How Much Do Democrats Have to Win The House Popular Vote By To Get A Majority In 2018?

The unsolved question for who controls Congress after 2018

The popular vote in the US Presidential election gets a lot of attention, due to its ability to be different from the actual result. But there’s another popular vote that functions very similarly, that doesn’t get a lot of attention. And that would be the House popular vote. This essentially is a total of all the votes for Republican and Democratic candidates in the 435 different congressional districts. Democrats are aiming to try to win back the house in 2018 and unfortunately, unlike senate races or the presidential race, it’s hard to poll individual house races and thus, it’s often to tell how close a party is to winning a majority. The best tool we have to measure the house however is by polling the house popular vote, also referred to as the generic congressional ballot. 

Unfortunately, the house popular vote isn’t perfect and due to things like geographic sorting and gerrymandering, winning a majority of the house popular vote in no way guarantees a majority of seats in the House of Representatives. Most political observers realize that Democrats are at a disadvantage when it comes to the House for those aforementioned reasons. In the 2012 Presidential election, Obama won the popular vote by 3.9% but did not win a majority of Congressional Districts (he won 211). Also in 2012, Democrats won the house popular vote by 1.2%, but did not win a majority of seats (they won 201). In 2016, Hillary Clinton won the presidential popular vote by 2.1% but did not win a majority of seats (she won 205). So, this raises a much debated question that is yet to be answered: how much do Democrats need to win the house popular vote by to get a majority in 2018? Let’s take a look: 

The data we have

This map has been (mostly) used in five different elections, three house elections in 2012, 2014, 2016 and two presidential elections in 2012 and 2016. If we put the five data points into a linear regression line, Democrats would need to win by 7.3% in order to capture 218 seats. Now there is an interesting difference when you separate it into house elections vs. presidential elections. In both 2012 and 2016, the Democratic candidate would have needed to win by 5.5% nationally, assuming a uniform swing (which never happens), to have won a majority. In 2012, the median seat was Barbara Comstock’s VA-10, which Obama lost to Romney by 1.6%, while in 2016, it was Scott Taylor’s VA-02, which Clinton lost to Trump by 2.4%. Thus, on a presidential level, it appears much easier for a Democrat to win than in the house. Using 2016’s house data, Democrats would have needed to win the house popular vote by about 11 points to get a majority. Why? Part of it has to do with the incumbent’s advantage. 

The Incumbent’s Advantage 

Everyone knows that incumbents have an advantage in elections, but how much does it factor into the house? A decent bit, no question as we noted earlier, the median seat was R+5.5 in both the 2012 and 2016 Presidential elections, but was R+11 in the house elections. But how do incumbents fair in wave elections? They aren’t safe, that’s for sure and the composition of their district is big. Let’s take a look at the 2006 Democratic Wave. In that election, 22 GOP incumbents were defeated:

2006 Wave: Uniform Swing = 11% 

CD	Candidate	1st Elected	04 margin	06 Margin	swing	pvi
AZ-05	JD Hayworth	1994	21%	-4%	25%	R+4
Ca-11	Richard Pombo	1992	23%	-6%	29%	R+3
CT-02	Rob Simmons	2000	8%	-0.04%	8%	D+8
CT-05	Nancy Johnson	1982	22%	-13%	35%	D+4
FL-22	Clay Shaw	1980	28%	-4%	32%	R+4
IN-02	Chris Chocola	2002	10%	-8%	18%	R+4
IN-08	John Hostettler	1994	9%	-22%	31%	R+9
IN-09	Baron Hill	2004	0.4%	-4.5%	5%	R+7
IA-02	Jim Leach	1976	20%	-3%	23%	D+7
KS-02	Jim Ryun	1996	15%	-4%	19%	R+1
KY-03	Anne Northup	1996	23%	-2%	25%	D+2
MN-01	Gil Gutknecht	1994	24%	-6%	30%	R+1
NH-01	Jeb Bradley	2002	27%	-3%	30%	EVEN
NH-02	Charlie Bass	1994	20%	-7%	27%	D+3
NY-19	Sue Kelly	1994	33%	-2%	35%	R+1
NY-20	John Sweeney	1998	32%	-6%	38%	R+3
NC-11	Charles Taylor	1990	10%	-8%	18%	R+7
PA-04	Melisa Hart	2000	27%	-4%	31%	R+3
PA-07	Curt Weldon	1986	19%	-7%	26%	D+4
PA-08	Mike Fitzpatrick	1992	12%	-1%	13%	D+3
PA-10	Don Sherwood	1998	Uncontested	-6%	106%	R+8
TX-23*	Henry Bonilla	1992	40%	-9%	49%	R+4

Now among these 22 defeated incumbents, we see that 7 were in seats with a Democratic PVI (interestingly enough, 7 incumbents hold D PVI seats in 2018), 1 was an even seat (there’s 4 of those in 2018), and 14 were seats with Republican PVI. Perhaps most interestingly, just two of the swings were smaller than the uniform swing as a whole, and many were quite large. Another interesting tidbit is that many of these incumbents were multi-term incumbents, showing that no amount of experience is enough to save you in a wave. That fact comes in handy for Democrats trying to defeat long-term incumbents like Frank LoBiondo, Fred Upton, and Dana Rohrabacher, to name a few.

I guess the point from this is that while incumbent’s advantage helps house members when compared to the presidential by CD, don’t automatically assume that someone is safe, even if they’ve served there for many years, as Jim Oberstar and Jim Leach would tell you. In my experience in looking at past results, it seems that the incumbent’s advantage helps the most during rather tame years, like 2016, where there aren’t wild swings. On the flip side, in wave years like 2006 and 2010, the incumbent’s advantage doesn’t do much for you and can be toppled rather easily. Don’t look at someone like Leonard Lance, who won by 11 points in 2016 even though Hillary won his district by 1.5 and assume that the incumbent’s advantage will protect him. As we saw in this table, and in a theoretical table I could make of 2010 data, it doesn’t give that much protection. 

Also, scandal helps. A few of those incumbents were scandal ridden and their ousting was greatly helped by the scandals taking them down. While you should never bank on scandal, the fact is that it does happen rather regularly and there are some incumbents who already look like they could be in trouble due to scandal (ahem, Dana Rohrabacher and Chris Collins)

Finding Good Candidates

The biggest key to winning the house is having good candidates, especially ones who can win in hostile territory. Tip O’Neill was famous for saying that “all politics is local”, which may not be true anymore, given how closely tied presidential results have become with down ballot races. That said, there is an exception to every rule. While Collin Peterson’s blue dog status annoys some progressives, he votes with Democrats far more often than you’d expect for someone representing a district that Hillary lost by 31 points. That type of representative, someone stealing a red seat, is crucial to cobbling together a majority. If Democrats want to win the house with a smaller end of the popular vote (somewhere around 5-6%), then they have to be able to seriously contest a lot of seats. The best strategy is to legitimately challenge somewhere in the neighborhood of 80 GOP seats, recruiting good candidates for each one. Remember Scott Taylor’s VA-02, the very seat that was the median house seat at the presidential level, one that Hillary narrowly lost? Yeah, Democrats barely even tried to win that seat last time around. That can’t be the case in any seats that are even remotely competitive in 2018. Recruiting so far has gone very well, but it needs to continue. 

A look at DDHQ’s Model

Yesterday, Decision Desk HQ unveiled their 2018 House model, which made me alter this piece. They concluded that assuming current polling, which has Dems +8 in Generic Ballot (according to HuffPo; some places, like FiveThirtyEight, have it at +10), would yield 206 Democratic seats, far short of a majority. Thus, this model essentially assumes that it is very hard (shall we say almost impossible?) for Democrats to win a majority, given that a +8 romp doesn’t even come close to a majority. Now what’s going on with this model? Well, for one, it isn’t very transparent. No methodology is listed and the author, G. Elliott Morris appears to be new at this (though he made a 2016 election model which had Hillary with a >85% chance of winning). Basically, there isn’t much of a track record to go off of here. 

As for the model itself, it assumes Dems will pick up the usual suspect seats: a bunch in California, the two in Minnesota, the two in SoFlo, Bacon, Hurd, Comstock, and Coffman, while losing Tim Walz’s seat. The model appears to be based on a combination of 2016 results at the prez level and congressional level. It then gives an expected Democratic vote share and thus a percent chance of winning. What’s a tad confusing is the discrepancy between vote share and chance of winning. Let’s take Pennsylvania’s 6th for instance. Democrats are projected to get 49% of the vote, yet only have a 41% chance of winning the district. In Pennsylvania’s 7th, Dems are projected to get 48%, yet win only 37% of the time when simulated. In Iowa’s 3rd, it’s a similar story: 48% of the vote, yet a 36% chance of winning. I’m no master statistician nor creator of statistical models, but something seems a tad funky about that. Regardless, this is one person’s projection and we shouldn’t put too much weight into it. FiveThirtyEight will be building their own model sometime next year and let’s just say, I will put a lot more trust into that one as opposed to this one. 

Conclusion

While it’s true that simply winning a majority of the popular vote won’t necessarily net Democrats a majority, it’s also true that the House could very much be in play. Geographic sorting and gerrymandering hurt, but they aren’t too much to overcome, as shown by the Presidential-level results that suggest a very competitive chamber given a solid margin of victory, one we’re seeing in current polling. As Nate Silver says in this piece, a good night could be anywhere from 15-50 seat pickup. The House is really wonky. There’s some Reps that we are sure are going to lose who somehow hold on, and there are Reps who we think are safe that somehow lose. 

Sorry to give an anti-climatic answer, but there is no magic number when it comes to winning the house. If I had to break it down, I’d say it goes like this:

- Win by 5-6 points: You’re competing, but don’t have a great shot, somewhere like a 25% chance of a majority. In the neighborhood though, and maybe you get lucky with a bunch of close wins 

- Win by 7-8 points: Toss-up. House majority probably is a nailbiter and rests on a late night of counting those California seats. Basically a 50/50 shot of who controls the House. 

- Win by 9-10 points: Democrats are the clear favorites to win the House, somewhere like a 75+% chance of winning the house. 

- Win by 10+: Very, very hard to imagine a scenario where Dems get a greater than 10 point win and don’t get a majority. In all honesty, this is probably 40-50 seat gain territory 

So there’s your answer. Rather than focusing on silly models, focus on campaigning and canvassing. Recruit good candidates, run on a strong message, and win by a lot.