Request a Review
MSR Decode · Elections · Polling data · Case 17

Fake Poll Fallout: How a Fabricated Mississippi Survey Still Lingers on Election Forecasting Sites a Week After the Exposure

Eight days after Wedgewood Polls admitted it was a fake polling firm, its Mississippi Senate poll still made up 87% of one site’s polling average, and an older VoteHub data feed still carried six of its Senate polls.

By MSR Decode · Published October 28, 2026 · 6 min read
Contents
  1. 01 · The Persistence of a Ghost Poll
  2. 02 · Campaigns and a PAC Use the Numbers
  3. 03 · Prediction Platforms and Media Fallout
  4. 04 · Who removed it, and the timeline
  5. What Remains Unknown
  6. Sources

Eight days after a firm called Wedgewood Polls admitted it was a fake polling firm, its fabricated poll of the Mississippi Senate race is still being counted on some election forecasting websites. Its numbers also appeared in a political fundraising email that is dated to the day after the admission. And a prediction platform’s news site is still citing another Wedgewood poll without a correction till present (Oct 28th, 2026).

While major news organizations and forecasters quickly removed the data, an analysis by MSR Decode found the fake poll by Wedgewood is still changing some polling averages just days before the Nov. 3 election.

At a glance

The fake poll a week after the exposure

Figures checked on Oct. 28, 2026

Polls vs. Odds87%of Polls vs. Odds’ Mississippi polling average came from the fake poll on Oct. 28.
Since the admission8 daysafter the firm’s admission, election sites are still counting the poll.
VoteHub6 pollsby Wedgewood are still in VoteHub’s older data feed for Senate races.

The Persistence of a Ghost Poll

Wedgewood released its Mississippi survey on Oct. 14, where it showed Republican Sen. Cindy Hyde-Smith leading Democratic challenger Scott Colom 52% to 48%, a close result in a state Donald Trump won by about 23 points in 2024.

On Oct. 20, the poll-analysis outlet FiftyPlusOne reported that the Mississippi data had been fabricated, and Wedgewood admitted that day that the firm was fake. Its statement said the vast majority of its releases had no real data behind them, according to Magnolia Tribune and the Michigan news service MIRS.

Major outlets moved quickly. By Oct. 21, The New York Times had removed 15 Wedgewood polls and Decision Desk HQ said it had removed 17, according to a syndicated news report. Nate Silver removed them from his forecast that day, Newsweek reported. However, some smaller election sites lagged behind:

  • Polls vs. Odds: On Oct. 28, the fabricated survey still makes up 87% of Mississippi’s polling average on the site. With the fake poll, the average showed Hyde-Smith ahead by 4.5 points. Without it, the average would have shown her ahead by 8 points.
  • Who Shows Up: The fake poll made up about 97% of the site’s Mississippi average on Oct. 23. After the site added a newer Change Research poll, Wedgewood’s share fell to 34%.
  • PredictionEdge: The site still listed the Wedgewood poll on Oct. 28 and counted it in a simple average of seven polls. There it moves Hyde-Smith’s lead only slightly, from 5.7 points without it to 5.4 points with it.
Evidence explained

Hyde-Smith’s lead with and without the fake poll

Mississippi Senate polling averages on Oct. 28, in points

Polls vs. Odds · with the fake poll4.52
Polls vs. Odds · without it8.00
Who Shows Up · with the fake poll4.37
Who Shows Up · without it4.19
Bars are drawn on a 0–10 point scale.
Evidence explained

The fake poll’s share of each site’s Mississippi average

Fake poll’s share of the average, Oct. 23 and Oct. 28, 2026

Election siteOct. 23Oct. 28
Polls vs. Odds86.9%86.9%
Who Shows Up96.5%34.0%
Source: MSR Decode’s recalculation of each site’s own average.

MSR Decode traced two of those sites, Polls vs. Odds and Who Shows Up, to an older VoteHub data feed, a list that other websites download automatically. VoteHub dropped Wedgewood’s polls from its own averages, but its older feed still carried six of the firm’s Senate polls on Oct. 28.

Campaigns and a PAC Use the Numbers

Political campaigns and a political action committee used the survey’s numbers, both before and after it was exposed.

  • The Colom campaign: On Oct. 16, Magnolia Tribune reported that Colom’s campaign cited the poll’s 4.3-point margin of error to call the race “tied.”
  • The Congressional Black Caucus PAC: According to an online archive entry dated Oct. 21, the day after Wedgewood’s admission, the PAC sent a fundraising email with a graphic showing the fake poll’s 52%-48% figures. The email’s subject line said Colom was “UNDERWATER!” and it asked for donations toward a $120,000 goal for Black voter turnout.

Earlier in the cycle, Democrat Abdul El-Sayed’s Senate campaign in Michigan listed a fabricated Wedgewood result in a July fundraising email, as the outlet Hanmi Ilbo reported.

Prediction Platforms and Media Fallout

Kalshi, a prediction platform whose odds Polls vs. Odds uses, also left uncorrected content online. An Oct. 13 Kalshi News article on the Michigan Senate race cited a Wedgewood survey showing El-Sayed ahead of Republican Mike Rogers, 53% to 47%. On Oct. 28, it still carried no correction.

After Wedgewood’s admission, Kalshi News published a response on Oct. 22 arguing that the fake polls barely moved Kalshi’s prices. An MSR Decode test of hourly prices on Kalshi and Polymarket broadly supports that claim. Across 19 Wedgewood releases, unusual price moves in the direction a poll pointed came up about as often as chance would produce: once on Kalshi and twice on Polymarket, never in the same race.

Who removed it, and the timeline

Evidence explained

Who removed the Wedgewood polls, and who still shows them

Status on Oct. 28, 2026

Site or publicationStatusWhat happened
The New York TimesRemoved or excluded itremoved 15 Wedgewood polls (by Oct. 21)
Decision Desk HQRemoved or excluded itremoved 17 Wedgewood polls (by Oct. 21)
Nate Silver’s forecastRemoved or excluded itdropped the polls (Oct. 21)
FiftyPlusOneRemoved or excluded itremoved them from its averages (Oct. 20)
WikipediaRemoved or excluded itdeleted the Mississippi poll (Oct. 20)
ScorvecRemoved or excluded itdropped the Mississippi poll (Oct. 20)
NewsweekRemoved or excluded itretracted its article about the poll
VoteHub’s own averagesRemoved or excluded itgive every Wedgewood poll zero weight
VoteHub’s older data feedStill showing it on Oct. 28still carries 6 Wedgewood Senate polls
Polls vs. OddsStill showing it on Oct. 28fake poll is 87% of its Mississippi average
Who Shows UpStill showing it on Oct. 28fake poll is 34% of its Mississippi average
PredictionEdgeStill showing it on Oct. 28fake poll is 1 of 7 polls in its average
Newsweek’s Oct. 15 newsletterStill showing it on Oct. 28online edition still shows the result
Kalshi News’ Oct. 13 articleStill showing it on Oct. 28still cites a Wedgewood Michigan poll
Evidence explained

Timeline

October 2026

  1. Oct. 13Kalshi News cites a Wedgewood poll of Michigan’s Senate race.
  2. Oct. 14Wedgewood releases its Mississippi poll: Hyde-Smith 52%, Colom 48%.
  3. Oct. 15Newsweek’s newsletter pairs the poll with Kalshi’s 90% odds for Hyde-Smith.
  4. Oct. 16Magnolia Tribune reports that Colom’s campaign is calling the race tied.
  5. Oct. 20FiftyPlusOne reports the Mississippi data was fabricated. Wedgewood admits it is a fake firm.
  6. By Oct. 21The New York Times and Decision Desk HQ say they removed Wedgewood’s polls. Nate Silver drops them.
  7. Oct. 21A PAC fundraising email carries the fake poll’s numbers, by the archive’s date.
  8. Oct. 22Kalshi says the fake polls barely moved its markets.
  9. Oct. 28Three election sites still count the poll, and VoteHub’s older feed still carries it.
Events are arranged in order; spacing is schematic.
Evidence tables, source files and corrections

Sources

The fake poll

FiftyPlusOne’s report (Oct. 20); MIRS report on Wedgewood’s statement (Oct. 20); Magnolia Tribune on the Colom campaign’s use of the poll (Oct. 16) and the admission (Oct. 20).

Newsweek: retracted article; Midterms Monitor, Oct. 15 (checked again Oct. 28).

Who removed it

Syndicated report on The New York Times and Decision Desk HQ (published Oct. 21); Newsweek on Nate Silver’s forecast; Wikipedia edit removing the poll (Oct. 20, 22:41 UTC); Scorvec’s change (Oct. 20).

Polling averages

Polls vs. Odds: the site, its published code and its current data file (updated 12:01 a.m. Eastern, Oct. 28). The fake poll’s share is 86.92% on Oct. 23 and Oct. 28; Hyde-Smith leads by 4.52 points with it and 8.00 without it.

Who Shows Up: its Mississippi page, published code and poll lists for Oct. 23 and Oct. 28. Oct. 23: share 96.50%; lead 4.86 points with the poll and 8.59 without. Oct. 28 (the site’s Oct. 27 forecast): share 33.97%; lead 4.37 with and 4.19 without. Its data lists VoteHub as the source of the Wedgewood poll and Wikipedia as the source of the Change Research poll.

PredictionEdge: Mississippi page (Oct. 28). Its average is the plain mean of the seven polls it lists: Hyde-Smith +5.43 with the Wedgewood poll and +5.67 without.

VoteHub: its older Senate feed, full Mississippi poll list, full Kansas poll list and notes for developers (checked Oct. 28).

The odds comparison

Polls vs. Odds’ forecast code and code for the disagreement list. Rebuilt from the site’s Oct. 28 data and matched to all 35 Senate races within 0.1 point. Colom’s polls-based chance is 5.3% with the fake poll and 2.2% without; Polymarket gives 13.5% and Kalshi 11.5%. The combined forecast gives Colom 8.9% with the poll and 7.3% without.

Fundraising

Congressional Black Caucus PAC email (archive date Oct. 21, 9:20 p.m.); El-Sayed campaign email of July 3; Hanmi Ilbo coverage.

Prediction platforms

Kalshi News: Oct. 13 Michigan article (unchanged on Oct. 28); Oct. 22 response. Polymarket: The Oracle (23 posts since June 20 checked) and its midterm market descriptions.

Prices: hourly history for each race’s Republican candidate from June 20 to Oct. 28, from Kalshi’s and Polymarket’s public data. Mississippi: Kalshi market; Polymarket market.

Ask MSR to investigate

Tell us which wallet, market, payout or profit claim you want checked. MSR accepts commissioned investigations using the evidence standards shown in this report.

Request a Review →

← Back to Case Work