Former President Barack Obama, center, celebrates at a campaign stop with Wisconsin Democratic Gov. Tony Evers, left, and U.S. Senate candidate Mandela Barnes Saturday, Oct. 29, 2022, in Milwaukee.
( Morry Gash
Melissa Harris-Perry: Welcome back to The Takeaway. I'm Melissa Harris-Perry.
Since 1938, Americans have voted in midterm elections 21 times, and in all but two of those midterm elections, the party of the incumbent president has lost seats. 19 out of 21 times. This year will mark the 22nd midterm election since 1938. According to The Economist Congressional Forecast, this year is probably going to follow historic trends.
According to the paper's latest forecasting model, Republicans are likely to gain a majority in the House and the Senate. Well, that's a dead heat. Joining me now is Elliot Morris, data journalist and a US correspondent for The Economist. Elliot, thanks for coming on the Takeaway.
Elliot Morris: Thanks so much for having me on.
Melissa Harris-Perry: All right, so in real life, there's only one 2022 election cycle, but every day you all over at The Economist hold 10,000 elections. How's that possible?
Elliot Morris: Yes. Well, what we're doing is trying to assess the uncertainty of the race. The big question we want to answer is pretty similar to this question you started out with which is, if we are in an election where, say, Democrats are up two percentage points or maybe they're down two percentage points on the generic ballot, that's the question that asks how people are going to vote in their local congressional races and then adds it all up nationwide.
How often have they gone on to win control of the House majority? To do that, you need statistics, you need computational power, you need all of these 10,000 simulations, as you mentioned. It's really just a mathematical way for us to account for any surprises in the election. That's where these fake elections, 10,000 of them come into play.
Melissa Harris-Perry: When you've been running these simulations, overall, in this final weekend before the final votes are cast, what is it telling you about control of the House?
Elliot Morris: On the House side, we find that 78% of the time we think Republicans are going to win a majority, 218 seats are more in the chamber. That's down to that generic congressional ballot. The aggregate situation in the national political environment right now is not very good for Democrats, but also it's due to local considerations we find even though the generic ballot is looking pretty tied, actually Republicans up nearly a point. That would signal, I think, in most people's minds a close election but because Democrats need, we find to win the popular vote by about two percentage points in order to really have a chance at that majority, then they don't have a great shot at holding their current majority.
Melissa Harris-Parris: Earlier this week, I know The Economist updated the forecasting model to account for some of the recent significant outliers in recent polling. Help us understand what that means.
Elliot Morris: Yes. When you're aggregating polling, when you're averaging different readings together, one problem that you run into is biased and routine outlier polls. It's not so much a consideration if, say the ABC Washington Post poll, which is really high-quality releases one outlier survey. The algorithms are designed to get around that. When you have one polling firm continually overestimating support for one party, say by 10, 15, 20 percentage points, that becomes a problem for your average. It biases your average.
The problem we ran into this year is that there's a lot more biased pollsters than there used to be, at least as a share of all polls. That's partly because polling is really expensive and really hard now, so non-partisan pollsters are doing less of it. Also, I think there's something about our information environment that incentivizes partisan pollsters to release these surveys. We haven't run into this problem of routine 10 or 15-percentage point outliers before. We had to go in and tweak our model to be a little more sensitive in calibrating or calculating what we call pollster house effects.
That's just the empirical bias of a single polling firm. If they release polls that have Republicans winning by five percentage points more than all the other polls, then we say they're biased by five percentage points and we subtract that. That's how it works under the hood.
Melissa Harris-Perry: If these historical trends, we're talking about only two midterms since 1938 where the president's party did not lose seats. In those two, which were 1996 and 2002, the way political scientists tend to think about that is that it was a backlash for the impeachment of President Clinton in '96 and that there was a rally around the flag effect in post 9/11 that impacted 2002. Is it possible that the Dobbs decision is going to behave like those historical outliers and that your forecasting model simply can't pick up something that is such a rare event?
Elliot Morris: Well, for a while it looked like the Dobbs decision was going to be the reason that this midterm didn't look like historical trends. You've mentioned this 2 out of 21 midterm elections. I think going back to 1938, Republicans have also-- or sorry, the out party. This year, Republicans have also gained 28 seats in the house on average and four in the Senate. That's about where our projection is at today. That's not what it looked like two months ago. Two months ago, it looked like Democrats were going to gain a seat, maybe 2 in the Senate and they would lose 15 seats in the House.
Now, all that's fallen back to this baseline historical expectation. I think the question of whether or not polls are picking up any of this enthusiasm from Dobbs is a legitimate one. However, polls are pretty good at this. They're not-- again, maybe they're not laser alike in their predictive accuracy, but they do ask people, are you enthusiastic about voting? They don't necessarily rule people out entirely based off of their past voter history. Maybe that's the theory. I think they're pricing in a pretty realistic picture of the political environment today. I wouldn't be surprised if there's a small polling error but if the polls are systematically missing 5 or 10 percentage points of Democrats because of enthusiasm from the Dobbs decision or anything else really, that would be quite shocking to me at least historically in how we understand polls today.
Melissa Harris-Perry: Then, Elliot, if we're looking at not only a historical trend but probably of about the same magnitude, an estimated $16.7 billion [chuckles] has been spent on federal and state races this year. Was that all wasted money? Do campaigns matter if the aggregate outcome is predictable, really basically since 1938?
Elliot Morris: I have no idea how to answer this question. I'm thinking about the way to start, maybe to say the United States is really the only democracy in the world that has campaigns for six months or a year. We'll have presidential election campaigns starting maybe next March, a year before the primary, a year and a half before the general election. That is an aberration. There's no real reason for us to believe that changes things or maybe even that it's a good idea or an important way to run your elections. Maybe coming back to the voting behavior here, the psychology of it all, I think a lot of this stuff doesn't matter.
Voters are not looking at swings in gas prices every single day. They're looking at the fact that most of this year they've had to pay more for gas than they did last time. They're not looking at the prospects for congressional action on Roe v. Wade or abortion rights after 2020 and whether or not those prospects are changing every single day like our model. They just have a baseline predisposition that they probably have around July or August, and that might change a little bit, but usually comes back to a pretty reliably predictable picture.
Yes, I think we waste a lot of money on campaigns. We waste a lot of time and energy on the reporting of them. Hey, I'm saying that as a reporter, so maybe this is not in self-interest at all, but evidently, this is true. This happens every single election cycle, this seesaw effect for one party. It almost always, at least recently comes back to be a more competitive election than we thought.
Melissa: Elliot Morris is a data journalist and a US correspondent for The Economist. Elliot, thanks for your time.
New York Public Radio transcripts are created on a rush deadline, often by contractors. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of New York Public Radio’s programming is the audio record.