Wednesday, June 22, 2022

BTRTN, the Midterms: A Close Look at the House

Tom is back with his latest look at the midterms.  His first midterm article, back in February, set the stage, and the second, in April, took an early look at the Senate.  Now Tom’s focus is on the House. 


Once again it is time for that ritual of American politics, the evisceration of a first-term president in his (yes, all “his” so far) first midterms.  Joe Biden seems right on course to suffer the fate of so many of his predecessors, to lose big in the House of Representatives on November 8, 2022.  Barring some truly monumental change of fortune, Biden will likely lose his very thin Democratic majority in the House and quite a few more seats beyond that. 

We rather immodestly point out that BTRTN has a very good track record at predicting House midterms gains and losses, as evidenced by the chart below.  We owe it all to a very good model, which we will explain further along.


Our updates, which will become more frequent as Election Day nears, will provide the BTRTN view on two questions:  1) if the midterms were held today, what are the odds that the Democrats will retain the House? and 2) just how many seats would the Democrats gain or lose?  (We will do the same for the Senate, and for the gubernatorial races as well, but today we focus only on the House.)

Today’s answer to question #1 is, well, almost zero.  Our model spit out a number that rounds up to 1% -- a 1% chance that the Democrats will retain a majority in the House.  It’s that unlikely, right now.

Today’s answer to question #2 goes something like this:  while all 435 House seats are up for reelection (as they are, of course, every two years), given the advanced state of the fine art of gerrymandering, only 88 races have any chance at all of being truly contested.  More realistically, only 55 races, as of now, appear to be truly “in play.”  Most districts have been sculpted into ungainly salamander shapes that all but guarantee Blue or Red status.  Of those 55 in which it remains truly possible for either party to win, the Democrats hold 42.  There are three other seats that they are almost guaranteed to lose (not included in either the 55 or the 88).  So the very worst the Democrats could do is to lose all 45 of those seats and not flip any current GOP seats.

We at BTRTN don’t envision that kind of bloodbath, at least not at this point in time.  Given the dynamics of the generic ballot and the net effect of redistricting, we think the Democrats, if the election were held today, would lose around 26 seats.  Remember and note well:  this is not a prediction, but rather a snapshot.   


There remain nearly five months until Election Day, five months in which it is theoretically possible for potential “catalysts” to move the macro-electoral environment in the Democrats’ direction and forestall the blowout.  Inflation could at least begin to recede without prompting a recession; gas prices could tumble.  Vladimir Putin could throw in the towel in Ukraine.  Omicron might run out of new mutations, with no new nasty variant worthy of its own Greek letter emerging behind it.  The baby formula supply crunch could disappear.  The Senate might find its way to some version of a Build Back Better Bill, and, in the wake of Uvalde, the Senate’s 25 years of embarrassing inaction on gun reform might finally end (which is looking particularly promising right now).  And, the combination of the January 6 Committee’s live hearings and final report, plus the final version of SCOTUS Judge Joseph Alito’s Roe-reversing opinion might actually set the Democrats on fire, galvanizing the Blue electorate to turn out as if Trump himself were on the ballot.

Any of this could happen.  But the odds of most of them happening -- and happening in time -- to prevent the GOP House takeover seems unlikely.   More likely is that Joe Biden’s approval rating and, more importantly, the almighty generic ballot will continue at their current levels, with Biden holding steady in the low 40% range, and the GOP up by +3 percentage points over the Dems.  And if that happens, the final outcome will indeed resemble where we stand today.  (And, of course, things could get even worse.)



Let’s drill down on the BTRTN House snapshot at this point, focusing on the three major variables driving our analysis (and our model).

1)     The generic ballot.  The first key variable is the generic ballot, in which voters are asked, essentially, if they prefer Democratic or Republican representation, without naming names (hence, “generic”).  The generic ballot is exceptionally accurate in predicting House midterm elections, at least when embedded in a multiple regression equation model such as our very own BTRTN model.  Once upon a time, through October of 2021, the Democrats led on the generic ballot, but that then reversed and the GOP has led since, and their narrow margin has widened a bit in the last few months.

Note that for the Democrats to retain the House, the generic ballot doesn’t simply have to favor the Democrats, but, given the inherent GOP bias in the congressional map, the Democrats actually have to lead by at least +5 points.  It is a long, long way from -3 to +5 in the remaining window before Election Day.

2)     Current seats held.  The second factor, which is fixed (more or less) through the campaign cycle, is how many seats the president’s party currently holds.  This has an enormous impact on how many seats the Democrats could lose.  That’s because, essentially, the more seats a party holds, the more seats it can lose.  Barack Obama lost 63 seats in 2010, but the Democrats held a remarkable 256 at the time.  Bill Clinton lost 54 seats in 1994, but they held 254.  In both those instances, Obama and Clinton switched a lot of middle-of-the-roaders to Blue when first elected, and many of those new-blue seats simply switched back at their first chance in the midterms.  On the other hand, George H. W. Bush lost only 8 seats in 1990 – but the GOP only held 175 at the time.  The GOP at the first Bush’s midterms held so few seats that they were down to all but a handful beyond the deepest red districts that they could lose.

Joe Biden’s Democrats only hold 220 House seats, plus one of the six vacancies was held by a Democrat, so call it 221.  That relatively low number alone will put a natural cap on the Dems’ potential losses – there is simply no way Biden can lose 63 or even 54 seats.  This variable – the number of seats the President’s party holds -- is also embedded in our regression model.

3)     Redistricting.  In most years, those two data points are all we would need to predict the House outcome – that is what we used to predict those House midterm outcomes in 2010, 2014 and 2018, in which we were virtually on the money each time.  We would simply plug those two data points – the 221 seats the Dems hold, coupled with the GOP +3 percentage point edge in the generic ballot – into our little black box and, voila, out comes the answer:  that the Dems would lose 22 seats.

But in 2022, we have another factor to consider: the redistricting that has been going on based on the 2020 census, a process that is just about complete.  Many feared that given GOP control of statehouses and state legislatures in many states, the GOP might gerrymander even more seats in their direction.  That has proven true, but not anywhere near the order of magnitude of the worst fears.  Most analyses show the GOP has, net net, picked up about 3-5 seats by virtue of redistricting alone.  Thus, to the 22 lost seats that our model suggests the Dems will lose, we add four more for redistricting, to get to our current estimate of 26 lost seats.

Of course individual races – the individuals the party’s elect to represent them via the primaries – make some difference.  We will be back a number of times in the election cycle to update our House estimates, taking the final match-ups into account.


Apart from simply keeping score, these assessments assist readers/volunteers to make informed judgments on where to spend their time and money in the electoral process to the best advantage.  Political donors and volunteers would do well to focus on races that are truly competitive and can be swayed with the precious resources that time and money represent.  Giving $50 to a favored congressperson who is running 80/20 over his opponent is not a good use of resource.

At some point – perhaps reasonably soon – Democratic donors and volunteers will have to decide whether the House outlook is so bleak that it will be better to focus entirely on holding the Senate, or perhaps on a few key gubernatorial races (or even state elections).  The Democrats have a far better chance of holding the Senate than the House, accordingly, as they are fielding generally excellent candidates in key swings states, while the GOP, not so much.  (Our Senate piece is only slightly dated and worth a read, and we will update it soon:

We will be explicit as the election season progresses with our guidance on this “resource allocation” question.

But if you are a Democratic volunteer/donor, and decided you simply must spend an hour or a dollar on a race right now, our advice is as follows – focus on three of the “Big Four” Senate races, the ones that will likely dictate the fate of Senate control this November where the Democratic headliner is set:  Arizona (Democratic incumbent Mark Kelly), Georgia (Democratic incumbent Ralph Warnock), Pennsylvania (Democratic primary winner John Fetterman).  (Wisconsin is the fourth, with the primary on August 9). Maggie Hassan of New Hampshire and Catherine Cortez Matso are also Democratic incumbents who are facing tough races and worthy of support. 

The Dems need to win four of these six races (or win fewer but flip a GOP seat or two, which is possible) to retain control of the Senate – and if they win all six, then the Dems will get to 52, and we won’t have to pay much attention to Joe Manchin and Kirsten Sinema anymore as the Dems drive through filibuster reform.

Right now we would say that Warnock probably needs the most help, as he is running about even with Georgia football demigod Herschel Walker, a colorful candidate who has zero political experience, one Heisman trophy and multiple personalities.



A midterm blowout would not be fatal to Joe Biden’s presidency, of course.  Many of his predecessors, as noted, suffered the same fate.  Three recent presidents – Ronald Reagan, Bill Clinton and Barack Obama – not only survived gigantic midterm blowouts, and were reelected two years later, but each rank among the top 20 most successful presidencies ever (according to the latest round of historian polls in 2021 – Reagan was #9, Obama #10 and Clinton #19).  All were able to adjust, receive a run of good luck, or both.  Donald Trump, on the other hand, who made no adjustments whatsoever to his loathsome, divisive style, lost 42 seats in the 2018 midterms and lost his reelection bid.

But there is little doubt that getting shellacked (Obama’s term for his loss of a whopping 63 seats) is a warning shot over the bow from the American people.  The midterms are a barometer of how a presidency is being received.  First-term presidents are not destined to get swamped in the midterms – George W. Bush picked up seats – rather they typically earn it through unpopular actions or presiding over a dip in the economic cycle.  Biden was dealt a horrific hand, and then drew two more events over which he had no control, the Omicron strain and the Ukraine invasion, which directly resulted in the inflation that bedevils him.  But whether he deserves blame or not, his approval rating is low, and needs to be rebuilt to at least 45% to feel confident about reelection.

The loss of the House would mean the loss of the Democrat’s “trifecta” – the control of the White House, Senate and House.  Now, the trifecta ain’t what it used to be, what with the ancient filibuster rules almost assuring the inability to pass landmark legislation, and the Supreme Court firmly in conservative hands.  But still, losing the House would negate the ability for the Dems to pass legislation via reconciliation in the Senate – if the Democrats were able to hold the Senate.

This is a good time to note that the Senate midterms are almost a completely different animal.  While that pesky micro-environment almost always dictates House fortunes, Senate outcomes, including the midterms, are far more driven by the candidates themselves.  The Democrats have a far greater chance of holding on to the Senate, roughly 50/50 odds at this point.  The primary season is underway, and we need to see what the final lineups will be, but we will return with many Senate updates in the coming months.

Stay tuned.

If you would like to sign up for email notification of our latest post, please contact us at


  1. In the cited examples of accuracy in prior midterms selection, what were the dates of those predictions - right at Election Day or at roughly the same time relative to Election Day as the prediction above?

    1. The night before Election Day. Remember, as I said repeatedly, this is only a snapshot, not a prediction, and should only be used to guide decisions on where to allocate donor/volunteer resources as of that moment,

  2. The ahistorical factor that complicates the prediction, in my reading, is Trump. Donald J. Trump, a two-time popular vote loser and two-time impeached President, is acting like no defeated President or defeated major party candidate of my lifetime. Nixon in '62, Humphrey in '70, McGovern in '74, Ford in '78, Carter in '82, Mondale in '90, Dole in '94, Gore in '02, McCain in '10, Clinton in '18 -- all of them stayed pretty far away from active campaigning for House members. When Trump is "on the ballot," it triggers negative partisanship. He is actively involving himself in primaries and shows every indication of continuing the involvement in the general election campaign.

    And on "generic ballot" surveys -- now lists 13 surveys with data gathered after the January 6 hearings started on June 9. 8 give Democrats the lead, 1 is "Even," 4 give Republicans the lead. [And 3 of the 4 are from YouGov or Rasmussen, using their approaches to determine who is a "Likely Voter."] Adding up the margins of all of the 13, it is almost dead even -- Democrats are one point up, or an average of 1/13th of a percent.

    1. John, in your first paragraph, you use the word "prediction." Again, as I state in the article several times, this was not a "prediction," but rather a "snapshot." I agree that Trump is more active than his predecessors, and that could affect the election, but I'm not sure how that would effect the methodology.

      As for the generic ballot, has the GOP up +2.3 right now, inclusive of the data you cite. I consider that to be a minor difference from the GOP +3 (actually +2.9) that I cite. RCP, BTRTN, 538 and others all use slightly different algorithms, and I'll stick with ours, though they all are similar, since we are all tracking the same polls. But even it is were "even," that would still suggest (via our model) that the Dems would still lose roughly 20 seats.

    2. re: Snapshot / prediction -- yeah, I was sloppy.

      On the other hand ... "But even it is were "even," that would still suggest (via our model) that the Dems would still lose roughly 20 seats." A model that says what an outcome would be 5 months in the future seems to be creeping pretty close to a mechanistic forecast.

      Colorado's primaries are tonight, and it appears that the Trump-iest candidates are generally NOT winning their races. Trump didn't formally endorse, as far as I know. But only Rep. "Bim" Boebert, an incumbent crazy, is winning, while the "stop the steal" candidates are losing races for the Senate, House 5 & 8, Governor and Secretary of State. I'm seeing Trump and the MAGA minions as divisive, and thinking he's a good reason why comparisons to previous election cycles are suspect.


Leave a comment