Happy Election Day! Here are my House predictions. I won’t go through every district. If you’d like to see how I chose individual races there’s a link to my map here. I predict Republicans will win the House 225-210. Yes, that is on the low end. Here’s how I got there.
First, as I pointed out in my Senate predictions, the election handicappers are being especially risk-averse this year. There is a lot more to lose from overestimating Democrats than overestimating Republicans. The data just doesn’t show a “red wave.” It never has. The FiveThirtyEight generic ballot average is at R+1 (R+1.7 with a likely voter adjustment). But the polling averages are made up of a disproportionate amount of partisan pollsters this year. Split Ticket has a generic ballot tracker that uses only non-partisan polls. That tracker is dead even. Then there’s the special elections from the summer, in which Democrats overperformed by an average of 9 points. That’s hard data. It’s not a poll. Not vibes. It’s how people actually voted.
So am I saying Democrats are going to win the popular vote by 9 points? No. Of course not. In fact, I think Republicans will likely win the House popular vote by 1-3 points. But a 1-3 point popular vote lead does not suggest Republicans will win 235 seats or more. For example, Democrats won the House popular vote by 3 points in 2020 and they only won 222 seats. The FiveThirtyEight forecast predicts Republicans winning 230 seats with a 4 point popular vote margin. In fact, alloftheforecasts, based on objective data, not vibes, show Republicans winning between 223-232 seats.
Anyway, that’s where I am. (225-210) There’s a lot of uncertainty around this election because there have been so few House polls but I think the forecasts are more likely to be right than the “vibes.” Frankly I think my prediction is probably on the low end and Republicans are probably more likely to win around 230 seats but it’s late and I’m tired and I couldn’t find the 5 seats to flip so this is what I’ve got. Enjoy!
Here it is folks! My final Senate prediction. I’m a bit more bullish on Democrats’ chances than most forecasters. Why? Maybe it’s because I’m biased. Maybe. Or maybe it’s because I think the professional forecasters are cowards and picking Republicans, not based on the data but because they have a lot more to lose from underestimating Republicans than underestimating Democrats.
What does the data say? Well it depends on the data you look at. High quality polls show Democrats still ahead in Pennsylvania and Georgia and within the margin of error in Nevada (a state where polls have actually underestimated Democrats in recent cycles). Low quality polls show Republicans with comfortable leads in all of these states and within the margin of error in Arizona. I don’t buy the Democratic conspiracy that “Republican pollsters are flooding the zone” with polls to skew the averages. I think the reason Republican pollsters are dominating the averages is because election handicappers, once again, are being cowards. There are simply far fewer polls being released by traditional, reputable pollsters. Put simply, they’re scared to be wrong.
So what to I think is going to happen? I think Democrats hold on in Pennsylvania and barely squeak out a win in Nevada. I, like other forecasters, am putting my trust in Nevada elections guru Jon Ralston who has predicted a Cortez-Masto win. Georgia is also going to be close and almost certainly headed to a runoff but I think Warnock will be able to hold on. The only pollster rated B+ or higher by FiveThirtyEight showing Herschel Walker with a lead is Trafalgar group, and they are aligned with the Republican party. I’m putting my trust in the higher quality polls. As far as Arizona is concerned, Mark Kelly seems pretty popular and is still leading in the polling. If Kelly loses in Arizona Democrats have almost certainly already lost Georgia and Nevada and probably Pennsylvania. Since we’re not in that universe here, I’m picking Kelly.
So in conclusion I do think Democrats are going to hold onto the Senate. My final map is 51D-49R. Could I be wrong? Absolutely. I wouldn’t be surprised by a variety of outcomes. Frankly, I think GA, NV, and PA are as close to even as possible and could go either way. I could also see Republicans winning in AZ and it’s not inconceivable that they could win New Hampshire as well. Conversely I think Democrats do have a shot at winning both Wisconsin and North Carolina, and hell, though I don’t think it’s particularly likely, Tim Ryan could still pull out a win in Ohio.
So there’s a wide range of possibilities. I wouldn’t be surprised to see anything from 54R-46D to 52D-48R. I would be surprised to see Democrats win more than 52 seats but it’s not outside the range of possibility.
A final thought: a state I’ll be watching is North Carolina. Not because I think Cheri Beasley is going to win but because I’ll be looking at the margin with 2024 in mind. Beasley and Budd are as close to neutral candidates as you can get. NC is an R+5 state. If the Republican margin is less than 5 points to the right of the national environment, that would suggest that the shift to the left we saw in 2020 could be continuing and that it could be in play in 2024.
A few weeks ago the New York Times’ Nate Cohn wrote a piece where he looked at the 2022 polling averages – where Democrats have been doing surprisingly well – through the lens of 2020’s polling error. His conclusion? As he writes, the “warning signs are flashing again.” According to the article, “Democratic Senate candidates are outrunning expectations in the same places where the polls overestimated Mr. Biden in 2020 and Mrs. Clinton in 2016.” If we applied 2020’s polling error to the 2022 Senate polls, Democrats would lose in places like North Carolina, Wisconsin, and Ohio, by fairly comfotable margins, despite leading in the polls.
Since the time of that writing Republicans have closed the polling gap, essentially running even now in those three states, but I’d still like to push back a little bit on Mr. Cohn’s article. There’s reason to believe 2022 is going to be closer to 2018 than 2020 or 2016. The 2020 election took place during the height of the pandemic and though we’re still technically in a pandemic (sorry Biden) November’s election is going to take place under much more normal circumstances. Plus, there’s reason to believe that Trump being on the ballot was a significant source of the errors in 2016 and 2020 – as his supporters may be much harder for pollsters to reach due to low social trust, and he seems to bring out infrequent voters, which messes with pollsters’ likely voter models. The polls in the Trump era when he hasn’t been on the ballot – 2018 and 2021 in the Georgia runoffs and later in Virginia – have been fairly accurate. The polls so far in 2022, especially in special elections after the Supreme Court overturned Roe V Wade, have actually been underestimating Democrats.
As such I decided to explore what the race would look like through the lens of 2018’s polling error and 2021 for Georgia (which didn’t have a Senate race in 2018). Unfortunately North Carolina hasn’t had a statewide race in the Trump era without him on the ballot so they’ve been left out of this analysis. The story doesn’t change in Ohio and Florida – where polls still overestimated Democrats in 2018– but it does change the view in places like Nevada, Pennsylvania, and Wisconsion. In Nevada and Pennsylvania polls actually underestimated Democrats in 2018, and in Wisconsin the polls were pretty darn accurate. Governor’s races are essentially the same story.
If we applied 2018s polling error Democrats would still win by comfortable margins in Michigan, Pennsylvania, and Arizona (though the AZ governor’s race would be close), and they would actually win by more in Nevada than polls currently show. The only state that would flip Republicans’ way would be Wisconsin, though both races would still be close.
We won’t know which direction polls are going to miss – if there’s a systematic bias at all – until after the election, but hopefully this chart adds one more data point to the conversation about what things may look like come November.
Note: I published this piece a week or so later than I inteded to so my chart isn’t directly comparable to the chart in Nate Cohn’s piece. Here’s a thread I wrote on September 12th, the day the Nate Cohn piece was written.
UPDATE (9/9/22): FiveThirtyEight updated their partisan lean metrics hours after I published this. The story has been updated to reflect the new data.
Special election charts and results as well as the above generic ballot polling average are by FiveThirtyEight. I wanted to credit them in the captions but WordPress is being dumb.
I think people are underestimating how likely it is that we’re in at least as favorable an environment for Democrats in 2022 as we were in 2020. Yes, Democrats are only ahead on the generic ballot by 1.2% at the time of this writing, whereas in 2020 they were ahead by 7.3%. But the polling in 2020 was notoriously bad. Democrats only wound up winning the House popular vote by 3.1%. That’s a 4.2% miss! In 2018 – when Trump was not on the ballot – the generic ballot got things exactly right. It predicted Democrats winning by 8.6% and they won by 8.6%.
The ‘Trump effs up the polls’ theory is going to be tested again this year (it makes sense since he destroys everything else that I love!) but in the meantime, let’s assume the generic ballot is correct, or at least close to correct (lets say for the sake of argument ±2%, which is in between the 4% error 2020 and the 0% error in 2018). Then the environment is anywhere from R+1 to D+3. So either slightly worse than 2020 or about the same.
But we also have other figures to go by, namely special elections. The advantage of special elections is that they aren’t polls. They’re actually elections where people go out and make a choice. The disadvantage is that they are usually low turnout affairs that attract only the most politically engaged. Still, they usually are pretty predictive. In the special elections since Dobbs, Democrats have been overperforming by between 7 and 11 points (depending on how you count Alaska).
So based on the generic ballot and special elections we can say that we’re in anywhere from an R+1 environment to a D+11 environment. To be clear, I don’t think we’re in a D+11 environment, so let’s use D+7 (the conservative calculation of special elections) as our high. So the environment is anywhere from R+1 to D+7, for a median of D+3. That puts us in basically the exact same environment we were in in 2020, when Democrats won 222 seats.
So Democrats should expect to win 222 seats? Not so fast. You forgot about redistricting! There’s a number of different ways to calculate the effect of redistricting but I think the easiest way is to look at how many districts Biden won. On the old map Biden carried 224 districts. On the new map he would have carried 225, according to Redistricter.
But Democrats ran 1.5% (or 2 seats) behind Biden; one of the reasons their House majority is so thin. So instead we can look at each district’s partisan lean or (PVI) using data from FiveThirtyEight. In a D+3 year Democrats should expect win every district with a PVI <R+3. This would give us a final result of 220D-215R. In an R+1 year it would be 206D-229R. A D+7 year would give Dems 237 seats, one more than they won in 2018, which makes me think we’re not in a D+7 environment (though stranger things have happened). So going by PVI we’re looking at Democrats winning anywhere from 206-237 seats, with the median being 221.5.
Another way of looking at it is to use FairVote’s Monopoly Politics Projections, which allow you to estimate each party’s share of seats based on different environments. I like this better because it uses toss-ups (so I have less likelihood of being wrong). In a D+3 environment there are 218 safe/lean D seats and 181 safe/lean R seats, with 36 toss-ups. 218 is the exact number you need to win the House. An R+1 environment gives us 209R-183D with 43 toss-ups. They don’t allow you to calculate D+7 but back of the napkin calulation tells me there’d be about 232 lean/safe D seats.
So going by FairVote, Democrats could expect to win anywhere from 183 seats (an R+1 environment where Republicans win all the toss-ups) to 254 (a D+3 environment where Democrats win all the toss-ups) with the median being 218.5. In a D+7 environment, Democrats would win close to 280 seats if they won all the toss-ups, but nobody’s had that kind of majority since the 1970s so I’m going to assume that’s not going to happen. Let’s stick to 254 (D+3 w/ all toss-ups) as our high.
So based on my read of the current environment Democrats are favored to win around 221.5 seats using PVI or 218.5 seats using FairVote’s projections. I guess this was all a long winded way of me saying that things are really freaking close! Also, someone’s going to have to be chopped in half. I suggest Marah Palintola.
Democrats are looking to defy history. Can they? Here’s a thread I wrote, analyzing how the current race compares to past races. Enjoy!
In good Republican years they tend to run away with the race around labor day as pollsters switch over to likely voter (LV) models. For instance, at this point in 2014, Republicans were ahead by 0.3%, but less than a week later they were leading by close to 4%
Democrats are currently leading on the generic ballot by 1 point, and Democrats and Republicans are essentially tied in LV polls taken in the month of September.
This is good news if you’re Democrats because it differs from 2010 and 2014 when Republicans were leading by 7.8 pts and 3.4 pts respectively in LV polls taken in the first half of September
Generally the party out of power holds an enthusiasm advantage, however this year, with the overturning of Roe v Wade, Democrats have closed the enthusiams gap, especially when compared with past midterms.
We’ve seen it in special elections where Democrats have overperformed by an average of 7-11 pts since the Dobbs decision (depending on how you count Alaska).
Whether this can hold for the next 2 months remains to be seen, but in past midterm cycles the party out of power has tended to run away with the race after labor day. Needless to say, the next few weeks should tell us a lot about where this race is headed.
In 2010 Scott Brown stunned the political world by winning a Special Election for the Massachussetts Senate seat vacated by Ted Kennedy after he died. Brown won 52-47, a 5 point margin in a state Obama had won two years earlier by 26 points. It was a stunning upset and a harbringer of things to come as Republicans picked up 63 seats in the 2010 midterms to easily retake the House in what then President Barack Obama called a ‘shellacking.’
Fast forward to 2022 and we have a very different situation. Democrats have overperformed in special elections by an average of 5.7pts since the Dobbs desicion overturning Roe V Wade. Democrats have overtaken Republicans on the generic ballot (at this point in 2010 it was R+4.5) and the Senate seems to be moving farther and farther away from Republicans – with Fetterman leading by 10pts in Pennsylvania and Mark Kelly leading by 8 in Arizona. Even Wisconsin and Ohio look within reach for Democrats (though we’ve been burned by polling in these states before).
The president’s party almost always loses ground in the midterms. In the 19 midterms that have taken place since World War II, the president’s party has lost seats in the House in all but two: 1998 (after the Clinton impeachment) and 2002 (after 9/11). Needless to say if Democrats don’t lose the House it would be a shocking turnaround from where we were just a few months ago. It would be less shocking if Democrats don’t lose seats in the Senate, but that’s still only happened 4 times since WWII.
So why are Democrats overperforming in a year where – if history is any guide – they should be losing ground? Well I’m sure the Dobbs decision overturning Roe v Wade certainly has something to do with it. As the Washington Post reported Monday 1 in 3 women have lost access to abortion, and more restrictions are coming. I think some of Democrats’ recent legislative victories have also helped though people usually don’t vote based on policy. We’re also seeing an improving economy with inflation and gas prices falling after (hopefully) peaking in June.
But I think the biggest reason is that Republicans have simply become too extreme. They’ve leaned into the Trumpist wing of the party, at the expense of regular voters. Republicans had a choice to make after January 6th. Trump or democracy. They chose Trump. That was a mistake and now they’re reaping what they’ve sewn. His voters simply don’t come out when he’s not on the ballot (just ask Republicans running in 2018) and it turns most Americans, at least the ones who vote in midterms, actually like democracy. They’re not okay with Republicans abdicating their responsibility to protect it. They like having rights and are not okay with Republicans taking those rights away, like the right to an abortion. And they’ll take boring competence that they don’t have to think about every day over the constant chaos, corruption, and conspiracies that come with Trump and the Republican party.
I never thought I’d see the day when one of our two main political parties simply stopped believing in democracy. When your response to someone trying to overturn the results of an election and a literal violent attack on our democracy is to look the other way or worse; actually defend these actions, you have turned your back on democracy and our not worthy of even being elected Dog Catcher (no offense to Dog Catchers). A democracy where voters don’t punish politicians for that kind of behavior is a country dangerously close to no longer being a democracy. I’m glad to see that maybe, just maybe, Republicans are in for the type of ‘shellacking’ they deserve.
Now that it’s officially 2022, it’s time for the midterms to kick into full gear! Republicans are bullish on their chances and, judging by the wave of Democratic retirements, so are Democrats. We all know by now that one of the golden rules of American politics is that the President’s party loses seats in their first midterm, and with Democrats already holding extremely slim majorities, their hold on power is hanging by a thread. But Democrats have a couple of things going for them that could help them buck history:
First, the Senate map actually looks good for Democrats. Republicans are defending 20 seats and Democrats 14. Moreover, Republicans are defending two seats in states that Biden won while Democrats are defending zero seats in states that Trump won. I’m not saying Democrats can’t lose the Senate, just that if they lose the Senate they’ve almost certainly already lost the House that’s the least of their worries.
Second, with a little bit of luck we’ll have COVID behind us and the economy roaring again by November. That should help put some wind behind the Democrats’s backs – so long as they make sure to get credit for it. We’re on track to have more than 80% of the country vaccinated by November, as well as a good portion of the world, and inflation is expected to fall precipitously this year. The latest projection from the Fed had core PCE falling to 2.6% by the end of the year.
Third, the Republican party is tying itself to an unpopular President who’s voters only seem to only come out when he’s on the ballot. That’s bad politics, which Democrats can take advantage of. Democrats need to continue challenging Republicans on the extremism within their ranks and their refusal to disavow Trump. Most voters don’t like Trump. Democrats should take advantage of that fact and remind voters that Republicans are the ones who empowered Trump, appeased Trump, and are ultimately responsible for January 6th. Thankfully it looks like the 1/6 Committee is going to get to the bottom of things, and while the investigation should not be done with the aim of helping Democrats, it nonetheless will remind voters of what Republicans created.
A note on Virgina: There’s a risk of Democrats learning the wrong lessons from Virginia. The lesson isn’t to not talk about Trump, it’s to do so wisely. Instead of trying to tie every single person with an (R) in front of their name to Trump, Democrats should simply challenge them on some of Trump’s more anti-democratic tendencies and remind voters that, with Trump likely to run again in 2024, who counts the votes really matters.
Finally, Democrats have a President who’s intent on doing popular things. This is a good political strategy and should help them heading into November.
So if Democrats hope to win in 2022 they need to get COVID and the economy under control, remind voters of Republicans’ extremism problem, and continue to govern effectively and do popular things. But there’s one more piece to the puzzle: democracy reform.
Democrats need to reform democracy so they can govern effectively and, more importantly, protect the country from the anti-democratic forces threatening to pull it down. That means reforming the filibuster first and foremost. In today’s politics it’s difficult for any President to live up to expectations when the out-party has a blanket veto over their entire agenda and no incentive to work with them. I’ve often said that my goal for 2022 is for Democrats to lose because they deserve it, not because the system is rigged against them. Yes that’s a low bar, but it hasn’t always been clear that they would be able to meet it. While it looks like Democrats are going to dodge a bullet with redistricting, there’s still plenty of voter suppression and election subversion laws being introduced and passed in states that could make it more difficult for Democrats to win. Passing the Freedom to Vote Act, the John Lewis Voting Rights Advancement Act, as well as updating the Electoral Count Act are key, not only to Democrats holding onto their slim majorities, but to America remaining a democracy (obviously an ECA update won’t affect the midterms but it’s important nonetheless).
So will Democrats hold onto the House and Senate? Probably not, but all hope is not lost. In 1998 and 2002 the President’s party bucked history and actually gained seats in the midterms. In 1998 it was backlash to Republicans’ impeachment of Bill Clinton. Democrats can hope for a similar backlash to Republicans’ continued embrace of Trump and Trump-backed extremists. In 2002 it was because of 9/11 and the rally around the flag effect. While it’s hard to see Americans rallying around the flag for anything these days, maybe coming out of the COVID emergency will foster some similar goodwill (I’m not holding my breath).
There have been manyreports about which party stands to gain or lose from redistricting this year, but any redistricting analysis that doesn’t take into account how gerrymandered the maps already were is going to fail to capture how skewed they will be going forward. For instance, Texas’ delegation is not likely to dramatically change next year, but that doesn’t mean it’s map is not gerrymandered. Instead, to understand a party’s advantage we need to compare their map to what would be considered a”fair” map. That’s what I do for this article.
I take a look at the 10 most gerrymandered states and compare them to what would be considered “fair.” For “fair” maps I use the proposed map with the lowest efficiency gap according to FiveThirtyEight or, if no “fair” map was proposed, the “most proportional” map from Dave’s Redistricting. Finally, for the sake of this analysis, I’m only using maps that have been enacted or are just awaiting the governor’s signature. Here’s what I found:
North Carolina: Fair map: 6R-4D-4C Actual map: 10R-3D-1C Difference: +4R, -1D, -3C
So Republicans have gained a total of 19 seats so far in those 10 states and Democrats basically break even. The biggest loser seems to be competitive districts, of which there are 19 fewer under the gerrymandered maps.
Note: This is an imperfect analysis and Democrats are likely to pick up a few seats in New York, though some of that could be cancelled out by Florida, depending on how agressive Republicans want to be there. The other pick-up opportunity for Democrats is in Maryland, but it looks as though they could only really net one seat at most from gerrymandering. The conclusion remains the same. Republicans are the big winners of redistricting and competition is loser.
That is, of course, unless Congress passes the Freedom to Vote Act, which would end partisan gerrymandering. But they need to get it done by the end of the year before maps are solidified for 2022. With Congress’s year end to do list piling up that’s looking less and less likely.
Another way of looking at it is to compare the percentage of the vote won to the percentage of seats won. This is called the “seat bonus bias.” In 2012 Republicans won 48% of the House vote but they won 53.6% of seats, giving them a seat bonus of 5.6%. That was the highest in two decades. Now, of course, some of this bias is due to geography, and Democrats being inefficiently clustered in big cities, but it’s worth nothing that just two years earlier, in 2010, when Republicans gave Obama and Democrats their famous “shellacking” their seat bonus was only 4.2%.
A seat bonus is not unusual. Generally whichever party wins the majority also tends to enjoy a seat bonus and the more you win by the larger your bonus, but that wasn’t the case in 2018. Despite Democrats’ “blue wave” Republicans still got a seat bonus of .4%. The median district bias was even worse that year. Democrats should count their lucky stars that they won the popular vote by 7.3%. Had they won it by 3.9% they likely would have lost the House. In 2020 Democrats finally enjoyed their first seat bonus of the decade. A whopping .4%.
Obama easily won in 2012 and Democrats actually gained seats in the Senate. The American people chose Democrats to control the House too but because of gerrymandering Republicans were able to hold onto their House majority. The rest is history. Instead of starting his second term with unified Democratic control, Obama spent his second term seeing his agenda frustrated by a unified Republican block.
That’s why it seems insane to me that Democrats aren’t showing more urgency to get the Freedom to Vote Act passed, which would end partisan gerrymandering. There’s no hard and fast deadline for when the bill needs to get passed in order for it to effect 2022, but with maps already being enacted and the first primary coming up in March, the window is closing fast. If Democrats don’t act soon they’ll almost certainly lose their majority, and they’ll deserve it.
There’s a long way to go until 2022 and a lot can happen between now and then but so far, we haven’t seen the telltale signs of a backlash against the President’s party. Even with the Delta surge and the withdrawal from Afghanistan weighing Biden down, Democrats seem to be holding steady. The next big tests will be in Virginia and New Jersey, where voting in those governor’s races is already underway. If you thought you were going to get a break from election season, think again.
Wonkness Blog 