To provide news coverage about the 2025 elections, ABC News, The Associated Press, CBS News, CNN, FOX News Media and NBC News pooled their resources into The Voter Poll by SSRS. Verasight was proud to contribute more than 7,000 survey responses from registered voters in California, New Jersey, Virginia, and New York City to The Voter Poll. While the major news networks rely on The Voter Poll for in-depth analysis of voters and the key issues on their minds, we at Verasight are also interested in how our data stands on its own. To answer this question, we assessed how accurately Verasight’s data predicted the statewide elections in Virginia and New Jersey, the mayoral race in New York City, and the Proposition 50 ballot initiative in California.
To evaluate Verasight’s accuracy, we compare all Verasight data delivered to the major networks as part of The Voter Poll to the actual election outcomes. Following standard survey practice, we weight the Verasight data in each area to match the known population of registered voters on age, gender, education, race, language (English/Spanish), region of the state/NYC, and 2024 presidential turnout/vote choice. (The design effects ranged from 1.12 to 1.33, which means the unweighted data were extremely close to these population targets.)
To compare the Verasight data and the actual election data, we follow the 2024 AAPOR Task Force on Pre-Election Polling and FiveThirtyEight and compute the “margin of victory,” calculated as the difference in vote share received by the top two candidates (or difference between the percent of yes and no votes for Proposition 50). For example, Winsome Earle-Sears, the Republican Candidate in the Virginia Governor’s race, received 42.22 percent of the vote. Abigail Spanberger, the Democratic Candidate, won with 57.58 percent of the vote, producing a margin of victory of 15.4 percentage points. Our goal is to see how closely the margin of victory based on reported votes in Verasight’s contribution to The Voter Poll by SSRS matches the actual election results. The results presented will not match The Voter Poll estimates released by the media outlets since our analysis is based exclusively on the Verasight portion of The Voter Poll sample.
The figure above shows the election results predicted by Verasight’s data in The Voter Poll side-by-side with each of the actual election outcomes. The central finding is that Verasight’s estimates closely matched the actual results across every contest polled. For example, based on voters in Verasight’s data, the expected margin of victory in the VA gubernatorial election for Democratic candidate Spanberger was 15.6%, just 0.2 percentage points above the actual margin of 15.4%. The average absolute difference between Verasight data and actual election results was only 1.5%, and across all the six races polled, the maximum difference was only 2.6%. Moreover, the margin of victory in every election was within the estimated 95% margin of error for Verasight’s results.
We view these results as a strong validation of Verasight’s multi-method recruitment and commitment to data quality. We anticipate presenting research that elaborates on these findings and additional analyses at the 2026 Conference of the American Association for Public Opinion Research in Los Angeles.
.png)
