4:22
The fundamentals are just overwhelmingly in Democrats' favor. The Economist's new model forecasting America's congressional elections gives the Democrats a whopping 98% chance of taking back the House of Representatives and an impressively high 48% chance of winning the Senate as well. The model eats every possible type of information that I can feed to it. So, that's historical election results, how every district or state voted in the past. Um it's uh all types of polling, national polling, presidential approval, polling of specific races. It looks at results of special elections to fill vacant legislative seats. It looks at fundraising.
You name it, we got it, we process it. One of the other things that has been, I don't know, the sort of thing that keeps you awake at night is about polling and polling errors. What kind of things are you looking at this time around? So, my role in this ecosystem is basically to calibrate that uncertainty in the polls and then uh combine it with other types of information and see how much we can reduce the uncertainty using that. There will be some polling error, the polls won't be perfect. Um something is going to be off by some amount somewhere. Probably everything is going to be off by some amount everywhere to some degree. What our model is supposed to do is calibrate all of that uncertainty and
say, "Okay, based on everything that we've seen in the history of elections and polling uh for Congress in the United States, um what are the chances that we get polling errors of X size in Y direction in this place plus in the uh Z size in the similar direction in another place, and how does that all add up?" Strong vibes of looking at the forecast and understanding what exactly it's telling me about whether it will rain or not. So, I won't be betting the farm on the basis of the models output at this point, but let's go through what it says for now, starting with the house. Uh talk me through it. Our model is very confident that the Democrats are going to flip the house.
Following the Virginia redistricting, and I should clarify here, one central and probably incorrect assumption that the model makes is the current maps are final. In fact, um uh the Virginia State Supreme Court could throw out the Democratic gerrymander. It's widely expected that Florida will uh attempt a final retaliatory gerrymander of its own. So, um this is all conditional on the current maps, but even a handful of further districting seat changes as a result of districting wouldn't vastly change this number or anything. The fundamentals are just overwhelmingly in Democrats' favor. First of all, it's a midterm.
[snorts] The whichever party is not does not hold the presidency tends to do extremely well in midterms. The polls are consistent with this. Uh Democrats lead by about six points on generic ballot polling, which measures um uh which asks people which party they plan to vote for in Congress. Um the president's approval rating is 20 points underwater. Um the maps are reasonably fair. Republicans can't count on gerrymandering to, you know, hold back the popular will. [snorts] So, it would I mean, look, pigs can fly. One in 100 of it One in 50 events happen one in 50 times, but uh I definitely think the Democrats are going to take the house. I couldn't quite tell you how big their majority is going to be, but like
I feel pretty confident in that. Okay. All right. Small side bet then for me, part of the farm on uh on the house. It's the Senate that's more of a nail-biter. Talk me through what the model says there. The Senate is really a nail-biter. It wasn't quote-unquote supposed to be because the map is super favorable to Republicans. Democrats need to flip four seats because Republicans hold the vice presidency, which break ties. And there's one of those that is not necessarily in the bag, but quite likely, which is that Roy Cooper, the former governor of North Carolina, will probably win a Senate race there. But the Democrats have a vulnerable open seat in Michigan that they need to hold, and there's quite a divisive primary happening that
happening there right now. And then they need to knock off Susan Collins in Maine, which is a blue state, but she's [snorts] the most moderate Republican in the Senate. And then after that, Democrats still need to flip two more states, all of which Trump won by at least double by double digits in the most recent presidential election. The reason why we still think that it's basically a coin flip that the Democrats could do it is a combination of two main factors. One, the national political environment is very favorable to not being the incoming party. And then in a few of these seats, Democrats got really strong recruits. In Ohio, Sherrod Brown, who is a long-time senator from that state. In
Alaska, Mary Peltola, who has already won a statewide race there. So, in theory, that would get the Democrats to 51. Now, both of those are tough races, but you know, if those fall through, they might have some backup plans elsewhere. Apart from the gerrymandering redistricting that you mentioned, what other curveballs might really change what the model's going to predict or indeed going to change the outcome? Well, the big known unknown is just who's going to run. Most primaries haven't happened. And our [snorts] model, it assumes that incumbents who are seeking re-nomination will get it. It doesn't have incorporate any information including polls about um
specific races until both nominees have been chosen. So, I would say so that and further redistricting are the two big known unknowns. And then the unknown unknowns are just change in the national political environment. Something big happens in the world or the country that is, you know, good for Trump [clears throat] and Republicans and bad for the Democrats and, you know, the Democratic lead on the generic ballot falls from, you know, six to four or two or zero and then we'd be in a very different universe. Got it. Well, I suspect we'll be checking back in with you as we get closer to the midterms, but for now, thanks a bunch. Thanks for having me.
