Public opinion polls in several high-profile primary races have shown significant deviations from final election results this season. In Michigan's Democratic Senate primary, Abdul El-Sayed held double-digit polling leads but defeated Haley Stevens by only one percentage point. Similar discrepancies occurred in Wisconsin, where David Crowley won the Democratic gubernatorial race despite Francesca Hong leading in earlier surveys.
Primary polling is described by experts as more difficult than national general election surveys due to lower funding and less rigorous methodology. Courtney Kennedy, senior vice president of U.S. affairs at Ipsos, stated that the margin of error in primary polls is typically three times higher than that of national polls. Kennedy noted that many state-level primary polls rely on "text-to-Web" surveys, which are less expensive but often fail to reach a representative sample of the electorate.
In Michigan, the text-to-Web method effectively reached younger, more progressive voters but failed to accurately capture support for moderate candidates like Stevens. Kennedy pointed to one higher-quality poll using phone interviews that more accurately reflected Stevens' actual support. In Wisconsin, the race was complicated by three candidates dropping out and one re-entering, leading to polling that showed 30 to 40 percent of voters remained undecided just weeks before the election.
The discrepancy in Michigan, where a double-digit lead translated into a one-point victory, highlights how specific groups—such as moderates or older voters—may be underrepresented in modern polling datasets. Kennedy explained that lower-cost methods like text surveys often miss these demographics, which can lead to results that do not reflect the actual voting population. For the average voter, this means that polling numbers seen in news reports or social media may not accurately predict the outcome they will see at the ballot box.
The issue is further complicated by the emergence of intentionally false data. An organization calling itself Median Strategies recently admitted that its Wisconsin polling data was faked as a "social experiment" to see how polling spreads. To verify the legitimacy of future data, experts recommend that voters check for a pollster's track record, look for a named person with credentials attached to the report, and ensure the organization provides transparent details regarding its sampling and weighting methods.
What happens next: As the primary season winds down, election observers are focused on how polling organizations will adjust their methodologies for future cycles. There are no reported dates for new industry-wide standards, but researchers continue to evaluate the effectiveness of phone interviewing versus digital text surveys. Voters and news organizations are advised to vet polling sources for transparency and credentials before treating survey results as definitive indicators of election outcomes.