“Soon as they like you, make them unlike you.”
Background
It’s been five months since Abigail Spanberger’s landslide victory in the election for Virginia Governor. She was all but certain to win against a weak opponent; the only thing to quibble about ahead of Election Day was what her margin of victory would be. It turned out to be the largest Democratic victory in a Virginia gubernatorial election since Black Americans legally earned the right to vote in the 1960s. Coupled with sky-high turnout, no other candidate for Virginia Governor has ever received more votes than Abigail Spanberger.
Fast forward: The Virginia General Assembly session drew conservatives into a frenzy (mostly over bills that either never either made it to the Governor’s desk or haven’t been signed into law). The Governor arguably broke her campaign promise not to change Virginia’s congressional lines, leading to an election to decide whether the new congressional maps she supports pass. Democrats have provided technical answers to this, but voters don’t care about technicalities.
National Battle: The pro-Democratic “Yes” campaign has spent over $50 million in its bid to gerrymander Virginia, heavily outspending its various “No” opponents thanks to support from Democrats in Congress who want to nab themselves an extra easy 4 seats in the House of Representatives this November. The Yes campaign argues that a Democratic gerrymander will counteract Republican gerrymanders already enacted or likely to be enacted in Republican-controlled states, while the No campaign argues that it’s an unnecessary power grab. For our Voter Guide on this election, click here.
Too late to the party?: The No campaign has received large influxes of cash, including from Peter Thiel, over the last two to three weeks. However, some of those millions of dollars arrived arguably too late, as the deadline for placing TV ads had passed. The No campaign will instead spend that money on digital ads, mailers, and door-knocking operations.
About State Navigate & Our Polling Team
State Navigate is a nonpartisan, 501(c)(3) nonprofit organization dedicated to becoming a compass for state legislatures everywhere. Our bipartisan polling team consists of alumni of FiveThirtyEight, political scientists, and polling experts. Last year, State Navigate was the most accurate pollster in the country for the 2025 election cycle. To learn more about our poll sponsorship opportunities, click here! Public polling on state governance, especially state legislatures, is rare, so we try to make our polling as affordable as possible for sponsors to help fill this gap.
We’d also like to thank the donors who made tax-deductible donations to our polling fund during February through April to help us conduct this survey. Whether you donated $5 or $500, we appreciate your help!
Topline Results
Virginia Redistricting Amendment: “Yes” leads by 5 points in our poll. However, rounding the “Yes” and “No” numbers yields 51-45; we do not include decimals in our poll releases because it casts the illusion of false precision. In the weighted output, Yes accounts for 50.7% and No accounts for 45.4%.
Unsurprisingly, partisanship is a strong predictor of Yes/No voters. Democrats are almost entirely Yes voters: 92% of Democrats are voting/have voted Yes, while 5% of Democrats are in the “No” camp. Republicans are almost entirely No voters: 95% of Republicans are voting No, and 2% are in the “Yes” camp. What is different compared to last year’s Virginia gubernatorial election is Independents’ vote choice: 56% of true Independents (meaning they don’t lean toward either party) are voting No, and 32% are voting Yes (a 24-point lead). Democrats put up a remarkable performance with these voters in our late October poll: Spanberger led them by 24 points in the race for Governor, Ghazala Hashmi led them by 15 points in the race for Lieutenant Governor, and Jay Jones led them by 9 points in the race for Attorney General.
Governor Approval: Our findings come on the heels of the Washington Post/Schar School (GMU) poll that rocked the Commonwealth, showing Spanberger as the most historically unpopular Governor at this point in the most recent Virginia Governors’ terms. 47% of likely voters approve of Governor Spanberger, and 47% disapprove. There’s much more hardcore disapprovers than hardcore approvers, too: 41% of likely voters strongly disapprove of the Governor, while 31% strongly approve; 4% somewhat disapprove, and 11% somewhat approve.
This makes Abigail Spanberger one of the most unpopular Governors in the country. While we don’t know exactly how she ranks due to a lack of an aggregator of gubernatorial approval ratings in each state, she’s possibly the third-most unpopular Governor in the nation based on Morning Consult’s Q4 2025 gubernatorial approvals; it’s never good practice to only cite one pollster, but Morning Consult’s gubernatorial approvals are in the ballpark based on approval ratings from other in-state pollsters. If anything, their approvals may be too bullish on incumbent Governors, so Spanberger may only be in the top 10 least popular Governors instead of the top 3; we don’t know for certain.
Spanberger’s honeymoon came and went almost as quickly as Donald Trump’s second-term honeymoon; Trump’s approval went from positive to even by his first month in office; Spanberger’s first approval poll came a month into her term conducted by Roanoke College, though it’s worth noting they had the worst performance of any pollster in the 2025 Virginia elections among those that asked about the three statewide contests up that year. One big reason Spanberger’s approval is struggling is because of overwhelming opposition amongst Republican voters to her Governorship, but Republicans are known to “boo louder” in polling when they don’t like someone, or something, than Democrats, who tend to be more ambivalent about showing their disdain on a person or issue (unless it’s Donald Trump).
Six days after the WaPo/Schar poll was released, Brandon Jarvis reported that the Yes campaign stopped funding the ad featuring Spanberger in which she asked voters to vote Yes. Whether this is due to the Governor’s frustration that her approval rating could be sandbagged by the divisive issue of Democratic gerrymandering, or if the Yes campaign felt that her support did more harm than good, we don’t know.
Key Findings
In order to get a better sense of how well voters understood the issues at play in this election, we showed respondents maps of the current and proposed district lines and asked for their impressions. Without telling them what the maps represented, voters were about equally split on which map they preferred for the upcoming midterms: 26% preferred the map displaying the current district lines, while 25% preferred the proposed new lines (the remainder were either unsure (46%) or skipped the question (2%)). And voters appear to have some sense of what the proposed lines look like. Voters that preferred the map showing current district lines plan to vote “No” by a margin of 42 points, while voters that said they preferred the proposed district lines plan to vote “Yes” by a margin of 27 points.
When asked which of the two maps looked gerrymandered, 13% said that only the current congressional district lines looked gerrymandered, 28% said that only the proposed lines looked gerrymandered, 17% said both looked gerrymandered, and 3% said neither looked gerrymandered. There does appear to be some connection between how voters perceived the maps and their vote choice in the referendum. Among voters who said that both or neither map looked gerrymandered, significant majorities plan to vote “Yes” on the constitutional amendment: “Yes” led by 36 points among voters who said both looked gerrymandered and 72 points among those who said neither looked gerrymandered. But among voters who selected one or the other map as gerrymandered, majorities say they have or will vote “No.” Among those who said that only Map A was gerrymandered, “No” led by 11 points, and among those who said that only Map B was gerrymandered, “No” led by 22 points.
We then informed respondents about which map was which, and asked how that made them feel about their vote. Among those that had already voted, almost nobody (0.3%) said that the information made them regret their voting decision. Among those that had not yet voted, 62% of respondents said that the information impacted their decision, with 24% saying it made them more likely to vote “Yes” and 38% saying it made them more likely to vote “No.” But 92% of voters who said that the maps impacted their decision told us it made them more likely to vote in the way they had already told us they planned to vote, so the information probably didn’t have all that much impact.
Map of Respondents
You may or may not have had the privilege of witnessing State Navigate’s first-ever live map of unweighted/raw responses for our poll. If you did, you got somewhat of a sneak peek at this poll. Below is a simplified map of the raw responses, with the unrounded percentage totals for the referendum displayed. In the future, we aim to have a map of weighted responses. You can see a clear divide among Yes and No voters: No voters are overwhelmingly rural/suburban and Yes voters are mostly urban/suburban.
Team Notes & Analysis
Mary Radcliffe on Message Testing: We tested four different statements about the constitutional amendment to determine what might impact voters’ decisions (or how they felt about their vote in retrospect, if they’ve already voted). One of these, presenting the maps of the current and proposed district lines, was discussed above. The other three statements described the partisanship of the projected congressional makeup under each set of lines, context about other states having redistricted their congressional maps ahead of the 2026 midterm, and a description of what will be the youngest congressional district in the country under the proposed new district lines.
Among voters who have already voted, there appear to be no (or at least very few) regrets: only 4% of these voters said that any of the statements made them regret their vote. But among voters who said they had not yet voted, we see something very different: on all four statements tested, at least 62% of voters who had not yet voted said that the statement would have at least some impact on their vote.
So what’s going on here? It’s a classic case of voters using this kind of poll testing to express the opinion they already have, rather than acknowledge that the facts presented don’t really change their minds. For example, 45% of voters who haven’t voted yet said that at least one of the statements presented made them somewhat or much more likely to vote “Yes” on the referendum. But 87% of these voters had already told us that they planned to vote “Yes” or were leaning “Yes,” so it seems unlikely that the messages really impacted their choice. (We see the same thing on the other side; 50% of voters who haven’t yet voted said that at least one statement makes them more likely to vote “No.” 89% of these voters already plan to vote “No.”)
Chaz Nuttycombe on Party ID: Our poll this time has a slightly better partisan lean for Democrats: in the late October survey, it was 49% DEM – 41% GOP – 10% IND (+8 DEM). That was likely a point too generous to the Republicans, given the actual result of the 2025 election. Our methodology in this survey is identical to the late October one: the 2025 Q3 Gallup national quarterly party ID (which informed our partisan weighting decisions in that survey) broke down 48% DEM – 41% GOP – 11% IND (+7 DEM); however, the 2026 Q1 Gallup national quarterly party ID breaks down as 49% DEM – 39% GOP – 12% IND. Thus, this survey has a 51% DEM – 40% GOP – 9% IND (+11 DEM) partisan composition.
Chaz Nuttycombe on Spanberger Approvals: Given the makeup of this survey, it’s more likely than not that a new poll of registered voters or all adults would show Spanberger with a negative approval rating. Spanberger carried this survey’s electorate by ~14%: that’s similar to her 15-point margin of victory in the gubernatorial election, where the electorate voted for Kamala Harris by ~10 points. People who didn’t vote in the 2025 election are more ideologically conservative and moderate, and thus vehemently dislike the Governor.
One key group that Spanberger did well with in last year’s election, as she tried to become synonymous with the word “affordability” (which national Democrats have since used heavily in their midterm campaign messaging), is with lower-income voters: our mid-October survey showed that Spanberger greatly overperformed with the lowest income voters compared to Kamala Harris’s performance with these voters in Virginia. Governor Spanberger has lost substantial ground with these voters, and with the wealthiest voters as well: our mid-October survey showed Spanberger winning voters making <$30K/year by 27%; her net approval rating with them is +6 in our April survey (a 21-point difference). She was winning voters who make $30K-$50K/year by 2% in the mid-October survey: her net approval rating with them is -13 (a 15-point difference). She won voters who make $50K-$99K/year by 2%: her net approval with them is -3 (a 5-point difference). She won voters who make $100K-$149K by 21%; her net approval with them is +5 (a 16-point difference). She won voters who make >$150K by 19%: her net approval with them is +10 (a 9-point difference).
While the change in Spanberger’s approval ratings among voters by income is mixed, we can directly compare the recalled vote to the current vote intention to see whether income matters. The chart below shows that the lowest-income voters, where Sears did the best, are not shifting by as much. The No campaigns gain instead with middle- and upper-income voters, which is what keeps it competitive.
Chaz Nuttycombe & Nick Goedert on Trump Approval: National polls show Donald Trump’s approval rating since Election Day 2025 has continued to decline: the President’s net approval was -15 that day compared to -20/21 now. Given what you’ve just read on the partisan weighting differences between our late October survey and this one, you can imagine our surprise when we checked back to see that the President’s net approval rating is actually slightly better in this April survey. In the late October survey, the President’s disapproval rating sat at 57% and approval rating at 41% (-16 net approval). In this survey, his disapproval rating remains at 57%, while his approval rating has climbed to 42% (-15 net approval). Either the national polls are off, or we are, or maybe Virginia, being a Southern state, has a polarized electorate that gives the President a floor/ceiling.
Chaz Nuttycombe on Yes/No Gubernatorial Approval: 6% of “Yes” voters are Spanberger disapprovers, while 86% are Spanberger approvers. The more interesting number comes from the “No” campaign’s opinions on the Governor: 5% of “No” voters approve of Governor Spanberger, while 93% disapprove of her. While this is a small section of the electorate, and thus the “cross-crosstab” has a larger margin of error, it still shows that the voters who disapprove of the Governor but are voting Yes are disproportionately African-American. With a 20.7% +/- MoE, 40.8% of Yes/Disapprove of Spanberger voters are White/Caucasian, and 43.3% are Black/African American. Our overall electorate is weighted to 71% White/Caucasian and 16% Black/African American. Maybe this group of voters is indeed 43.3% Black, or it could be as low as 22.6%, or as high as 64%.
Chaz Nuttycombe on Voting Plans: Republicans are pretty uniformly opposed to this, but it seems like the Spanberger coalition is showing more cracks. However, almost all the Spanberger-to-No movement is not coming from voters who have actually shown up. This could mean that a lot of the defectors could potentially just sit out, or that they are holding their nose at the ballot box and switching to yes. Something like that effect could add another point or two to the Yes margin by the end.
43% of respondents say they already voted in this election. This matches roughly what we had in our late October survey for the Virginia gubernatorial election last year (41% say they already voted in that survey), which we fielded 9 to 7 days before Election Day; we fielded this survey 11 to 8 days before Election Day.
The biggest difference, however, is the percentage of voters who say that they plan to vote on Election Day. In our late October survey last year, 41% of voters said they planned to vote on Election Day. In actuality, the electorate was 57% Election Day, a 16-point difference. This is almost certainly due to a natural phenomenon of voters believing highly in their voting behavior in that they’ll be early birds, but end up deciding to vote on Election Day.
This time, 32% of voters say they plan to vote on Election Day on April 21st. If we see the same level of error as we did in our late October survey, that would mean that ~48% of the electorate will be Election Day voters, and ~52% will be early voting voters, which would be similar to 2024’s election vote composition when 53% of Virginians voted early, and 47% voted on Election Day.
Compared to our late October survey’s analysis of the vote plan in the Attorney General race, which is the closest 2025 race to the upcoming election by margin, our survey shows the No campaign is outpacing Jason Miyares’s early-vote performance thus far. However, it also shows that the final stretch of the early vote is expected to be more favorable to the Yes campaign than what was expected for Jay Jones. This may be due to a reduction in satellite voting access: by the time our late October survey entered the field, it was immediately after the first weekend that most satellite voting locations opened. This survey, in comparison, entered the field the day before most satellite voting locations opened.
Methodology, Toplines & Crosstab Links
This survey was fielded from April 10-13, 2026, among 707 registered likely special election voters in Virginia. Respondents were chosen from a voter file provided by i360. Text messages containing a link to an online survey designed by the State Navigate Polling Team and hosted on the SurveyMonkey platform were sent to respondents by AllianceForge. The survey was weighted by age, race, gender, education, race by education, region, and party identification, with benchmarks derived from the voter file, Catalist, AP VoteCast, Gallup Quarterly National Party ID, and data provided by the U.S. Census Bureau. The margin of error is 3.7% for a 95% confidence interval. The survey was designed and conducted by State Navigate. This survey was not sponsored by any party, candidate, or candidate’s committee.
Complete toplines can be found here, and selected crosstabs can be found here.
