Skip to main content
ABC News
Resist The Pundits: What The 2021 Elections Can (And Can’t) Tell Us About 2022

A sample size of two is never sufficient to get an accurate read of what’s happening in politics — let alone what will happen. But in the wake of the New Jersey and Virginia gubernatorial elections last week, that’s what everyone is trying to do. Still, we’re almost exactly one year away from the 2022 midterm elections, so the temptation is understandable. So, what can last week’s results tell us about the 2022 midterm elections? And what can’t they tell us?

CAN: Right now, the political environment is significantly more friendly to Republicans than it was in 2020

This is the simplest, surest and maybe most important takeaway from the 2021 elections. As we wrote last week, the electorates in both states shifted toward the GOP, leading to only about a 3-percentage-point reelection victory for Democratic Gov. Phil Murphy in fairly blue New Jersey and a 2-point win for Republican Glenn Youngkin in comparably purple Virginia, which marked the first statewide win for the Virginia GOP since 2009. 

Having looked through the results, we can be really confident that Republicans have an edge right now. And keeping in mind the historical tendency of the party holding the White House to face substantial headwinds in the subsequent midterm election, there’s plenty to suggest that Republicans could have a turnout edge in 2022 as the nonpresidential party. Speaking of …

CAN: Turnout is likely to be supercharged in 2022 — especially among Republicans

Last Tuesday’s vote demonstrated that Republicans are strongly motivated to turnout. We can see that in the chart below, where the greater the Republican vote share was in a county or city, the better turnout was compared to the 2020 election. While turnout fell everywhere from last November (you’d expect that going from a presidential contest — the highest-profile election in the U.S. — to an off-year election), participation tended to decline less in places that voted more Republican.

History shows that this pattern is unsurprising considering there’s a Democratic president in office — and an unpopular one at that. All else being equal, we’d expect the average Republican to be more likely to show up right now than the average Democrat. This “differential turnout” will be an important factor in the 2022 midterms, although just how much of a turnout edge Republicans will end up having remains to be seen.

But the broader turnout picture also suggests that our era of high turnout elections seems likely to continue in 2022. Among the nation’s voting-eligible population, more Americans cast ballots in the 2018 midterms (50 percent) and 2020 presidential election (67 percent) than in any midterm or presidential race since the voting age was lowered to 18 in 1971, according to the United States Elections Project. Yet while the 2017 elections in Virginia and New Jersey served as initial markers of high voter engagement early in Donald Trump’s presidency, those contests actually had significantly lower turnout than what we saw this year. Compared with 2017, Virginia’s turnout jumped from about 43 percent to around 53 percent of the voting-eligible population, a modern record for the state, while New Jersey’s increased from around 36 percent to roughly 41 percent.1 This doesn’t guarantee that 2022 turnout will outpace the 2018 midterms, but it does portend that participation levels will still be quite high compared with pre-2018 midterms.

Additionally, the results in Virginia and New Jersey should put paid to the idea that high turnout is a panacea for Democrats. New Jersey and Virginia lean Democratic as a default: According to FiveThirtyEight’s partisan lean metric, the Garden State is about 12 points more Democratic than the nation as a whole, while Virginia is D+5.2 As such, you might expect a higher turnout off-year election to come closer to falling in line with those partisan baselines. Yet with turnout up in both states from 2017, Republicans won in Virginia and fell just short in bluer New Jersey.

CAN: Democratic turnout may not be low in 2022, but matching Republicans will be tough

Still, considering the high overall turnout, it’s apparent that Democrats also showed up at a relatively high rate in many places — just not quite to the same extent as Republicans. For instance, Fairfax County, Virginia’s most populous locality and one Biden carried by 42 points last year, saw its vote total decrease by about the same rate as raw turnout fell in Virginia overall from 2020 (down about 27 percent). But looking ahead to 2022, turnout figures carried warning signs regarding two groups who usually lean pretty strongly toward the Democrats: voters of color and young people.

When it came to turnout and race in last Tuesday’s elections, places that were whiter tended to see higher turnout compared with the 2020 presidential race. For instance, in the cities and counties in New Jersey and Virginia that are majority nonwhite, turnout from 2020 fell by 39 percent, compared with a decline of 29 percent in places that were 80 percent white or more. Of course, given the swing toward the GOP in the 2021 elections and the tendency for more Republican-leaning places to see higher turnout, this isn’t surprising. After all, race is one of the strongest predictors of partisan identity — nationally, fewer than 1 in 5 Republicans are nonwhite, while about 2 in 5 Democrats are, according to a 2019 Pew Research study — so if GOP turnout was higher, we’d expect greater participation in whiter areas.

Virginia Lt. Governor-elect Winsome Sears

Related: Why White Voters With Racist Views Often Still Support Black Republicans Read more. »

Similarly, we see evidence that fewer young voters showed up to vote. Turnout dropped precipitously in college towns. Radford and Harrisonburg, small cities in western Virginia with sizable state universities,3 for instance, saw the ninth- and 10th-largest drops in turnout from 2020 across all 133 Virginia cities and counties. And while exit polls are imperfect, 18 to 29-year-olds made up only 10 percent of the electorate in Virginia’s exit polls, a decline from their share in 2020 and likely compared with 2017 as well, both comparatively more Democratic environments than the current one.

With all this being said, a disproportionate dropoff in turnout among voters of color and young voters in nonpresidential elections isn’t unusual. In midterm elections, the electorate has traditionally been whiter and older because a disproportionate number of voters who are young and/or from racial or ethnic minorities tend to drop out of the electorate. But because Democrats are more reliant on these voters as whites without a college degree — the largest part of the electorate — move toward the GOP, it’s become more critical for Democrats to find ways to ameliorate this midterm turnout decline.

CAN’T: The 2021 elections signal a GOP swing in 2022 — but not how much of one

For all the strong reactions to the New Jersey and Virginia results, however, it’s worth pointing out that it’s entirely normal for those two states to swing away from the president’s party — hell, it would have been strange if they hadn’t. In all but one of the two states’ 16 total gubernatorial elections since 1993, the party in the White House has performed worse than it did the year before in the presidential election. The one exception wasn’t exactly a dramatic one either, as the GOP lost the 2001 New Jersey governor’s race by about 15 points one year after losing by 16 points in the 2000 presidential contest. 

Still, Virginia is a more competitive state at its baseline and has more regularly moved away from the president’s party, so observers tend to look at it as a potential indicator of future midterm results. But for all the focus on Virginia, since the early 1990s it’s been an inconsistent predictor of the next midterm result. From 1993 to 2018, the average difference between the swing in Virginia’s gubernatorial election from the state’s partisan lean and the margin in the midterm national popular vote for the House of Representatives was just shy of 7 percentage points. That error margin — and remember, our sample of elections here is relatively small; there could be an even bigger difference in 2022 — represents the difference between a “meh” or even OK year for Democrats in 2022 and a complete wipeout in both the House and the Senate.

Virginia’s gubernatorial race doesn’t beget midterms

Difference between the swing in Virginia’s gubernatorial race from FiveThirtyEight’s partisan lean and the national popular vote in the subsequent midterm election for the U.S. House of Representatives

Cycle VA Gubernatorial swing MIDTERM Nat’l House Margin Gap
1993-1994 R+7.7 R+6.8 0.9
1997-1998 R+3.3 R+0.9 2.4
2001-2002 D+15.0 R+4.6 19.6
2005-2006 D+15.2 D+7.9 7.3
2009-2010 R+10.3 R+6.6 3.6
2013-2014 D+7.2 R+5.8 13.0
2017-2018 D+8.9 D+8.6 0.2
Average 6.7

The Virginia gubernatorial swing metric compares Virginia’s FiveThirtyEight partisan lean to the actual gubernatorial result.

FiveThirtyEight’s partisan lean is the average margin difference between how a state or district votes and how the country votes overall. This version of partisan lean, meant to be used for congressional and gubernatorial elections, is calculated as 50 percent the state or district’s lean relative to the nation in the most recent presidential election, 25 percent its relative lean in the second-most-recent presidential election and 25 percent a custom state-legislative lean based on the statewide popular vote in the last four state House elections.

Sources: Dave Leip, U.S. House of Representatives

Simply put, the Virginia result can’t tell us exactly what the national House vote is going to look like next year. It gives us only a rough idea of how the political winds might blow as we head into the midterm year.4

That makes sense, as the factors that boosted Republicans in 2021, especially Biden’s poor approval rating, could push voters toward them next year, too. Indeed, the scope of that rightward swing will depend greatly on Biden’s standing — if it improves between now and next fall, that could fend off some of the Democrats’ potential losses; if it stays where it is or gets worse, Democrats could be looking at another 1994 or 2010, when their party took a thorough shellacking in the House and Senate.

CAN’T: How coalitions will shift in 2022

Another thing the Virginia and New Jersey races can’t tell us with certainty is just which parts of the two parties’ coalitions are actually sticking with them or shifting disproportionately toward the other party. 

For instance, might white voters with a college degree stick with Democrats? The exit polls from Virginia suggest Democrats won a narrow majority among them, more or less the same as Biden performed in 2020.5 So it’s possible this group, which has trended toward Democrats in recent years, will hold up for the party come 2022. At the same time, an analysis of county-level data in the Virginia and New Jersey races didn’t show much of a relationship between how much Republicans improved over their 2020 showing in a place and how much of the population is white with a college degree. In other words, if Democrats’s support among white voters with a college degree held steady, we might have expected those sorts of places to see comparatively smaller Republican gains, but instead there wasn’t much of a pattern in the data to suggest that happened. 

And for all the talk of Democratic losses in the suburbs and rural areas, it’s worth noting that there was a relatively uniform swing toward the GOP across rural, suburban and more urban parts of both New Jersey and Virginia. Using FiveThirtyEight’s urbanization index,6 we categorized localities in both states into places that are more urban, suburban/exurban, or rural in nature, and found that while suburban and exurban places tended to move the most toward Republicans, the magnitude of those shifts were not all that different from urban and rural areas.7

Democrats didn’t just lose ground in the suburbs

Shift in vote margin between the 2020 presidential election to the 2021 governors races in New Jersey and Virginia, grouped by FiveThirtyEight’s urbanization index

2020 election 2021 election 2020-21 Shift
Type of place N.J. Va. N.J. Va. N.J. Va.
Urban -31.1 -41.5 -21.6 -30.3 9.5 11.3
Suburban/Exurban -1.6 -13.6 12.5 -1.4 14.1 12.3
Rural NA 29.4 NA 39.2 NA 9.8

New Jersey has no counties that fall into the rural category.

FiveThirtyEight’s urbanization index, a calculation of how urban or rural a county is, is the natural logarithm of the average number of people living within a five-mile radius of every census tract in that county. Places with an index of 12 or more were categorized as urban, 10 to 12 as suburban or exurban, and below 10 as rural.

Perhaps if suburban areas had shifted even more markedly to the right while more urban places stayed in the Democratic camp, we could more convincingly claim that the suburbs in particular swung hard for the GOP. Or if rural areas had moved further to the right than other types of places, we could better make the case that it was rural areas that cost Democrats in Virginia. But the lack of more sizable variations across types of place leaves us with a less obvious conclusion about the role of population density, at least beyond “the environment was good for Republicans.” Although, obviously, if the swings are like this in November 2022, that’ll be all the GOP needs to capture the House and Senate.

Lastly, we got little clarity on whether Hispanics continued to notably shift toward the GOP after moving to the right in 2020. The exit polls found that Democrat Terry McAuliffe won about two-thirds of the Hispanic vote, which didn’t differ that much from Biden’s showing in 2020. However, the Associated Press’s VoteCast caused a stir when it reported that Youngkin had actually carried Latino voters. The truth might lie somewhere in between, but fundamentally, only about 10 percent of Virginia’s population is Hispanic, so we also know the sample sizes of Hispanics were not especially large in either survey, which produces results that have a larger margin of error. And just to add to the confusion, consider some of the actual vote totals. On the one hand, after taking into account the overall swing toward the GOP, McAuliffe didn’t do exceedingly worse than previous Democratic statewide candidates in a handful of majority Hispanic precincts in Fairfax County, and the only place where Murphy more or less matched Biden’s margin was Hudson County (Jersey City), which is 40 percent Hispanic. At the same time, Murphy lost about the same amount of margin in Passaic County, which is 42 percent Hispanic, as he did statewide compared to Biden (14.5 points vs. about 13 points). There’s just not a clear picture.

The results out of New Jersey and Virginia signal that the electoral environment is clearly favorable to Republicans right now, although it can’t tell us just how pro-GOP it’ll be in a year’s time. There’s little question that Republican voters are energized and motivated to turn out to a greater extent than Democratic ones, and Democrats do have to worry about turning out important parts of their base, like young voters and voters of color. But the New Jersey and Virginia contests failed to really tell us just how certain demographic groups might vote next year, or whether they’re continuing to trend toward or away from either party. Part of that is down to having just two major elections to go off of. But it’s also down to a mixed bag of data that offers somewhat contradictory conclusions.

I don’t buy that Tuesday’s election night was good for Trump’s 2024 prospects: Silver


  1. These are estimates based on voting-eligible population data from the United States Elections Project, which will not publish new population figures until 2022.

  2. Partisan lean is the average margin difference between how a state or district votes and how the country votes overall. This version of partisan lean, meant to be used for congressional and gubernatorial elections, is calculated as 50 percent the state or district’s lean relative to the nation in the most recent presidential election, 25 percent its relative lean in the second-most-recent presidential election and 25 percent a custom state-legislative lean.

  3. Radford University and James Madison University, respectively.

  4. The correlation between that swing and the House vote margin since 1993 is .56, which suggests there’s a somewhat positive relationship between them.

  5. Exit polls are always at least a little bit prone to error — they are still surveys, after all — but this was especially true in the 2020 presidential election given how much of the electorate voted absentee. It’s likely this comparison isn’t apples-to-apples but it’s still a useful benchmark for understanding the shift among suburban voters in Virginia.

  6. FiveThirtyEight’s urbanization index is a measure of how densely populated a county or county-equivalent is based on the average number of people living within a five-mile radius of a given resident.

  7. We categorized places with an index of 12 or more as urban, 10 to 12 as suburban or exurban, and below 10 as rural.

Geoffrey Skelley is a senior elections analyst at FiveThirtyEight.