Portland Neighborhood Resources
US House 2022 Election Analysis
The US House has 435 members, all of whom are up for re-election every two years. However, incumbency is a significant advantage in the House, so that many of these elections are not competitive.
Currently the Democrats hold 221 seats, the Republicans 212, and there are two vacancies. So if the Republicans gain 6 seats they will hold the majority 218-217. It is typically in the off-year (non-Presidential) election year for the party opposing the President to gain 20 to 40 seats, so it had been anticipated that the Republicans would gain a significant majority this year. However, so far this has not been a typical year. Many believe that the typical low turnout of the off-year election will not occur this year.
In the spring, general belief was that Republicans would be highly motivated to vote this year. While that still appears to be true, Democratic turnout s expected to be large because of the continued presence on the political scene of the former President, and the US Supreme Court ruling overturning Roe v. Wade. Therefore, let’s examine two trusted election prediction sites, the Cook Political Report and fivethirtyeight, and compare their predictions for common ground.
While of course their methodologies and data are different, we can see from their summary data that Cook rates 31 seats as toss-ups. 538 rates 5 races as leaning Democratic, 13 as toss-ups and 11 as leaning Republican, for a very similar total of 29. Indeed, Cook calls 212 races for the Republicans and 192 for the Democrats (212+31+192=435), while 538 calls 209 for the Republicans and 197 for the Democrats. Since a majority is 218, the predictions agree that the Republicans need to pick up a single-digit (6 or 9) net gain of seats to gain the House majority.
The remaining question is whether the actual races identified as toss-ups (31 for Cook, 29 for 538) have substantial overlap or not. So I have gone into each firm’s (published) data to list and compare these races. Because 538 ranks all 435 races from most to least competitive, I have added races 30 and 31 in their list to this analysis to increase (potential) comparability.
In the table at the right, data is presented on the 18 races both Cook and 538 call toss-ups. The data is organized by the races 538 calls the closest, in order. The district column gives the state and the House district within that state. The next two columns give the Republican and Democratic candidates; an (I) represents an incumbent and an * names the current leader in the race. The final column gives 538’s analysis as of September 25, 2022. The “leader win percent” is the number of times their model shows the leading candidate winning in 100 runs of the model. So, for example, in the top line, the closest of all 435 House races, the Democrat Riley wins in 51 runs and the Republican Molinaro wins in 49.
But why does the 538 model get different answers in different runs? Some models are “deterministic”, meaning when you put in the data, the model calculates an answer, and that’s the end of the process. However, life has variability, as we all know. So some models use “Monte Carlo simulation” (named after the famous gambling location in Monaco), where random numbers are used to simulate different probable outcomes of each of the terms the model recognizes as variable. The model generates a different result each time. Those results are summarized in a probability distribution, and the “most likely” result (roughly speaking, the middle trial in the distribution) is considered the result of the modeling.
Notice in this table that the 6th most competitive race in the nation is in Oregon’s Fifth District. This district was noticeably adjusted in redistricting. In the Democratic primary, Congressman Kurt Schader was solidly defeated by Jamie McLeod-Skinner. In the general election, McLeod-Skinner faces the former mayor of Happy Valley, Lori Chavez-DeRemer, a MAGA Republican. This race covers part of Multnomah County, and so is of considerable interest to local Democrats. In short, because Democratic control of the US House is at risk in this election, we must work to preserve this long-held Democratic seat. In particular, we must not send an election denier to Congress. Like never before in US history, this election challenges our right to have our voices, and our votes, heard. Literally, we must vote to save our democracy.
Races Both Firms Call Close
| District | Republican | Democrat | Leader Win Percent |
|---|---|---|---|
| NY-19 | Molinaro | Riley* | 51 |
| NY-22 | Williams* | Conole | 51 |
| VA-02 | Kiggins* | Luria (I) | 52 |
| CA-27 | Garcia (I)* | Smith | 53 |
| CA-22 | Valadao (I) | Salas* | 54 |
| OR-5 | Chavez-DeRemer* | McLeod-Skinner | 54 |
| KS-3 | Adkins* | Davids (I) | 55 |
| NM-2 | Herrell* | Vasquez | 56 |
| AK-01 | 2 candidates* | 2 candidates Peltola (I) | 59 |
| ME-2 | Poliquin | Golden (I)* | 63 |
| IL-17 | King | Sorenson* | 66 |
| CO-08 | Kirkmeyer* | Caraveo | 68 |
| NC-13 | Hines* | Nickel | 69 |
| NY-1 | LaLota* | Fleming | 70 |
| PA-17 | Shaffer | Deluzio* | 72 |
| PA-08 | Bognet | Cartwright (I)* | 74 |
| OH-09 | Majewski | Kaptur (I)* | 74 |
| MI-07 | Barrett | Slotkin (I)* | 75 |
Races Only 538 Calls Close
| District | Republican | Democrat | 538 Win Percent | Cook Rating |
|---|---|---|---|---|
| AZ-02 | Crane* | O'Halleran (I) | 52 | Lean R |
| PA-07 | Scheller* | Wild (I) | 54 | Lean R |
| TX-15 | DeLaCruz-Hernandez | Vallejo* | 54 | Lean R |
| IA-03 | Nunn* | Axne (I) | 55 | Lean R |
| MD-06 | Parrott | Trone (I)* | 62 | Likely D |
| NJ-07 | Kean* | Malinkowski | 63 | Lean R |
| MI-03 | Gibbs | Scholten* | 68 | Strong D |
| NY-18 | Schmitt | Ryan (I)* | 68 | Lean D |
| CA-45 | Steel (I)* | Chen | 70 | Lean R |
| NY-03 | Devolder-Santos | Zimmerman* | 72 | Lean D |
| OR-06 | Erickson | Salinas* | 74 | Lean D |
| MI-10 | James* | Marlinga | 75 | Likely R |
| WI-03 | Van Orden* | Pfaff | 75 | Lean R |
Continuing with our analysis, we now look at the 13 races only 538 calls competitive. Notice that in 12 of the 13 races, the 2 firms call the race for the same candidate. While they may differ on how competitive the race is, for the twelve races on which they agree, we can attribute those races to the predicted outcome: 7 Republicans and 5 Democrats. That leaves TX-15 as the only district on which the firms disagree.
Note also that this table includes Oregon’s Sixth District. This is a new district, formed when Oregon’s population growth earned it another seat in the 2020 Census. Here, Andrea Salinas, a Democratic member of the Oregon House since 2017 representing the 38th District (much of Lake Oswego), is running against Republican political newcomer Erickson. Erickson was previously a Republican candidate in Oregon’s Fifth District, losing to incumbent Darlene Hooley in 2006 and to Kurt Schrader in 2008.
So now we look at the 13 races only Cook calls toss-ups. Because unlike 538, Cook does not provide the details of their “toss-up” conclusions, at least not publicly (they, like 538, are of course a for-profit business). So here we simply adopt the 538 predictions for these 13 seats: 9 Democrats and 4 Republicans are elected.
Note that these 13 races include 7 incumbents, 4 Democrats and 3 Republicans. 538 expects all of these incumbents to be re-elected. Given the power of incumbency, this prediction seems quite reasonable.
To put 538’s predictions in context, in the Chavez-DeRemer/McLeod-Skinner race described above, the 54% win percent attributed to Chavez-DeRemer is associated with a final vote prediction of 50.4% Chavez-DeRemer and 49.6% for McLeod-Skinner. The 74% win percent attributed to Andrea Salinas involves a final vote prediction of 52.8% for Salinas and 47.2% for Erickson. The 82% win percent attributed to Magaziner in RI-02, in the middle range of win percents in this table, reflects a final vote prediction of 51.9% for Magaziner and 44.1% for Fung, with a third candidate taking 3.9%.
These races are relatively close. In most cases, we have not yet had candidate debates, but will have them in October. So the analyses of Cook and 538 are appropriately called predictions, because these races are yet to be run.
Races Only Cook Calls Close
| District | Republican | Democrat | 538 Win Percent |
|---|---|---|---|
| CA-13 | Duarte | Gray* | 77 |
| IN-01 | Green | Mrvan (I)* | 86 |
| MI-08 | Junge | Kilde* | 81 |
| MN-02 | Kistner | Craig* | 80 |
| NH-01 | Leavitt | Pappas (I)* | 85 |
| NV-03 | Becker | Lee (I)* | 74 |
| NV-04 | Peters | Horsford (I)* | 78 |
| OH-13 | Gilbert* | Sykes | 85 |
| RI-02 | Fung | Magaziner* | 82 |
| WA-08 | Larkin | Schrier (I)* | 85 |
| AZ-01 | Schweikert (I)* | Hodge | 91 |
| NE-02 | Bacon (I)* | Vargas | 86 |
| OH-01 | Chabot (I)* | Landsman | 84 |
Let’s summarize what we’ve found, using 538’s numbers. Solid Republican 209 seats; solid Democrat 197 seats; leaving only 29 seats in question (paragraph 4). We added two back in paragraph 5, races 30 and 31 in their ranked list (WI-03 and MI-07), leaving 31 to resolve, 18 in Table 1 and 13 in Table 2. In Table 1, 538 has 10 Republicans and 8 Democrats leading (even though these races are very close). In Table 2, 538 has 7 Republicans and 6 Democrats leading. Thus, if all races end as 538 predicted, the House would have 226 Republicans and 211 Democrats. But we must remove WI-03 (R) and MI-07 (D) as they are double counted. Thus, 538 predicted in September a House with 225 Republicans and 210 Democrats, giving Republicans a 7 seat majority.
The 3 tables above identify the races the Democrats must concentrate on to retain their majority, or at least limit the size of the Republican majority. One obvious place to look is the 10 Republicans in Table 1, and in particular the McLeod-Skinner/Chavez-DeRemer race in Oregon’s District 5.
Sources: https://projects.fivethirtyeight.com/2022-election-forecast/house/