Will Abdul El-Sayed win the 2026 US Senate election in Michigan?
Latest forecast
Generated Nov 3, 2026, 3:24 AM
Write-up
Executive Summary
Abdul El-Sayed has an 83.8% probability of ultimately being elected to Michigan’s U.S. Senate seat. He is the clear favorite over Republican Mike Rogers, but the upset risk remains meaningful.
The strongest evidence is El-Sayed’s lead across numerous late polls. The main risks are a shared polling error, an unfavorable turnout mix, and Democratic defections following Haley Stevens’s November 1 endorsement of Rogers.
This is a pre-election assessment dated November 3, 2026, before Michigan’s Election Day voting began. The question resolves according to the final lawful election outcome—not an election-night media projection.
Routine update
El-Sayed’s chance of winning rose from about 79% to 84%, supported by a broader polling assessment and stronger limits on late-endorsement damage. He remains the clear favorite, while Rogers’s upset chance falls from roughly one in five to one in six.
Key drivers
A broad polling lead, with important disagreement about its size
RealClearPolling’s final average showed El-Sayed ahead by 5.3 percentage points. The advantage was not confined to one unusually favorable survey: polls from YouGov, CNN/SSRS, Mitchell, Research Co., Rasmussen and others also placed him ahead.
Nevertheless, the spread between surveys restrains confidence. Trafalgar’s final survey showed a 1.2-point lead, and election-eve Zogby figures showed a 2.8-point lead among decided respondents. NYT/Siena showed a much larger eleven-point advantage. These are published polling results, not alternative forecasts. Their disagreement supports aggregation and a substantial allowance for error rather than selecting a single headline. (Polling record; Zogby tables)
Michigan’s recent history argues against complacency
Democrats won the recent Michigan Senate elections examined, but the contests became much closer: the certified Democratic victory margin was 1.68 points in 2020 and 0.34 points in 2024, when Rogers was also the Republican nominee.
Recent Senate polling tended to overstate Democratic margins. The longer historical comparison, however, also included a large error in the opposite direction. The forecast therefore makes a cautious adjustment toward Rogers without assuming Michigan has a permanent, precisely measurable polling bias. (Official results archive)
Stevens’s endorsement is a genuine but poorly measured late risk
Stevens’s November 1 endorsement could give some Democrats permission to support Rogers or leave the Senate contest blank. Most published surveys did not cleanly measure its aftermath.
Two factors limit the case for assuming a decisive collapse. First, Michigan had already recorded more than 2.3 million early in-person and returned absentee ballots. Second, the endorsement prompted reported Democratic fundraising and volunteer activity. Those mobilization reports are imperfect indicators—not measured changes in votes—but suggest the reaction was not uniformly favorable to Rogers. (AP reporting; Michigan Department of State)
The political environment helps El-Sayed, but does not eliminate candidate-specific weakness
Marist found Trump unpopular in Michigan and the Democratic gubernatorial candidate running ahead of El-Sayed. This supports a favorable Democratic environment while showing that voters distinguish between the candidates on the ballot. The forecast treats those conditions as context already reflected in Senate polling, not an additional independent bonus. (Marist)
Scenarios
| Route to the outcome | Why it carries weight | How it resolves |
|---|---|---|
| El-Sayed’s polling advantage survives | This is the dominant scenario because his lead spans many surveys and methods. Ordinary polling error and a limited net endorsement effect leave that advantage intact. Ballots already cast constrain how much last-minute movement can change the result. | El-Sayed wins and the final lawful result produces YES. |
| Rogers wins through combined polling and turnout error | This is the main alternative. Several late polls show a close race, Rogers nearly won statewide in 2024, and recent Michigan Senate polls have overstated Democrats. A shared polling miss, weaker Black or younger-voter participation, or further Democratic defections could erase El-Sayed’s advantage. | Rogers wins and the final lawful result produces NO. |
A very close count could lead to recounts or litigation under either route. That would affect the timing of resolution, not its governing rule. A minor-candidate victory was treated as negligible given the observed polling.
Implications
- Use Democratic retention of Peters’s seat as the planning baseline, not a certainty. Decisions dependent on the winner should retain a credible Rogers contingency.
- Do not confuse a likely winner with a likely landslide. The evidence supports El-Sayed more consistently than it supports any particular victory margin.
- Separate electoral confidence from procedural finality. A projection or certification delay does not itself resolve the question. Certification—and any qualifying final legal determination before the winner takes office—controls.
Signs to watch
| Timing or trigger | Development and direction |
|---|---|
| November 3: Election Day turnout | Strong participation among Black and younger voters would make YES more likely; weakness in those groups, or a turnout mix more favorable to Rogers’s older-voter coalition, would make it less likely. Early participation totals alone cannot establish candidate support. Reliable evidence of the final turnout mix would settle a major uncertainty and make the forecast worth rerunning. |
| New evidence about the endorsement’s aftermath | Credible evidence of broader Democratic defections or Senate-ballot abstention would make YES less likely. Demonstrated counter-mobilization would move it in the opposite direction. Additional crossover endorsements could arrive without warning, but fundraising or volunteer claims alone would not establish a vote shift. A verified broad change would warrant rerunning the forecast. |
| November 3 and subsequent counting | Broadly representative returns showing El-Sayed’s polling advantage holding would make YES more likely; consistent Rogers overperformance would make it less likely. Once returns meaningfully replace assumptions about turnout and polling error, the forecast is worth rerunning. |
| Certification and any final legal proceedings after Election Day | A final lawful outcome electing El-Sayed settles YES; one electing another candidate settles NO. Delay alone provides no directional evidence. A verified, material change in the vote count or a binding ruling would warrant immediate reassessment. |
Research and methodology
The forecast was built from election evidence rather than by averaging published win probabilities.
An original weighted calculation assembled 18 recent survey observations. It checked fieldwork dates, sample definitions and candidate percentages against available releases. Polls were placed on a common two-party basis—comparing El-Sayed and Rogers after excluding other responses from that denominator—while unresolved preferences remained part of the outcome uncertainty.
More recent surveys received greater weight. Sample-size influence was capped, and politically aligned surveys or observations with dependence or documentation concerns received reduced weight. The calculation then applied a modest historical-error adjustment and a separate judgmental allowance for late disruption. Its uncertainty allowance covered polls missing in the same direction, turnout mistakes and late movement; it did not treat all interviews as one enormous survey with negligible error.
The sources with the greatest bearing were:
- Original polling releases and tables: NYT/Siena, YouGov, Zogby, Trafalgar and Research Co. established both the broad lead and the disagreement about its size.
- Michigan’s official results archive and historical polling: Certified results were compared with final polling averages for Senate elections from 2012 through 2024. Newer elections received greater emphasis, while the historical correction was reduced toward zero to avoid treating a small sample as a permanent bias.
- AP reporting and official voting statistics: These informed the assessment of Stevens’s endorsement, possible backlash and the constraint imposed by completed voting.
- External assessments: Cook’s November 2 move to Lean Democratic corroborated the polling direction. Polymarket and Kalshi’s election-eve commentary favored El-Sayed but were more cautious. They served as calibration checks, not independent polling observations.
Sensitivity checks—removing Siena, applying a more adverse historical correction, or increasing the endorsement penalty—preserved El-Sayed’s favored status while reducing confidence.
Limitations
The largest missing evidence was a reliable measurement of movement after Stevens’s endorsement, separated from differences between pollsters. Final turnout among the candidates’ key constituencies was also unknown; early-ballot counts could not resolve it.
Some survey documentation remained incomplete. Competing MSU/Verasight specifications could not be fully audited, so that observation received reduced weight. A reported final Emerson release was excluded because its topline and sample could not be sufficiently verified.
Finally, the historical sample is small, and the error allowance and late-event adjustment are modeling judgments rather than precisely fitted effects. The reported decimal should not be read as comparable precision in knowledge of the true probability.
Forecast history
Hover a data point to trace its series, or click to view the forecast generated at that time.
Signed forecast receipt
Signed Nov 3, 2026, 3:24 AM with ed25519 key preseen-prod-ed25519-20260523 and externally timestamped Nov 3, 2026, 3:24 AM.
sha256:66fa40c9985e5b...0e5e6dcffcWant forecasts like this?
Preseen researches your questions and answers them with calibrated, source-backed forecasts.