AI for WisconsinPublic record · operated by Party Party

Reviewed polling evidence · starting profile

Wisconsin data-center costs and benefits

Baseline proposition: The benefits of the data centers outweigh the costs

Evidence outcome: Oppose

The deterministic population-consensus-without-cross-poll-averaging.v1 method classified the reviewed evidence as oppose for the stated baseline proposition. It did not average unlike polls or establish a campaign position.

What this does not mean

This profile reports reviewed public-opinion evidence for the exact proposition above. It is not factual proof that benefits or costs prevail, a participant mandate, a policy recommendation, or an official campaign position. Recommendation: none. Campaign position: not established.

This exact starting profile is published so the public can inspect the admitted question, reported results, population, methodology, uncertainty, limitations, missing voices, source links, deterministic method, and hashes. It establishes neither a campaign recommendation nor a campaign position.

Adults age 18 or older who were currently registered to vote in Wisconsin or planned to register; reported toplines identify the result as the Wisconsin registered-voter sample.

There are several large data centers being proposed or under construction in Wisconsin. Which comes closer to your opinion, even if neither is exactly right:

The benefits of the data centers outweigh the costs
22%
The costs of the data centers outweigh the benefits
76%
Don't know
2%
Refused
0%

Source and fieldwork

Marquette Law School Poll registered-voter toplines, July 8-16, 2026. Marquette University Law School; sponsored by Marquette Law School. 838 respondents, fielded July 8, 2026 through July 16, 2026. Evidence is current.

Official toplines · Official methodology

Method boundary

The first option exactly states the baseline proposition, the second states its opposite, don't know expresses no direction, and refusal is excluded. This orientation classifies the reported opinion result only; it does not decide factual net effects or a campaign position.

The Marquette methodology statement reports 838 Wisconsin registered voters and an overall margin of error of plus or minus 4.5 percentage points. The data-center item is not listed among the random half-sample items. Sampling error is only one potential source of error.

Sampling, weighting, and limitations

Sampling: Full-probability hybrid sample: 621 respondents from a Wisconsin registration-based sample procured from L2 and 217 respondents from the probability-based SSRS Opinion Panel; 782 interviews were completed online and 56 by live-interviewer telephone in English.

Weighting: Design weighting, source-specific nonresponse adjustment, compositing, and calibration to Wisconsin registered-voter demographic benchmarks, the 2020 presidential result, and partisanship. The methodology reports an overall design effect and incorporates it into sampling error.

  • Published percentages are rounded to whole numbers; frequencies are also rounded, while percentages were calculated from non-integer weighted sample sizes.
  • The survey was administered in English only by web and live-interviewer telephone.
  • The hybrid sample, nonresponse, mode, question-order, measurement, and other unmeasured errors may affect the result beyond the reported sampling margin.
  • This item compares perceived aggregate benefits and costs. It does not measure specific policy remedies, technical impacts, host-community experience, statutory authority, or a candidate position.
  • The statewide registered-voter result must remain separate from affected-community evidence and any later participant-mediated result.

Missing voices

  • Tribal governments and members affected by energy, water, land, or infrastructure decisions
  • Residents, workers, utilities, local officials, and ratepayers in specific host communities
  • People who may benefit from or bear costs of employment, tax, grid, water, housing, or environmental changes

Conflicts

No conflict exists within the single admitted item. That does not establish broader agreement or replace missing evidence.

Reproducibility record

Method
population-consensus-without-cross-poll-averaging.v1
As of
2026-08-24T04:40:24Z
Profile version
01a03211-bff6-7e5c-bd04-0c0e95c8cd94
Profile digest
sha256:d7af8c66c5f9fe38ab3d6d3202e874cb1d488682f2355102efc3b26ba921941c
Issue version digest
sha256:f39ab861afef559d319c6fc6ad1498e63d6db06a3f9af29dc4aa41affab59b31
Published
2026-08-24T05:20:39.668413Z