Wisconsin Governor: Divergent Pollsters, Same Direction

Three polls, one race, and a reminder of why methodology matters as much as the topline number. In the Wisconsin governor race, Democrat David Crowley holds a lead over Republican Tom Tiffany in the two surveys built on independent samples. Marquette University put Crowley ahead 51 to 44 among 668 likely voters, a seven-point edge. A separate release from Wick, a Republican-aligned pollster, surveyed a larger pool of 1,213 likely voters and found a tighter contest, Crowley at 48 and Tiffany at 45. Different house effects, different sample sizes, same direction: Crowley ahead, margin somewhere between three and seven points depending on the firm.

The outlier is the third entry. A survey released by the League of American Workers shows Crowley collapsing to 19.5 percent against Tiffany's 44.6 — a result that would represent a wholesale reshaping of the electorate if taken at face value. It does not stand on its own, however. That poll draws from the identical 1,213-respondent sample used in the Wick release, meaning the two are not independent confirmations of anything; they are two interpretations, or two weighting schemes, applied to one underlying pool of respondents. When a shared sample produces a 29-point swing in the margin between releases, the discrepancy says more about modeling choices — likely-voter screens, turnout weighting, undecided allocation — than about any actual movement in voter sentiment. Readers should treat the League of American Workers number with real caution until there is clarity on how it was derived from the same raw data that produced a near-tied Wick result.

The practical takeaway is that the credible range in this race currently runs from a competitive mid-single-digit Crowley advantage to a more comfortable one, not to a Tiffany-favorable environment implied by the outlier figure. Marquette's academic polling has a long track record in Wisconsin statewide races, which gives its seven-point Crowley margin added weight relative to partisan-sponsored work. The Wick survey, despite its Republican affiliation, lands closer to Marquette than to the League of American Workers number — itself a signal that the shared-sample release is the one out of step, not the other two.

With five days until Election Day, Wisconsin voters and the campaigns alike are left parsing not just who is ahead, but which pollsters' assumptions to trust. The data converges on a Crowley lead; the size of it remains the open question.