In a result that confounded pollsters and party strategists alike, Democratic state representative Francesca Hong — who was reported by multiple surveys to be leading the Wisconsin gubernatorial primary by double digits — lost the nomination by roughly half a percentage point to Milwaukee County Executive David Crowley.
How a wide lead disappeared
Polls in the weeks before the Aug. primary showed Hong ahead by margins ranging from 18 to 36 percentage points, according to aggregated reports. On election night she fell short, losing by 0.5 percentage points. The mismatch between pre-election polling and the final tally has reopened debate over several long-standing issues in public opinion research: sample representativeness, likely-voter models, late-deciding voters and turnout forecasting.
Political scientists and pollsters have spent years adjusting methods to reflect changing communications habits and diverse electorates. But when a large, sustained lead like Hong’s fails to hold, it highlights both technical and political questions about the extent to which surveys capture who actually shows up to vote.
What may explain the gap
The specific reasons behind the Wisconsin outcome will be the subject of after-action reviews by campaign teams and pollsters. Broadly, analysts point to a few recurring factors that can produce the appearance of a large lead that narrows or vanishes:
- Likely-voter models: Polls weight respondents by who they think will vote. If the model overestimates turnout among enthusiastic supporters of a candidate, margins can be overstated.
- Late movement: Voters can break late for a different candidate after the last polls are taken, especially in low-information or crowded primaries.
- Nonresponse and sampling bias: Certain demographic groups may be underrepresented in samples despite weighting, producing skewed results.
- Turnout composition: Even if overall turnout is close to expectations, shifts in which groups participate — by age, race, geography or ideology — can change outcomes dramatically.
Implications for Washington state
Though the contest was in Wisconsin, the lessons matter for Washington voters, campaigns and local media. Washington’s politically diverse electorate and expanding use of mail ballots and vote centers mean pollsters and campaigns must carefully account for how different voter groups engage with voting methods and with late campaign developments.
Washington elections officials and campaigns frequently point to Washington’s vote-by-mail system as promoting higher participation, but the composition of who returns ballots early versus who waits to vote in person or at drop boxes can create similar turnout dynamics to those that surprised pollsters in other states.
Practical takeaways for voters and stakeholders
For voters, the Wisconsin result is a reminder that pre-election poll coverage is an imperfect snapshot — not a prediction etched in stone. For journalists and analysts, it suggests caution when interpreting single polls or repeating margins without context about methodology and likely-voter assumptions.
| Metric | Reported range or value |
|---|---|
| Pre-election lead reported for Hong | 18–36 points |
| Final margin | 0.5 point loss |
Campaigns should also prepare for volatility. Organizers who rely on robust early polling signals to allocate resources may find those signals misleading if underlying turnout assumptions are wrong. That calls for continuous voter contact, real-time turnout tracking and contingency planning for late shifts.
Questions for pollsters and party leaders
Immediate questions for pollsters include how their likely-voter screens are constructed and whether weighting adjustments adequately compensate for nonresponse among crucial subgroups. Party leaders will likely examine whether messaging or coalition-building efforts failed to convert broad poll support into votes.
For voters and officials in Washington, the Wisconsin case is also a prompt to look at local polling and turnout data with fresh skepticism and to demand transparency from firms conducting public-facing surveys.
The broader lesson is not that polls are useless — they remain valuable tools for understanding trends and testing messages — but that large reported leads can be fragile when they rest on assumptions that do not match who actually turns out. As Washington prepares for its next rounds of primaries and general elections, campaigns, journalists and voters should treat polling margins as informative, not definitive, and pay close attention to turnout indicators in the closing days of a race.
Grace Yun is the state correspondent at WE NEWS.
