the subject, every day we get questions like this one from reader R.D. in Austin, TX:

I've noticed this year what seems to be a lack of polling in several states since Harris got the top spot on the Democratic ticket. I've also noticed a couple of new polling outfits being mentioned in the recent news articles about polling trends, such as something called Outward Intelligence. Are you concerned 2024 is going to be a bad year for polling accuracy, given some of the new players in the market and the fact that everyone is only able to guess as to what the model electorate will look like?

We wish would could say, with confidence, that the polls will be OK this year. But, of course, we cannot say that. Nobody, not even Ann Selzer, will know for sure until after the election is over.

Many people, and many, many readers, are concerned that because the polls missed on Donald Trump in 2016 and 2020, that the polls could miss by a similar margin again this year. If so, then that could mean that the small-but-steady lead that Kamala Harris has opened up is a mirage, and the race is tied (or, maybe, Trump is actually still ahead). This is certainly possible.

That said, let us put forward a few things for your consideration. First, take a look at our final posting before the 2020 election. We had a bunch of guesses, from ourselves and others, as to how the EV tally would shake out. From high to low, the guesses for Biden's EV total were 351 (Z), 350 (The Economist AND our map), 348 (FiveThirtyEight), 321 (Sabato), 293 (V), 290 (Cook), 279 (PredictIt AND Politico) and 278 (Election-Projection). That averages out to 313.9 EVs, let's round it to 314. Biden's actual total was 306. So, the wisdom of the crowds (which blended polling data with a little gut feel) was not far off.

And now, let's take a look at our final map for 2016, which had 317 EVs for Hillary Clinton. She actually got 227, of course. Oops. We did not think, back then, to aggregate other outlets' projections, or to make our own, independent-of-the-map, guesses. So, we can't do what we just did for 2020. That said, we did have an item that day on national preference polls, all of which had it close. We also had warnings that week, from us, and from others, that Trump might just pull out the victory. And the "miss" on the projections was not entirely because the polls were inaccurate. It was also because polling of the "blue wall" states had virtually ceased, which meant a late, post-James Comey break toward Donald Trump largely went undetected.

The point here is that polling was certainly imperfect in the last two presidential cycles, but the extent of the problem should not be overstated. On top of that, the pollsters are well aware of their misses, and are doing everything they can to correct for them. We do not know exactly what they are doing, since they keep their "secret sauce," well, secret. However, you can be certain that, to take one example, the "blue wall" states are going to be polled right up to the bitter end, and that error will be avoided. Other errors will be, too, as best as possible.

There has already been some evidence that pollsters have corrected, or even over-corrected, this cycle. A number of pollsters, most obviously Siena and Emerson, have had unusually GOP-friendly numbers this year as compared to past years. There have been cross-tabs that were a bit hard to swallow, in particular showing Trump with historic levels of support among Black voters. Why would they break in his direction in his third election?

R.D. also raises a good point about this year's electorate being hard to model. On the whole, pollsters have to work with what the electorate actually looked like in past elections, and then maybe tweak from there. And the tweaks are going to be on the conservative side (as in, "not going too far," not "in favor of Republicans"). Think about what groups might show up in unusually high numbers this year. Women, right? Young voters? Black voters? Asian/Indian voters? All of those groups can be expected to break Democratic. And so, if pollsters make errors in their models of the electorate, there's a fair chance they will err to the detriment of Kamala Harris' numbers by not fully accounting for unusual turnout among those groups.

We are not saying the pollsters will get Trump right this year—again, unknowable. What we are saying is that there are enough moving parts here that any of three outcomes are entirely plausible: (1) the polls understate Trump's support, (2) the polls basically get it right and (3) the polls understate Harris' support. And, if we had to guess, we'd say each of these three possibilities is about equally likely.

Oh, and as to new pollsters? There are new pollsters every cycle. We take a pretty close look at any new polling house, and we don't let them into our sample unless we have good reason to believe they know what they are doing, and that they are not a partisan house.

It is well known that the election administrator's prayer is: Lord, let it be a landslide. Pollsters may be adopting it this year, too. (V & Z)