KEY POINTS FROM THIS ARTICLE
— The non-presidential party often performs better in off-year elections than it did in the previous cycle’s presidential race. The recent Virginia and New Jersey gubernatorial races are no exception, both this year and historically.
— This likely has more to do with an enthusiasm advantage generating better turnout for the non-presidential party than that party’s ability to attract considerable numbers of voters who supported the presidential party in the most recent presidential race.
— Based on exit polls, the Virginia electorate in 2021 was markedly more Republican-leaning than the 2020 presidential electorate.
Conversion vs. mobilization in 2021
Republican Glenn Youngkin’s victory in the 2021 Virginia gubernatorial election along with the surprisingly narrow win by Democratic incumbent Phil Murphy in the New Jersey gubernatorial contest shocked many observers of American politics. Both states had strongly supported Joe Biden in the 2020 presidential election, New Jersey by 16 points and Virginia by 10 points. The results of these two high-profile off-year elections have led to speculation that some key voting groups such as college-educated suburbanites that had swung Democratic in 2020 out of opposition to Donald Trump were swinging back to Republicans — a shift that could bode well for Republican prospects in the 2022 midterm elections.
The assumption underlying such analyses of the 2021 contests is that most of the shift between 2020 and 2021 was due to conversion: 2020 Democratic voters moving into the GOP column in 2021. There is another possible explanation for Republican gains in 2021, however — disproportionate partisan mobilization. According to the partisan mobilization hypothesis, off-year elections often produce a shift in turnout in favor of the party that does not control the White House. While turnout drops for supporters of both parties compared to a presidential election year, out-party voters are more motivated to turn out to express their discontent with the leadership of the newly elected or reelected president.
The mobilization hypothesis provides an explanation for the familiar pattern of midterm election gains for the out-party in the U.S. House and Senate. It also provides an explanation for the pattern of results seen in Virginia and New Jersey’s off-year gubernatorial elections: Between 1981 and 2017, the out-party’s candidates won 17 out of 20 gubernatorial elections in these two states. In these 20 elections, the average shift in margin toward the out-party from the presidential election to the following year’s gubernatorial election was 16 points in Virginia and 12 points in New Jersey — very similar to the shifts of 12 points in Virginia and 13 points in New Jersey between 2020 and 2021.
In this article, I will take advantage of exit poll data from the 2021 Virginia gubernatorial election (there was no exit poll in New Jersey) to examine the sources of the swing toward the GOP in that contest. These exit poll results — though imperfect, just like any other survey — allow us to estimate the proportion of 2020 Biden and Trump voters who shifted parties in 2021 and the impact of changes in the demographic composition of the electorate between the two elections.
Before turning to the exit poll results, it is important to note that there was a remarkable degree of continuity in the results of the 2020 and 2021 elections in Virginia. Figure 1 displays a scatterplot of Virginia Democratic gubernatorial nominee Terry McAuliffe’s vote share by Biden’s major party vote share across all 133 Virginia cities and counties. The results show that there was a correlation of .996, very close to the maximum possible value of 1.0, between Biden’s vote share and McAuliffe’s vote share. In other words, Biden’s vote share almost perfectly predicted McAuliffe’s vote share. McAuliffe lost because he consistently ran between 1 and 9 points behind Biden’s vote share. In 56% of Virginia cities and counties, McAuliffe ran between 4 and 6 points behind Biden (or 8 to 12 points in terms of margin). A scatterplot of the New Jersey results shows a very similar pattern. What this tells us is that the voting patterns in both elections were highly nationalized: gubernatorial voting patterns followed presidential voting patterns very closely.
Figure 1: McAuliffe 2021 vote share by Biden 2020 vote share in Virginia cities and counties
Source: Data compiled by author
The question raised by the data in Figure 1 is why McAuliffe consistently ran behind Biden across Virginia’s cities and counties. The answer, based on the Virginia exit poll data, is disproportionate partisan mobilization rather than conversion. Ninety-two percent of 2021 Virginia voters reported supporting either Donald Trump or Joe Biden in the 2020 presidential election and very few switched parties in the gubernatorial election. Ninety-eight percent of Trump voters reported supporting Youngkin while 95% of Biden voters reported supporting McAuliffe.
The shift of 5% of Biden voters to Youngkin versus 2% of Trump voters to McAuliffe would only account for a small fraction of the overall shift in results between 2020 and 2021. The vast majority of the shift in results was explained by disproportionate partisan mobilization. Simply put, 2020 Trump supporters were overrepresented and 2020 Biden supporters were underrepresented in the 2021 Virginia electorate. Thus, among respondents in the 2021 exit poll, 48% reported voting for Joe Biden while 44% reported voting for Donald Trump in 2020. Biden’s margin of 4 points among these voters compares with his margin of just over 10 points in the presidential election.
Table 1: Comparing the 2020 and 2021 Virginia electorates
Source: 2020 and 2021 Edison Research Exit Poll results reported by CNN
The reasons for the Republican tilt of the 2021 Virginia electorate are clear in Table 1, which compares the demographic characteristics and candidate preferences of the 2020 and 2021 Virginia electorates based on exit poll data from the two elections. The data in this table show that the 2021 Virginia electorate was substantially older, Whiter, and more rural than the 2020 Virginia electorate. The most dramatic difference between the 2020 and 2021 electorates involved their age distribution. Not only were the 2021 voters considerably older, but those young voters who did turn out in 2021 were far more Republican in their preferences than the much larger group that turned out in 2020.
The data in Table 1 show that the 2021 electorate was also considerably Whiter than the 2020 electorate. In addition, White voters who turned out in 2021 were considerably more Republican than those who turned out for the presidential election. There is less evidence here of any swing toward Republicans among nonwhites although the results for Latinos and Asians should be viewed with considerable caution due to the relatively small numbers of Latino and Asian respondents in these surveys.
Discussion and conclusions
The results of the 2021 gubernatorial elections in Virginia and New Jersey showed a strong swing toward the Republican Party. A 16-point Democratic margin in New Jersey in 2020 turned into a 3-point Democratic margin in 2021 and a 10-point Democratic margin in Virginia in 2020 turned into a 2-point Republican margin in 2021. However, county-level data show that voting patterns in the gubernatorial elections were very similar to those in the presidential election while exit poll data from Virginia show that very few Biden voters actually switched sides in the 2021 gubernatorial election. Instead, it appears that the shift in election results between 2020 and 2021 was due largely to disproportionate partisan mobilization — stronger turnout among Republican voters than among Democratic voters in the off-year elections.
Republicans won in Virginia and came surprisingly close to winning in New Jersey because Republican voters were more energized than Democratic voters. In the California recall election, in contrast, Republicans apparently did not enjoy a major advantage in turnout and the GOP-sponsored recall effort fell flat. These results suggest that the results of the 2022 midterm elections will depend primarily on the ability of Republican and Democratic candidates to mobilize their own party’s supporters more than their ability to convert supporters of the opposing party.
 In this analysis, I am using results from Edison Research Exit polls conducted for a consortium of media outlets. There was a separate exit poll in 2021 conducted by the Associated Press and Fox News. However, I was unable to obtain detailed crosstabs for the 2021 gubernatorial vote for the AP/Fox News Poll.
|Alan I. Abramowitz is the Alben W. Barkley Professor of Political Science at Emory University and a senior columnist with Sabato’s Crystal Ball. His latest book, The Great Alignment: Race, Party Transformation, and the Rise of Donald Trump, was released in 2018 by Yale University Press.|