Political poll results in New York, Michigan, and Wisconsin have come under scrutiny following discrepancies between survey data and election outcomes. Recent reports highlight issues ranging from sampling methods in legitimate research institutes to the admission of fabricated data by a firm known as Median Strategies.
The discussion follows a 2024 legal dispute where then President-elect Donald Trump sued the Des Moines Register and pollster J. Ann Selzer over a survey showing him trailing in Iowa by 3 points; Trump eventually won the state by 13 points. More recently, the Siena Research Institute published a poll showing New York City Mayor Zohran Mamdani with a 69 percent approval rating within the city, a figure that has drawn criticism regarding its methodology.
The Siena poll, which focused on the New York gubernatorial race, surveyed 811 likely voters but included only 227 respondents from New York City, representing approximately .003 percent of the city’s population. Steve Greenberg of the Siena Research Institute stated that the sample was tailored for a state-level race, leading to an overrepresentation of white voters (71 percent) and households with incomes over $100,000. In Michigan, polling for a Democratic Senate primary showed Abdul El-Sayed with a double-digit lead, but he ultimately won by only 1 percentage point over Haley Stevens.
The use of fabricated data presents a distinct challenge to the reliability of political information. Median Strategies, a firm whose website has since been removed, admitted to creating "completely bogus polls" as a "short-term social experiment." One such poll claimed Francesca Hong was leading a Wisconsin gubernatorial primary by over 20 points, though she ultimately lost to David Crowley. Another false poll placed incumbent Karen Bass ahead in the Los Angeles mayoral race by 12 points. These figures were shared by candidates on social media platforms like X, spreading unverified data to the general public before the firm's admission of fabrication.
The impact of these polling errors and fabrications sets a precedent for how campaigns and media outlets verify data before distribution. In Wisconsin, the false Hong poll was reported by The New York Times as having spread through the political ecosystem without independent verification. The long-term effects on public trust remain to be seen, as different groups have used these instances to question the motives and accuracy of the polling industry at large.
Moving forward, the accuracy of future polling will be tested in upcoming election cycles. While specific court dates or new legislative deadlines were not reported, the closure of Median Strategies and the retirement of J. Ann Selzer mark recent shifts in the industry. Voters and analysts are encouraged by experts to examine "cross-tabs"—the detailed breakdowns of who was surveyed—to assess whether a sample accurately reflects the demographics of the electorate. No specific effective dates for new polling regulations were mentioned in the source material.