This article was co-written by Lucy Ding, Peter Jones, and Dominic Skinnion
Background
Tuesday’s election for Governor of Virginia is a state race in name only. As Republican Glenn Youngkin takes on Democrat and former governor Terry McAuliffe, the other 49 states look on, well aware of the implications that the election will bear for Democratic and Republican leadership. A decisive McAuliffe win would offer Democrats a chance to exhale and regroup; conversely, a Republican victory could foreshadow the party’s resurgence and add momentum to an agenda that currently derives its fuel in large part from President Biden’s declining popularity.
A State Race with National Reach
Because Virginia always holds its statewide elections one year after the presidential election, its gubernatorial result has long served as a referendum on the president’s first year in office. President Biden’s introduction to the White House, needless to say, has seen moments of turmoil and confusion — crime, inflation, and the contentious withdrawal from Afghanistan have all given Republicans and moderates reason to criticize the Biden mission. Still, because the U.S. has not been home to any truly competitive elections since last winter, the jury remains out on the extent to which Biden has lost his grip on reliably blue states like Virginia. Until this Tuesday, that is.
Elections — whether state or national — do not occur in vacuums. The result in Virginia will likely trigger a political shockwave that exerts some influence on later statewide elections, similar to that of Iowa and New Hampshire in the presidential primaries. With midterm elections on the horizon for Biden and the Democrats, this effect has the potential to rip control of Congress out of their hands.
Last Tuesday, lawmakers from Virginia and New Jersey (which will also hold its gubernatorial election this week) met in Nancy Pelosi’s office to discuss their anxieties regarding the pivotal nature of both elections. As Jonathan Martin of The New York Times reported, the representatives warned Pelosi that a McAuliffe loss “could have a cascading effect on the party, prompting Democrats to pull back from President Biden and his ambitious agenda, and perhaps even drive some to retirement.”
Moreover, a Republican presidential candidate has not netted Virginia in nearly two decades. Turning red in favor of Youngkin might put this trend in jeopardy come 2024.
Terry McAuliffe
McAuliffe presents himself as a classic Biden Democrat, though he remains unafraid to express his frustration with the presidential administration’s slipping hold on the general American populace. Having held the governor’s office in Virginia between 2014 and 2018, McAuliffe is not a political outsider nor is he lacking in experience with state politics. In the 2013 election, McAuliffe narrowly won the election with 51.3% of the two-party vote. Because Virginia, unlike any other state, prohibits its governors from serving consecutive terms, McAuliffe has spent the last four years outside the Virginia capitol. Now, he looks to retake it.
The former governor, a longtime friend of the Clintons, said earlier this month that his campaign faced “headwinds” born from President Biden’s worsening popularity. He added: “The President is unpopular today unfortunately here in Virginia, so we got to plow through.” Republicans ran with the remark, while Democrats fired at McAuliffe for denigrating the national party. Shortly afterwards, McAuliffe downplayed the statement.
Still, McAuliffe’s politics align quite closely with the left-of-center president. He has focused his campaign on civil and voting rights, healthcare, and climate change, all issues pertinent to the Democratic agenda. Additionally, McAuliffe has promised an “unprecedented investment in education,” likely a response to his opponent’s conservative, education-centric campaign.
Glenn Youngkin
Businessman and political unknown Glenn Youngkin is about as polar an opposite to McAuliffe as it gets. Youngkin spent 25 years at Carlyle Group, a private-equity firm, of which he would eventually become co-chief executive. “I’m not a politician,” he has noted correctly in campaign materials.
Youngkin walks a thin line between mirroring Donald Trump’s rhetoric and appearing moderate enough to retain Virginians with less radical political ideologies. Although Trump announced an online rally last week to garner pre-election Republican momentum, Youngkin avoided it, likely afraid McAuliffe would use it as campaign ammunition. Similarly, Youngkin has emphasized “election integrity” as a key issue throughout his campaign, but balks at saying fraud defined the 2020 presidential election. Both instances exemplify Youngkin’s tightrope walk between moderation and Trumpism — or as Times reporter Lisa Lerer put it, “giving the conservative Trump base 80 percent of what they want.”
Otherwise, the Youngkin campaign is based on traditional Republican principles: job creation, reduction of taxes, and law enforcement. His campaign has particularly honed in on crime, saying “defend not defund” in response to the defund-the-police movement.
Trump and Education: Two Points of Contention
The apparent quasi-alliance between Youngkin and Trump has been among McAuliffe’s favorite weapons on the campaign trail. The former governor loves to tie the two together whenever and wherever he can, a logical strategy given that Biden won Virginia by 10 points last November. When Youngkin RSVP’d “No” to Monday’s virtual Trump-Youngkin rally, McAuliffe taunted, “[Youngkin is] willing to pledge his loyalty to Trump in private. Why not in public? What’s he trying to hide? … Is he embarrassed?”
Meanwhile, Youngkin’s campaign arsenal consists largely of fiery rhetoric regarding education policy. Whereas McAuliffe supports vaccine mandates in schools and a curriculum that addresses race, Youngkin decries mandates, though not the vaccine itself, and dismisses the teaching of critical race theory.
Education has since risen to the forefront of Virginian electoral politics and comprised a large part of the battle between Youngkin and McAuliffe. The former governor called Youngkin’s dismissal of critical race theory a “racist dog whistle,” and, potentially in response to Republican attacks, underscored investment in education as a primary campaign issue.
Youngkin has also dug into McAuliffe’s voting record around education policy to support his offensive tactics. In 2017, McAuliffe vetoed the “Beloved Bill,” a piece of legislation that would have allowed parents to exert influence over whether their children study “sexually explicit” material in school. The bill’s nickname was born from a mother’s failed attempt to prohibit their child from reading Toni Morrison’s Beloved. Youngkin has tried to use the veto to his advantage, spouting advocacy for “parent’s rights” and releasing a campaign ad depicting the mother who attempted to ban the Morrison novel.
Long-Term Implications
The Biden administration is at a critical point in garnering midterm support for Democratic candidates. Four of Virginia’s last five governors have been Democrats. Accordingly, a Youngkin, Republican win would be devastating for the left.
A wave of conservative enthusiasm would be likely to wash across the country if Youngkin can pull off a victory on Tuesday. In its wake would come increased donations, a possible party-wide strategy shift to emulate Youngkin’s, and, possibly, a Republican Congress. Moreover, it could present Trump with a jumping-off point for a 2024 presidential campaign.
This election extends far beyond Virginia.
Forecast
Our forecast predicts that Glenn Youngkin will win the Virginia Governorship with a two-party vote share of 52.1%. The 95% confidence interval for this prediction is (43.1%, 61.1%), meaning that we are not confident that Youngkin will win the race; there is substantial uncertainty in our model. Our simulations predict that his chances of winning the race are 67.9%. Even though this may sound like a solid win, it is truly anything but. McAuliffe’s nearly 33% chance of winning, according to our simulation, is still a large probability, and thus we cannot rule out the possibility that McAuliffe pulls away with a win.
Methods
Using demographic data from the Iowa Community Indicators Program, the Centers for Disease Control and Prevention, and the University of Virginia Weldon Cooper Center, poll results compiled by RealClearPolitics, and historical election data from the Virginia Department of Elections, we performed three different simulations of the upcoming Virginia Gubernatorial Election.
Weighting polls by their sample sizes and their recency, we created a weighted running average of polls for each election for which poll results were available; as RealClearPolitics had compiled polls since 2005, there are four elections on which we can train models which use polls. Demographic data are available from 1993 to 2020. The demographic variables used are the percent of the population that identified as white and the unemployment rate in the year prior to the election, as we do not have data available for 2021. The election results included the Democratic percent of the vote share and Republican percent of the vote share. We transformed these variables to create a response variable which is the two-party vote share of the Democratic candidate. This is equal to the Democratic percent of the vote share divided by the sum of the Democratic and Republican vote shares.
The first model predicts the Democratic two-party vote share based on the final weighted running average. To take into account that the most recent polls available this year ended six days before the election, we excluded weighted running averages which were calculated after this time in past years. Then, the closest weighted running average was used as the final weighted running average. We used a simple Ordinary Least Squares model, trained on the latest four elections.
The second model predicts the Democratic two-party vote share based on the percent of the population that identified as white and the unemployment rate from the year prior. We used a logistic regression model, trained on the latest seven elections.
The last model uses a combination of poll results and demographic data. This model, which we use for our final election forecast, predicts the Democratic two-party vote share based on both the final weighted running average and the percent of the population that identified as white. We used a logistic regression model, trained on the latest four elections. If we had a greater sample size, we would have been able to use unemployment as an additional predictor; however, because there are only four usable observations, we had to limit the number of predictors to less than three.
For each of the models, we ran a simulation of the election with 10,000 repetitions, predicting the outcome of the 2021 Virginia Gubernatorial Election from the aforementioned variables. We used the estimate and the standard errors of the predictions for our simulations.
Results
The simulation results are shown below:
The polls-based model predicted that Youngkin would win the election with a two-party vote share of 52.4%, and McAuliffe would lose the election with only 47.6%. The simulation based on this model resulted in a Youngkin win 65.6% of the time and a McAuliffe win 34.4% of the time.
The demographics-based model predicted that McAuliffe would win the election with a two-party vote share of 52.6%, and Youngkin would lose the election with only 47.4%. The simulation based on this model resulted in a McAuliffe win 75.1% of the time and a Youngkin win 24.9% of the time. This is nearly the exact opposite as was predicted by the polls-based model.
When we look at our chosen model, which includes both polling and more fundamental variables, the story looks much closer to the poll-only model. The last model predicted that Youngkin would win the election with a two-party vote share of 52.1%, and McAuliffe would lose the election with only 47.9%. The simulation based on this model resulted in a Youngkin win 67.9% of the time and a McAuliffe win 32.1% of the time.
Overall, our chosen model and simulation indicate that Youngkin has a much better chance of winning than McAuliffe, though this is likely overstated. It is important to note that our models are severely limited by small sample sizes. The reliability of polls has also been called into question in recent years, particularly after the repeated bias in 2016 and 2020. From a qualitative lens, the race appears much closer than our simulations suggest. This might be a function of the race’s significance to national party organizations — Republican leadership appears eager to galvanize red voters by portraying the race as a nail-biter, while Democrats seek to maintain blue voters’ confidence in their ability to pull off a win. These intentions are visible in the media, wherein political agents from both sides of the aisle have spoken to the make-or-break nature of this election. That said, Virginia is a tried-and-trusted blue state with a known population of Democratic-leaning voters. The deciding factor here will be whether liberal activists can get them to the polls.
Correction: Nov. 3, 2021
An earlier version of this article incorrectly stated that our final model resulted in a Youngkin win 67.9% of the time and a Youngkin win 32.1% of the time. The second second statistic should have been attributed to McAuliffe, not Youngkin.
