Introduction
Split Ticket is devoting December to detailed analyses of the latest House election results to shed light on the origins of the Republicans’ smaller-than-expected net gain and its potential 2024 implications. To better define the lower chamber’s districts and members, we are developing three interactive tools.
The first, a new WAR model, uses a wide variety of data to examine how nominees did relative to expectations – just like in baseball. Utilizing the WAR scores to judge candidate quality, Split Ticket’s second tool, which calculates District Splits, offers an in-house take on the partisan leanings of the newest congressional seats that should assist us during the race-rating process. The last resource, gauging House seat trends and quantifying flippability, will be introduced below.
We have also used House returns to better understand past election cycles in the environmental sense. Our new SHAVE metric calculated state-level generic ballots by projecting how bipartisan-uncontested seats might have voted had Republican and Democratic nominees been on the ballot. The corresponding articles proved that lower-quality Republican candidates lost many of 2022’s winnable statewide races and that the 2020 national environment was redder than President Biden’s 4.6 point popular vote victory implied.
What Is The “Flip Index”?
To clear up confusion about our Flip Index, we should start by describing how it differs from other political volatility metrics like our upcoming District Splits. The index exists to quantify how likely all 435 House seats are to change hands. PVI calculations, meanwhile, merely standardize baseline district partisanships. The difference is clear: one metric highlights the incumbent party’s risk level in each seat, the other just gauges competitiveness.
Given the lack of tools probabilistically measuring flippability, Split Ticket’s standard fills a valuable niche in the forecasting community. While the index scores should not be considered a perfect solution, they have intuitive predictive power. Together with the District Splits and WAR model, the Flip Index will make our 2024 House forecast more well-rounded and statistically-sound.
Before breaking down the methodology underlying the index, it is important to mention how we captured ongoing trends in our model. The Trend Score, as one might guess, contextualizes which party each House seat is currently trending toward. These shifts do not always immediately affect election outcomes, especially in safe seats, but they remain important to the long-term picture painted by the Flip Index.
Methodology
As the Flip Index probabilities depicted above suggest, “flippability” is quantified along a 0-5 numerical scale, with the likelihood of a party change increasing as the score rises. A median score of 2.5, for instance, would place a seat’s chances of flipping at 50%. Each seat’s score is derived from a weighted average of risk level values 0-5 assigned to four separate variables on a district basis: district lean, margin of victory, trend score, incumbent party.
Because flip scores are intended to focus more on individual seats than candidates, a divergence from the WAR model, district leans received the majority weighting in our calculations. We determined that variable using a weighted average of the 2020 and 2016 presidential partisanships for every seat, with more recent numbers receiving a higher weight. Risk levels were attributed to these leans on a five-level scale corresponding to the size of a party’s marginal advantage in each district. Because uncompetitive districts, those with presidential leans greater than +16 for either party, are inherently less likely to flip, they received numbers closer to zero.
Margin of victory and incumbent party, both receiving similar weights, provided a partial control for incumbency-driven candidate quality effects. To consistently define the first variable, we used our 2022 generic ballot calculations to account for uncontested seats and create a weighted average with 2020 presidential lean. This method slightly reduced variations in 2022’s actual results caused by midterm elasticity, ensuring candidate quality deltas had an important, but not overwhelming, impact on the final scores.* A similar system decided risk levels for each district in terms of the MoE variable, with larger victory margins reducing the likelihood of a future flip.
The incumbent party assignment distinguished between crossover seats and the rest of the House. Republicans and Democrats representing non-crossover districts predictably received lower flip scores than their counterparts. Remember, also, that the model’s crossover seat calculations are based on the adjusted 2022 leans described previously, not the actual results.
The last variable proved to be the most challenging to assign risk levels despite simple baseline calculations. Each seat’s Trend Score is a weighted average between its 2016-2020 presidential shift and the difference between the adjusted 2022 results and district leans. Risk levels were given first by grouping seats into crossover and noncrossover categories and then distinguishing between Republican and Democratic incumbents based on whether a seat’s Trend Score benefitted them.
*accounting for candidates is important to the score because an incumbent’s strength plays a fundamental role in determining how likely a seat is to flip at present*
Analysis
The results, which can be viewed here, generally align with expectations. Most of the House’s 435 seats are not competitive, a reality represented by the chamber’s low average flip score of 1.1 – sitting around 25%. Keep in mind, though, that that figure does not represent the much higher likelihood of House control changing hands at some point this decade. In terms of seats with scores above 2.5, those with a greater than 50% chance of hypothetically flipping, the vast majority were among the most competitive in the nation in 2022.
Arizona’s 1st district, with a score of 4.1, fit into this category because of its marginal baseline partisanship. Though Republican David Schweikert outran statewide nominees like Blake Masters and Kari Lake, his meager victory suggests that he will be vulnerable in future cycles. Other seats, like California’s 40th, show the limitations of the Flip Index with regard to candidate quality. While the district’s fundamentals place it in the same category as AZ-01, Republican Congresswoman Young Kim’s recent victory implies she should be able to win future reelections comfortably. In other words, astute observers must look at ulterior factors to determine where the flip score is most useful.
New Jersey’s 7th (4.0) and Ohio’s 9th (3.2) are two great examples of seats where the scoring system does provide valuable insight regarding long-term trends. The former district, encompassing some of the nation’s most affluent and well-educated suburbs, flipped to the Republicans last month. Former State Senator Tom Kean, whose primary campaign was criticized for moving too far to the right, defeated Democratic incumbent Tom Malinowski. While Kean still outran Trump across the board, his ancestrally-Republican ticket splitting potential in Somerset and Union counties declined. Given the fact that Malinowski matched or exceeded his margins in much of the turf carried over from the 2012 lines, Kean almost certainly would have lost without the favorable redraw. It is fair to assume, as the score suggests, that he will be in hot water come a neutral or Democratic national environment.
In OH-09, a Trump seat home to many traditionally-Democratic white working class voters, veteran Democratic Congresswoman Marcy Kaptur crushed J.R. Majewski by 13 points. While the size of Kaptur’s win suggests she would have beaten any Republican this year, Majewski’s personal flaws obviously made the job easier. The modest flip score notes that the seat has roughly a 60% chance of flipping based on its characteristics alone. Assuming Republicans find a better nominee in 2024, especially if Kaptur unexpectedly retires, presidential dynamics could mitigate the effects of down-ballot lag while increasing turnout. Of course, redistricting in the Buckeye State could change dynamics entirely. It’s all guesswork until then.
Conclusion
Both of these metrics are meant to provide context for House elections, which have historically been quite difficult to predict. While there is no reason to consider our flip scores flawless, since the data driving them must ultimately be subjectively interpreted, we believe that they do mesh with conventional wisdom enough to retain value. These figures will be updated for mid decade redistricting in Ohio and North Carolina, assuming those plans come to fruition.
Author’s Note
This index was updated on January 19th, 2023 to reflect the certified 2022 House election results.