There shouldn’t have been many elections in 2023 other than special elections, but unfortunately, five states elect their state governments in odd-numbered years.1 In 2023, this included statewide races in Louisiana, Kentucky, and Mississippi as well as legislative races in Virginia and New Jersey.
The lessons that can be read in the tea leaves of the 2023 election results in the pursuit are limited. The main takeaway is that there are few signs of any real shifts in the electorate. The most important elections to examine in detail are the statewide results in Virginia and Pennsylvania. The national implications of these election results are limited but mixed: Pennsylvania’s 2023 judicial election results are a positive indicator for Democrats, Virginia’s 2023 legislative elections are a positive indicator for Republicans.
We should not accept the argument that Democrats “won” the 2023 and 2022 elections, nor the closely related claim that Republicans cannot win the popular vote based on election results from 2016 to 2023. Both major parties remain viable, and strategists on both sides should not stop trying to persuade a true majority to their side.
Virginia
In Virginia, all seats in both chambers of the state legislature were up for election. Democrats gained three seats in the House of Delegates and lost one seat in the Virginia Senate, ending up with the narrowest possible working majorities in both chambers (51-49 in the House of Delegates, 21-19 in the Senate). These majorities will have to contend with a Republican governor, Glenn Youngkin.
Due to the exceptionally large number of uncontested seats in the House of Delegates, the Virginia Senate election is the best proxy for a popular vote; adjusting for uncontested Senate seats, the best estimate of the “true” popular vote margin is about 52% to 48%, with a margin of error that makes it hard to differentiate from a tie.2 A small but real share of the American electorate prefers a divided government.
In 2020, Joe Biden won Virginia 54% to 44% and Mark Warner won his Senate race 56% to 44%. Relative to 2017-2020, Democrats have now underperformed in Virginia for three elections in a row.3 Donald Trump may have made Virginia look like a safely Democratic state, but the election of Glenn Youngkin in 2021 no longer looks like a wild outlier. Democrats cannot take Virginia for granted.
Pennsylvania
While Pennsylvania does not elect its governor or legislature in off-year elections, it had partisan judicial elections this year. Democrats won three key statewide judicial elections by margins ranging from 50,000 to 200,000 votes. Turnout was quite low, but this was the only one of the major 2023 elections that took place in a state that will very likely be a key battleground in 2024.4
US politics is countercyclical; a small but significant number of swing voters prefer divided government and usually vote against the party in power. There is a Democratic president at the head of the federal government, and Pennsylvania has a Democratic governor with one legislative chamber controlled by Democrats. The countercyclical advantage should belong to Republicans, so in addition to being a victory of material importance, this is a genuinely good indicator for Democrats.5
Additionally, conventional wisdom has been that Republicans were generally favored in lower-turnout environments. The fact that Democrats won statewide judicial elections by margins of up to 7% in a low-turnout election in a battleground state suggests otherwise. What this means is that Republicans should not hope for low voter turnout as a strategy for winning elections.
Ordinary outcomes
The most newsworthy election of the cycle were two key statewide referenda in Ohio, where voters added abortion rights to the Ohio constitution and legalized marijuana. These were major defeats for the Republican cultural agenda in Ohio, which has been a safely Republican state. However, they were expected defeats. Republicans, including Ohio Republicans, already knew that a majority of the population supports legal marijuana and a right to abortion before fetal viability.
Referenda passed by Californians over the objections of the California Democratic Party did not indicate a broader rejection of the Democratic Party in California. Similarly, Ohio’s 2023 referendum results do not indicate a broader rejection of the Republican Party in Ohio.6
In Kentucky, Democratic governor Andy Beshear was re-elected. Due to the fact that turnout was lower than in 2019, he earned 15,000 fewer votes than he did in 2019. His margin was larger, but the simple and boring explanation for Beshear’s improved performance is that he was a popular incumbent with high approval ratings.7
In Mississippi, incumbent Republican Tate Reeves was re-elected by a margin of 30,000 votes. His 2019 victory had been by a margin of 45,000 votes, with turnout being higher in 2019. Unlike Andy Beshear, Tate Reeves had low approval ratings, partly linked to a major scandal over the misuse of welfare funds. This provides a simple and highly individual explanation for his poor performance at the polls.
Supporting evidence includes the fact that Republicans improved their performance in many other statewide races in Kentucky and Mississippi. There were no broad gains for Democrats in Kentucky and Mississippi, nor any signs of a backlash against Republicans linked to abortion.
In Louisiana, the incumbent Democratic governor was term-limited. While Republicans won the governor’s seat, the result was very similar to Republican performance in other statewide races in 2019 and 2023, with few indications that much has changed in Louisiana. Here, Republicans picked up an open seat that would normally be Republican by baseline partisanship.
In New Jersey, there was no gubernatorial election, but there were legislative elections. With turnout substantially lower than in 2021, Democrats gained a small number of votes in a state that has been trending strongly in a Democratic direction for roughly thirty years, with a net gain of zero seats in the upper house and six in the lower house.8 This returns the state legislature to the same balance of power it had after the 2019 election.
The Republican Party is not doomed
I have recently heard many talking heads and opinion-mongers claim that the Republican Party is doomed, and that the Republican party has done nothing but lose recently.9 It’s true that if Republicans choose to nominate Trump a third time, they are likely to win fewer votes for president a third time in a row. Almost every Republican politician running against Trump for the nomination would do better at the polls than Trump would.
While Trump lost the popular vote in 2016, the Republican Party as a whole won the popular vote if we look at the US House of Representatives.10 Republicans also won more votes in the 2022 midterms. These have been framed as a “loss” for Republicans because it was a narrower victory than Republicans wanted, but the fact of the matter across the United States in 2022, Americans voted for Republicans by about 51% to 48%.11
Republican presidential candidates may have only won the popular vote once in the last eight presidential elections, but Republican congressional candidates have won more votes in ten out of the last sixteen federal election cycles.12 Republicans winning more votes is not a rare anomaly; it just happens that Democrats have done a better job of choosing electable presidential candidates.13
Democrats should not count Republicans out and take votes for granted; Republicans should not give up on trying to gain the support of a true majority of voters.
As I discussed here and here, choosing to put elections in lower-turnout years is a terrible idea. State and local elections should synchronize as much as possible with the federal schedule, with offices elected during presidential election years for governors and other statewide officials with four-year terms and at least during even-numbered years otherwise.
The raw popular vote margin was actually only 0.4%, but this includes three races not contested by Democrats and only two races not contested by Republicans. The extra uncontested seat can be crudely estimated as contributing about 48,000 votes to the Republican side of the tally. With such a small margin, it’s not in any way clear that Glenn Youngkin would have lost if he were on the ballot; overall, he outperformed fellow Virginia Republicans in 2021, and a small but real percentage of Americans are fans of divided government.
The gubernatorial election in 2021 was R+2, and the net popular vote for Virginia’s US House seats was D+2 in 2022. This stands in contrast Ralph Northam winning the governor’s mansion by a 9% margin in 2017 and Tim Kaine winning a US Senate seat by a 16% margin in 2018. Adjusting for uncontested seats, the net margins in the US House looked like about D+7 in 2020 and D+10 in 2018.
Pennsylvania has been a perennial political battleground state since 1796 with rare interruptions, so this is a fairly safe prediction to make.
Note in particular that if the 2023 election results are litigated, the Pennsylvania Supreme Court is likely to be involved.
That said, if Ohio Republicans double down hard enough on trying to reject the will of the voters, it could cost them badly at the polls.
Another notable potential factor is that he restored voting rights to 180,000 Kentuckians convicted of non-violent felonies. Some number of those voters might have felt personally grateful.
This is complicated by uncontested races, incredible variation in vote totals from district to district, and the fact that New Jersey just redistricted with a
Pointedly, this was a major talking point at the last Republican debate. I’ve been hearing about the inevitable demographic demise of the Republican Party for a couple of decades at this point.
The raw vote totals are 49% to 48%; if we adjust for seats in which Republicans did not have a candidate, the projected national generic Republican vote in 2016 exceeds 50%.
Adjusting for uncontested seats increases this margin.
To address a popular talking point, Republican congressional majorities were not elected by gerrymandering in 2016 - Republicans not named Trump simply did that much better at the polls.
Whatever their reputations are now, both Bill Clinton and Barack Obama were highly charismatic on the campaign trail, well-organized, and importantly ran as moderates (regardless of what their reputations are among partisans after their presidencies). Similarly, Donald Trump was very clearly unpopular before he won the Republican nomination.
Those three candidates’ exceptional attributes account for six of the last seven times that Republicans lost the popular vote for president. For the seventh, I would contend John McCain was the better choice of candidate in 2000 & likelier to win the popular vote against Al Gore.