It’s not too presumptuous to start talking about a “wave election” for this November’s midterm. For one thing, wave elections aren’t even that unusual in midterm elections; going back to the Civil War, the average loss for the party that controls the White House in a midterm is 32 seats. The presidential party invariably thinks “but this time it’s going to be different!” And then, invariably, partisans for the party in power get complacent or get their feelings hurt, members of the party out of power get enraged and feel their very existence is threatened, and swing voters freak out that the party that they just voted for is either failing to fulfill its promises, or is fulfilling its promises too aggressively. It’s a phenomenon political scientists call thermostatic public opinion.
But we’re also seeing all sorts of flashing red lights on the instrument panel. Retirements in the House—especially among Republican members, who can probably sense things are going to go not so well this November—are starting to pile up. In fact, they’re coming at an unusually high rate; with the retirements of Darrell Issa and Ed Royce this week, we’re already up to 30 GOP retirements, and most states haven’t hit their filing deadlines yet.
Generic ballot House polling shows Democrats opening up a wide advantage; different aggregators will give you different results, but most see a low-double-digit lead. FiveThirtyEight, for instance, currently sees a 10 point advantage for the Democrats, though some recent polls have seen spikes as high as 17 percent. And the recent track record for Democrats in special elections last year, both for House seats and state legislative seats, is also consistent with a generic ballot lead in the low teens, based on how much vote shares have swung since previous elections.
So, you might be wondering, which Republicans are most likely to lose? There isn’t a lot of information, unfortunately, that tells us much about individual races. There are a handful of recent Public Policy Polling polls of key races, though many of those polls only test our good friend “Generic Democrat.” (Unlike some previous recent years, there’s no shortage of Democrats willing to challenge GOP incumbents. There’s a lot of uncertainty because of that surplus, in fact; in many races, we simply don’t know yet which of the various talented Democratic candidates will emerge from the primary.) What we can do, though, is turn to the Daily Kos Elections House Vulnerability Index, which combines data from previous elections to assess which seats are in the greatest danger in the next election.
If you’re a regular Daily Kos Elections reader, you’re probably already familiar with the House Vulnerability Index; we’ve deployed it in 2016, 2014, and 2010, and each time it worked pretty well at pointing toward which seats were in the most danger. Keep in mind, it doesn’t predict how many seats each party will lose, and it doesn’t even predict whether or not a particular seat will be lost. What it does do is predict the order in which seats are likely to fall.
For instance, in 2016, it turned out there wasn’t much of a wave one way or the other; the Democrats gained seats in the House, but it was only a net gain of six seats, as they picked up nine but also lost three. However, of the eight Republican-held seats that the HVI deemed most vulnerable heading into 2016, the Dems picked up five of them. That started with easy pickups in FL-10 and VA-04, two seats turned Dem-friendly in court-ordered mid-decade redistricting that the prior Republican occupants fled from, leaving them as open seats. These seats were No. 1 and No. 2 on the list. Democrats also picked up IL-10 (at No. 3, the most vulnerable seat that still had an incumbent, Bob Dold), NV-04, and the open seat in NV-03. (The three seats in the top tier where the Republican survived were IA-01, where, to almost everyone’s surprise, Rod Blum held on, thanks to presidential coattails in Iowa, FL-26, and the open seat in NY-19.)
The four other seats that Democrats gained were a little higher up the list; one wasn’t that much higher, in the form of NH-01 at No. 13, a swingy seat that has flipped almost every two years for a decade. Higher up were two long-time incumbents who were coasting and didn’t realize until too late that they were in difficult races against talented opponents (Scott Garrett at No. 35 in NJ-05, and John Mica at No. 60 in FL-07), and one race that was in a very Dem-friendly district—FL-13, which Charlie Crist picked up at No. 91—but where the HVI score was skewed by David Jolly running without Democratic opposition in 2014.
And the three seats that the Democrats lost were all within the six most vulnerable slots on the list: FL-02 at No. 1 (an open seat that Gwen Graham vacated after the seat was made much redder in redistricting), FL-18 at No. 3 (also an open seat, vacated by Patrick Murphy for his Senate run), and NE-03 at No. 6 (where Brad Ashford was the only Dem incumbent to lose). Between those, the Dems managed to hold the open seats in AZ-01 and NY-03, and protect their most vulnerable incumbent, Ami Bera in CA-07.
So let’s take a look at who’s most vulnerable in 2018, starting with, of course, the Republicans, since they’re likely to take the brunt of the blow this year. The way the index works is deceptively simple: it relies on only two factors, the average lean of the district in the previous two presidential elections, and the incumbent’s margin of victory in the last House election. Each factor is rank-ordered, and the rank orders are combined into one figure; those figures are then re-ordered to show vulnerability within each party. There’s one important tweak: for open seats, we always switch to “0” for the margin score; historically, open seats are when a party is at its most vulnerable, because the entrenched incumbent’s goodwill no longer applies.
The goal is to pinpoint members who are in both a swingy district and had a rough previous election. It downplays members who are in a district that’s pretty safe for their party but had a narrow election of their own last time (often because they were running for the first time; when you win an open seat or defeat an incumbent, it’s usually by a much narrower margin than once you’re entrenched). It also downplays members who are in swing districts but have gotten entrenched and have usually won by large margins anyway.
So here are the 50 most vulnerable incumbents. (This doesn’t mean I’m necessarily predicting a 50-seat gain for the Democrats this fall. A likelier gain is, for example, 34 seats, based on a +10 aggregate generic ballot advantage, according to poli sci professor Alan Abramowitz’s model for converting generic ballot polling into seat gains. Again, all this is predicting is that if there is, say, a 34-seat gain, the large majority of those seats that flip will be among these 50.)
HVI Rank |
DISTRICT |
REP. |
DIST. LEAN |
‘16 MARGIN |
TOTAL HVI |
1 |
FL-27 |
OPEN (Ros-Lehtinen) |
3 |
0 |
3 |
2 |
CA-39 |
OPEN (Royce) |
10 |
0 |
10 |
3 |
WA-08 |
OPEN (Reichert) |
12 |
0 |
12 |
4 |
NJ-02 |
OPEN (LoBiondo) |
13 |
0 |
13 |
5 |
CA-10 |
Denham |
9 |
5 |
14 |
6 |
AZ-02 |
OPEN (McSally) |
14 |
0 |
14 |
7 |
CA-49 |
OPEN (Issa) |
15 |
0 |
15 |
8 |
VA-10 |
Comstock |
8 |
9 |
17 |
9 |
TX-23 |
Hurd |
16 |
2 |
18 |
10 |
IA-01 |
Blum |
6.5 |
12 |
18.5 |
11 |
CO-06 |
Coffman |
5 |
15 |
20 |
12 |
CA-25 |
Knight |
11 |
11 |
22 |
13
|
FL-26 |
Curbelo |
1 |
22 |
23 |
14 |
MN-02 |
Lewis |
21 |
3 |
24 |
15 |
CA-21 |
Valadao |
2 |
28 |
30 |
16 |
NY-19 |
Faso |
20 |
13 |
33 |
17 |
PA-08 |
Fitzpatrick |
19 |
16 |
35 |
18 |
MN-03 |
Paulsen |
6.5 |
30 |
36.5 |
19 |
NE-02 |
Bacon |
37 |
1 |
38 |
20 |
MI-11 |
OPEN (Trott) |
42 |
0 |
42 |
21 |
ME-02 |
Poliquin |
24 |
17 |
41 |
22 |
PA-15 |
OPEN (Dent) |
42 |
0 |
42 |
23 |
IA-03 |
Young |
17 |
30 |
47 |
24 |
NJ-07 |
Lance |
27 |
21 |
48 |
25 |
KS-03 |
Yoder |
34 |
19 |
53 |
26 |
NM-02 |
OPEN (Pearce) |
58 |
0 |
58 |
27 |
PA-06 |
Costello |
25 |
35 |
60 |
28 |
NY-22 |
Tenney |
54 |
7 |
61 |
29 |
FL-18 |
Mast |
46 |
18 |
64 |
30 |
PA-16 |
Smucker |
47 |
20 |
67 |
31 |
OH-12 |
VACANT (Tiberi) |
68 |
0 |
68 |
32 |
FL-06 |
OPEN (De Santis) |
70.5 |
0 |
70.5 |
33 |
NY-24 |
Katko |
4 |
73 |
77 |
34 |
OH-16 |
OPEN (Renacci) |
78.5 |
0 |
78.5 |
35 |
IL-06 |
Roskam |
22 |
56.5 |
78.5 |
36 |
PA-07 |
Meehan |
18 |
60.5 |
78.5 |
37 |
NC-13 |
Budd |
56 |
23.5 |
79.5 |
38 |
CA-45 |
Walters |
31 |
52 |
83 |
39 |
IL-12 |
Bost |
48 |
36 |
84 |
40 |
GA-06 |
Handel |
78.5 |
6 |
84.5 |
41 |
CA-48 |
Rohrabacher |
41 |
46 |
87 |
42 |
TX-07 |
Culberson |
63 |
25 |
88 |
43 |
NY-01 |
Zeldin |
44 |
44 |
88 |
44 |
MI-08 |
Bishop |
39 |
50.5 |
89.5 |
45 |
FL-15 |
Ross |
53 |
38 |
91 |
46 |
NJ-03 |
MacArthur |
23 |
69 |
92 |
47 |
CO-03 |
Tipton |
60 |
33.5 |
93.5 |
48 |
IL-13 |
Davis |
29 |
64.5 |
93.5 |
49 |
NY-23 |
Reed |
55 |
40 |
95 |
50 |
NJ-11 |
Frelinghuysen |
32 |
63 |
95 |
As you can see, open seats are extremely valuable. This year’s open seats include seats that have bedeviled Democrats for a decade or more, where there’s been a consistent Democratic lean at the presidential level but where incumbents with a carefully cultivated image and selectively moderate votes have been able to hang on (like Ileana Ros-Lehtinen, Dave Reichert, and Frank LoBiondo). They also include the two that popped up this week (Darrell Issa and Ed Royce), which only very recently shifted into the blue column at the presidential level, and where the very conservative incumbents seemed ill-equipped to try and handle that.
You might be surprised to see that Barbara Comstock, of Virginia’s 10th district, isn’t the most vulnerable incumbent who’s still running for re-election. After all, you probably saw that, in the 2017 Virginia election, nearly every Republican House of Delegates member that overlaps her district in Loudoun and Prince William Counties got bounced out in a spectacular housecleaning.
However, she’s only the second-most vulnerable incumbent, trailing a somewhat more obscure member, Jeff Denham, who represents the 10th district in California’s Central Valley. Denham’s district, in the Modesto area, is one of those California districts with a Latino plurality but low turnout, and a white minority that’s pretty conservative. This has insulated him in previous midterms even though he’s had a rougher time in presidential years … but he hasn’t faced a midterm with a Republican in the White House before, so we’ll have to see how much his electorate changes this year.
Keep in mind that this list is subject to change, especially because of new open seats (we adjust our spreadsheet with every retirement, so whenever there’s a big one, check back and see how much the numbers change). Some high-profile retirements don’t change the calculus much; for instance, Darrell Issa’s retirement didn’t move CA-49 much because he was already near the top of the list even as an incumbent, after barely squeaking by in 2016. In others—like Frank LoBiondo, who flattened his 2016 opposition by 22 points—retirement makes a big difference.
In fact, you can try and identify which future possible retirements would be the biggest difference-makers by subtracting the “president rank” column from the “margin rank” column. John Katko, a locally-popular Republican in NY-24 is a prime example: his district is No. 4 for “president rank” but No. 73 for margin rank. If he retired, NY-24’s score would plunge from 77 to 4, and it would suddenly become the GOP’s second-most vulnerable seat (instead of its 33rd most vulnerable). David Valadao, another elusive target in the mostly-Latino seat of CA-21 in California’s Central Valley, is another example: it’s the second bluest district held by a Republican, but his last margin score was 28. If he retires, the seat is basically gone; if he continues to stick around, though, it’s certainly no sure thing for the Democrats.
There’s one other more remote possibility for future changes: big changes to the presidential numbers if the districts themselves change in North Carolina and Pennsylvania, thanks to pending litigation over those states’ gerrymandered maps. If those states get redrawn (in ways that would create several more Dem-friendly seats in each state), our list would get very scrambled.
Let’s unpack the underlying Republican data a bit, so you can better see the inner workings of the model. On one side of the equation, we have the presidential numbers. Here are the ten “bluest” districts represented by Republicans:
DISTRICT |
REP. |
‘16 ReSULT |
‘12 RESULT |
AVG. MARGIN |
FL-26 |
Curbelo |
56.7/40.6 |
55.4/43.9 |
- 13.8 |
CA-21 |
Valadao |
55.2/39.7 |
54.6/43.5 |
- 13.3 |
FL-27 |
OPEN (ex-Ros-Lehtinen) |
58.6/38.9 |
53.0/46.3 |
- 13.2 |
NY-24 |
Katko |
48.9/45.3 |
57.0/41.1 |
- 9.8 |
CO-06 |
Coffman |
50.2/41.3 |
51.6/46.5 |
- 7.0 |
IA-01 |
Blum |
45.2/48.7 |
56.2/42.5 |
- 5.1 |
MN-03 |
Paulsen |
50.8/41.4 |
49.6/48.8 |
- 5.1 |
VA-10 |
Comstock |
52.2/42.2 |
49.2/50.8 |
- 4.2 |
CA-10 |
Denham |
48.5/45.5 |
50.6/47.0 |
- 3.3 |
CA-39 |
OPEN (ex-Royce) |
51.5/42.9 |
47.1/50.8 |
- 2.5 |
It’s a different cast of characters, for the most part, from the representatives who ran the closest races in 2016:
DISTRICT |
REP. |
‘16 RESULT |
MARGIN |
CA-49 |
Issa |
50.3/49.7 |
0.6 |
NE-02 |
Bacon |
48.9/47.7 |
1.2 |
TX-23 |
Hurd |
48.3/47.0 |
1.3 |
MN-02 |
Lewis |
47.0/45.2 |
1.8 |
CA-10 |
Denham |
51.7/48.3 |
3.4 |
NY-22 |
Tenney |
46.5/4.0 |
5.5 |
VA-10 |
Comstock |
52.7/46.9 |
5.8 |
CA-25 |
Knight |
53.1/46.9 |
6.2 |
IA-01 |
Blum |
53.8/46.2 |
7.6 |
NY-19 |
Faso |
54.2/45.7 |
8.5 |
You can see a difference between the ones who are vulnerable because of their district (Erik Paulsen and Mike Coffman are two other examples besides Katko and Valadao), versus the ones who are vulnerable because of their own selves (Darrell Issa was the top example, though he’s already hit ‘eject’ rather than run again; Claudia Tenney and Don Bacon other examples, though their numbers in 2016 are down partly because that was the first time they were elected). And finally, there are the ones who make both lists (Denham and Blum, for example).
Finally, let’s take a quick look at the most vulnerable Democrats. I’m not expecting a lot of them to lose, so we’ll just do a table of the top 15; anything beyond that looks really implausible. This isn’t to say that it’s impossible for any of them to lose, wave election not withstanding. Even, for instance, in 1994 and 2010, the Democrats did manage to pick off a few seats, either open seats or hopelessly overmatched incumbents. In 2010, for instance, they won the open seat in Delaware vacated by Mike Castle while also picking off accidental Rep. Joe Cao in New Orleans; in 1994, they even managed to pick off a non-accidental Rep., Ron Machtley, who merely had the misfortune of trying to represent Rhode Island as a Republican.
HVI Rank |
DISTRICT |
REP. |
DIST. LEAN |
‘16 MARGIN |
TOTAL HVI |
1 |
MN-01 |
OPEN (Walz) |
2 |
0 |
2 |
2 |
MN-08 |
Nolan |
3 |
1 |
4 |
3 |
NV-03 |
OPEN (Rosen) |
6 |
0 |
6 |
4 |
NH-01 |
OPEN (Shea-Porter) |
7 |
0 |
7 |
5 |
MN-07 |
Peterson |
1 |
8 |
9 |
6 |
NJ-05 |
Gottheimer |
4 |
6 |
10 |
7 |
FL-07 |
Murphy |
12 |
3 |
15 |
8 |
AZ-01 |
O’Halleran |
5 |
12 |
17 |
9 |
NV-04 |
OPEN (Kihuen) |
21 |
0 |
21 |
10 |
CA-07 |
Bera |
20 |
2 |
22 |
11 |
PA-17 |
Cartwright |
8 |
14 |
22 |
12 |
FL-13 |
Crist |
19 |
4 |
23 |
13 |
NH-02 |
Kuster |
17 |
6 |
23 |
14 |
AZ-09 |
OPEN (Sinema) |
24 |
0 |
24 |
15 |
NY-03 |
Suozzi |
14 |
10 |
24 |
Again, open seats are the most vulnerable, though here it’s mostly Democrats running for higher office (MN-01’s Tim Walz running for Minnesota governor, NV-03’s Jacky Rosen running for Nevada Senate, and AZ-09’s Kyrsten Sinema running for Arizona Senate), rather than running away from the 2018 election in terror. Beyond that, we have several old-timers running in rural Minnesota districts that trended sharply away from the Dems in 2016 (Nolan and Peterson, who I suspect will be much safer this year), and some newbies who only narrowly picked up seats in 2016 (Josh Gottheimer and Stephanie Murphy) but who might be insulated this year by running in well-educated suburban districts.