A quick guide to determining your predictions for the upcoming House midterms.
In nearly every midterm election, the party in power loses seats. Over the last four consecutive midterm elections (2006 – 2018), the president and his party have either lost the House or further lost the House. Before that, strange events disrupted the usual pattern of midterms multiple times.* So for now, we’ll stick to just the last four:
The average number of House seats the party in power lost was 37.
2006 (Bush 43): Dems net gained 32 seats & won the House.
2010 (Obama): Rs net gained 63 seats & won the House.
2014 (Obama): Rs net gained 13 seats & kept the House.
2018 (Trump): Dems net gained 40 seats & won the House.
In a modern blue wave year, Democrats essentially get the average. A red wave year, however, is very unpredictable. In the above list, the outlier is the 2014 election. Republicans already had the House. They gained back the few seats they had lost in 2012 and added a few more. That was their ceiling (247 seats total – the largest Republican majority since before Franklin Roosevelt). The Democrats in the House generally have a ceiling of about 235. Last time they had more than that was in the ’70s.
Red waves have only existed in modern history since 1994. Before that, Democrats controlled the House without interruption for 48 years. In 1994 (the next previous midterm in the pattern where the party in power loses in the House), the Republicans net gained 54 seats and won the House. So to the extent we can discern a pattern, the chances of Republicans winning big in the House are far greater when they don’t already have control of it.
Today Republicans have a narrow minority of 212 seats in the House of Representatives. Everyone knows they’ll win enough to be in the majority for the 118th Congress. But since the end of September, Real Clear Politics has steadily increased its projection of the average number of net gains we can expect by ten, and this looks to be continuing. The ceiling for Republicans went from 38 to 48. On the other hand, though Cook Political Report has also steadily been increasing the number of seats it projects to be competitive, its outlook is a little colder. If they’re to be believed, merely an average midterm net gain of 37 seats would mean that Republicans will win every toss-up race except one. For Republicans to win a “red wave average” (43 seats based on just the red waves since 1994), they would have to win every toss-up race and every race that is considered “Lean Dem,” except one. Possible? Certainly. But it’s a tall order. Keep in mind also the fact that Republicans must net gain 35 seats to reach the majority ceiling they were shown to have in 2014.
The trajectory described above is what momentum looks like, which brings me to the elephant in the room: polling.
It’s an open secret that polling is ineffective. Over the last four elections (2014 – 2020), it has consistently underestimated Republican turnout. Conservatives generally don’t talk to pollsters, even those they like. As such, we get polls that range from D+2 to D+8, and others where the Democrat is so far ahead, the sample simply cannot be believed. There are two rules for adjusting to this. The first comes from Ed Rollins: if you don’t believe a poll, order a new one. The second is my own: if the polls show a race tied or where the candidate from the party with the national momentum is down by (at most) 3 points, the party with that momentum wins. If the party that has the momentum is down by 4, the race is tied and it’s anyone’s guess. This rule is especially true when Democrats are ahead and without momentum, which was usually what we saw in most of those last several races, most notably 2016.
The wind is clearly behind Republican sails in this midterm cycle. But it’s still an open question how grossly the polls are underestimating Republican turnout. Do I expect this time to be different? No. But do I expect that they’re so far off that I need to modify my rule? Also no.
In short, the Republican Party is out of power in all three branches of government, not just the House, and has the advantage of being underestimated by the polls and the momentum of the electorate on its side. It has a recent history of making massive House gains in years exactly like this one, and keep in mind also that this entire analysis hasn’t mentioned even once the issues and policies being voted on in this cycle.
I’m not going to tell you how many seats to predict based on all of this. We’ll all find out in a couple of weeks. Here’s to it.
* In 1998 the Republican efforts to impeach Bill Clinton backfired, resulting in net gains for Democrats in the House. In 2002, the first midterm after 9/11, George W. Bush became the first president since Franklin Roosevelt in 1934 to see his party gain seats in both branches of Congress.