Polling Miss in Wisconsin Democratic Primary Highlights Persistent Challenges in Predicting Primary Outcomes

In the August 11, 2026, Wisconsin Democratic primary for governor, Milwaukee County Executive David Crowley narrowly defeated State Rep. Francesca Hong, a democratic socialist, by roughly 0.5 percentage points (about 4,000 votes), despite Hong leading most polls by double digits for weeks.

Multiple reputable outlets confirmed the upset and the polling discrepancy. The New York Times reported Hong's loss after leading surveys, noting her campaign's struggles with electability concerns in the swing state.

PBS NewsHour and the Guardian similarly detailed Crowley's comeback win over the progressive frontrunner, attributing the outcome partly to late voter shifts and questions about Hong's viability against Republicans.

Polling data from Marquette Law School (July 22-27) showed Hong at 38-46% with high undecideds, while State Navigate's early August poll had her at 44% to Crowley's 22%. Nate Silver publicly questioned the models on X, noting primaries' inherent modeling difficulties compared to general elections.

Analysts cited in coverage, including references to a similar Michigan primary miss days earlier, pointed to potential oversampling of younger or progressive voters and failures to capture turnout or late-deciding moderates. Hong's past comments resurfacing also factored into narrative shifts.

This episode underscores broader polling limitations in open primaries with fluid fields, high undecided shares, and turnout variability—issues that can mask underlying voter preferences for electability and stability over ideological alignment. Such misses reveal how assumptions in survey design and modeling may not fully capture the pragmatic calculations shaping Democratic voter behavior in battleground states.