Introduction
Understanding the extent to which the national environment changes from election to election is a critical component of post-election analysis. While changes between presidential elections can easily be understood and quantified because of the nearly-uniform election systems in each of the 50 states, the swing from a presidential election to its following midterm involves additional degrees of complexity.
Firstly, comparing the vote for the President to the vote for the Congress in itself is a flawed comparison. Voters generally know more about the specifics of their presidential candidates than the ones vying for their congressional vote. Presidential candidates often have characteristics which differentiate themselves from a “generic” congressional candidate, which in some locales may inflate the baseline for performance in either direction. In the 2020 Presidential election, former President Trump ran behind congressional Republicans in many suburban areas, but ran ahead of them in many rural areas.
Secondly, a national round of redistricting and a number of uncontested seats prevent observers from making a clean comparison between the Novembers of this year and two years ago. Not only are district toplines unusable as a means for comparison, many seats simply did not have the Democrat-versus-Republican matchup that is necessary for a proper year-to-year comparison. Whatever “topline” result determined by adding official district-by-district results together hides the fact that many partisan voters did not see their favored choice on the ballot.
These two points necessitate the creation of a standardized method to determine a true generic congressional ballot, one that can be understood as the result of asking the entire country, “Would you prefer a Republican or Democrat as your representative in the House of Representatives?” That’s exactly what Split Ticket did earlier this week. The “Substitutional House Aggregated Vote Estimate” (SHAVE for short) is a modeling method which determines the generic popular vote by substituting vote totals in uncontested races with an estimate of the outcome of a generic matchup. These are then aggregated into state- or national-level figures.
This has been explained in a previous article, but for the uninitiated, the way we calculated the generic ballot involved modeling the results in districts where voters did not see either a Democrat or Republican on their ballots. We started with vote totals from either the concurrent or most recent Presidential election. We then shifted these totals according to the shifts seen between the presidential and congressional vote in other districts in each respective state. Then we applied a turnout modifier, again derived from other districts in each state, to normalize the generated total to a congressional total. We then re-inserted each new total into a table to read off the adjusted national generic congressional ballot.
In that previous article, we examined the results of the 2022 election, which concluded earlier this month, as Georgia conducted its runoff election for a Senate seat, and California finally finished counting its votes. In this article, we look at the results from the 2020 election, and look at an objective measure of how the national environment truly swung across those two years.
The 2020 Election
A lot of material has been, could be, and will be devoted to the 2020 Presidential election, but this article will limit its scope to the final topline figures. Our goal is simply to understand the shifts from then to last month’s election.
While the figure listed on Wikipedia indicates that the Democrats won the popular vote by 3.13%, the SHAVE indicates that the true popular vote margin lands around 2.08% in favor of Democrats. This 1.05% red shift is largely due to the unregistered votes from unrepresented Republicans in the 18 seats without GOP representation on the ballot. Compare that figure to the presidential contest, where President Biden bested Trump by 4.28% (4.45% only when you include Washington, D.C., which does not have a voting representative in the House of Representatives). Biden, therefore, is estimated to have outperformed his party’s congressional candidates by a margin of just over two percent. While two percent may not seem like a lot, Wikipedia lists eight Republican-won House districts as being within that margin in 2020, which makes a world of difference when a party only has a buffer of four seats in their House majority.
The above map shows the results of this metric applied to each of the 50 states. The most salient aspect to note is the Republican victories in Wisconsin, Georgia, Pennsylvania, and Arizona — four states which would have carried President Trump to reelection had he won them as we estimate the sum of generic House Republicans would have done.
Beyond that glaring difference, it should also be noted that the gap between the modeled congressional and presidential votes exceeds five percent in both directions among the states. In Hawaii and Maine, Biden ran over five percentage points behind Congressional Democrats. In Nebraska, North Dakota, Louisiana, Ohio, Washington, and Utah, the opposite is true. These tend to be smaller states with incumbent representatives running for reelection, which seems to have helped local delegations outperform their presidential correspondent.
The Swing
And now, we unveil the centerpiece of this article: the shift between our estimates for the generic ballot between 2020 and 2022.
At this stage, only simple arithmetic is needed. The SHAVE of 2022 is R +1.64%, which combines with the prior D +2.08% SHAVE to yield a total swing of 3.72% to the right from 2020. This figure is far below the 7.11% gap between the Presidential margin of D +4.28% and the apparent margin of R +2.83% from the Cook Political Report’s National House Voter Tracker. Again, this is because uncontested seats in 2020 would have net Republicans votes, while in 2022 that category would have net Democrats a larger share of the popular vote, evening out to reveal a milder shift between election years.
Of course, this effect was pronounced differently in each state. The map above gives a state-by-state “swing” map showing to which party, and to what extent, each state’s vote swung. The majority of the country favored Republicans more than it had in 2020, helping the House Republican secure a narrow majority, most notably in Florida and New York, where Republicans gaineda combined total of seven seats. There are some exceptions to this trend, of course.
Specifically, readers will note that Alaska’s Democratic representative, Mary Peltola, posted what amounts to a D +19% swing from 2020. We can say, based on other election results, the environment in Alaska certainly did not swing to that extent, but a more frigid environment was certainly in store for Republicans this year, where Gov. Dunleavy won with only 50.3% of the vote, Democrat-backed Republican Lisa Murkowski won against the Republican-favored candidate, and bipartisan coalitions now have control over both chambers in the state legislature.
Statewide candidates seem to have had coattails across the country. Among gubernatorial candidates, Democrat Gov. Polis of Colorado pulled his state further left than in 2020, and Republicans Rep. Zeldin and Gov. DeSantis helped bring red waves to New York and Florida, respectively. Federal candidates may have also had an effect. Senator Hassan (NH) and Rep. Ryan (OH) likely contributed to moving their states’ congressional votes to the left from above, on the Senate line of the ballot.
As for broader regional trends, Republicans performed stronger in the South, in the New York City Metro, the Upper Mississippi River Valley, and the Far West. Democrats improved their performance in some of the Plains States, the Mountain West, and in many of the Great Lakes states. These regional trends could be explained as the result of a whiter electorate and a greater Republican vote share among nonwhites who did turn out, helping Republicans in states where nonwhites are a larger share of the Democratic base. White liberals did not stay home, however, and were seemingly energized in purple and red states, where abortion rights were understood to be under greatest threat from a possible conservative wave.
The above trend map shows each state’s swing normalized to the national swing. Large states like California, New York, Florida, and even Texas trended Republican, and crucial, mid-sized swing states trended towards the Democrats. Alaska, Colorado, Florida, and especially New York stick out as outliers, swinging drastically while most of the country generally remained close to the national average.
Developing a model for the national congressional vote helps political observers understand, contextualize, and most importantly, quantify, the changes in each state’s environment between elections. By developing the SHAVE and proving its utility in comparing results between cycles, Split Ticket now adds another data processing method to our toolkit, which can now be used to examine past and future elections with an objective lens.
Those who are interested may consult our 2020 calculation spreadsheet, found here.
Addendum: tabulation errors led to a incorrect SHAVE of D +2.25. The corrected SHAVE is D +2.08, and our numbers and maps have been corrected to reflect this figure.
I make election maps! If you’re reading a Split Ticket article, then odds are you’ve seen one of them. I’m an engineering student at UCLA and electoral politics are a great way for me to exercise creativity away from schoolwork. I also run and love the outdoors!
You can contact me @politicsmaps on Twitter.