Will Abdul El-Sayed win the 2026 US Senate election in Michigan?

TL;DR

Abdul El-Sayed has a 72% chance of winning Michigan’s 2026 Senate election. My weighted average of the post-primary polls puts him ahead by about 2.7 points, including a new NYT/Siena result showing him up five and a co/efficient poll showing a tie. The race remains competitive because El-Sayed runs well behind other Michigan Democrats and a normal statewide polling error could erase his lead.

Context

El-Sayed won the August 4 Democratic primary by only 13,792 votes, defeating Haley Stevens 48.45% to 47.55% in a 1.54-million-vote contest (Michigan official results). He now faces Republican Mike Rogers, who lost Michigan’s 2024 Senate election to Elissa Slotkin by 19,006 votes, or 0.34 percentage points (Michigan official results).

The newest poll at the cutoff was NYT/Siena’s 49%–44% El-Sayed result. Siena’s release said its director was available to discuss the poll from 7:30 a.m. Eastern on September 24, or 11:30 UTC, so I include it before the client’s 12:14 UTC cutoff, while giving it no extra leverage because the accessible page did not expose its full sample details (Siena Research Institute).

Evidence

Michigan is a swing state, but Democrats have repeatedly found ways to win its Senate elections. These are the most relevant recent statewide comparisons:

The 2024 comparison is the cleanest analogue. Rogers lost even though Trump carried Michigan. Rogers ran about 1.75 margin points behind Trump in the same electorate. The 2026 environment is more Democratic, but El-Sayed is also a weaker fit for the median Michigan voter than Slotkin.

I used the following 15 non-internal post-primary ballot tests. Margins are El-Sayed minus Rogers.

I weighted these by poll quality, the square root of sample size and recency, using a 21-day half-life. Partisan and low-transparency polls received less weight. Siena received a normal high-quality-poll weight rather than an assumed large-sample bonus. The result is El-Sayed +2.7 points. Reasonable changes to the half-life and quality weights put the center between roughly +2.4 and +3.1.

The national environment supports that lead. Silver Bulletin’s September 23 likely-voter trend put the generic congressional ballot at Democratic 49.8%, Republican 41.6%, or D+8.1 (Silver Bulletin). Emerson found Trump at 40% approval and 55% disapproval among Michigan likely voters (Emerson). Inside Elections’ structural Michigan baseline is D+4.5, although it rates this specific race a toss-up because of candidate effects (Inside Elections).

El-Sayed’s candidate penalty is real. In the same respondents, Jocelyn Benson led the governor’s race by 12 points while El-Sayed led by three in the Washington Post poll; Benson led by 6.7 while El-Sayed led by 2.5 in Emerson; and Benson led by three while the Senate race was tied in co/efficient. The Post also found that 53% wanted Democratic Senate control, but only 48% backed El-Sayed; 48% considered him too liberal; and 83% of Black likely voters preferred Democratic Senate control while only 67% supported him (Washington Post). El-Sayed also lost Wayne County’s Democratic primary 51.34% to 45.49%, confirming a coalition problem in the state’s largest Democratic county (Michigan official results).

Rogers has not converted those vulnerabilities into a polling lead. Emerson found El-Sayed at 49% favorable and 44% unfavorable, versus Rogers at 42% favorable and 46% unfavorable (Emerson). Democratic leaders have also consolidated behind El-Sayed, with Kamala Harris campaigning in Detroit to help repair his support among Black and establishment Democrats (Associated Press).

Both sides have enough money to compete. Candidate filings available at the cutoff cover only through July 15: El-Sayed reported $14.51 million in cycle receipts and $2.55 million cash, while Rogers reported $10.83 million in receipts and $4.41 million cash (El-Sayed FEC; Rogers FEC). AdImpact estimated that candidates and allied groups had spent almost $125 million since the primary, including an $18.5 million Michigan commitment from a Trump-linked group (Associated Press). Senate Majority PAC’s additional $10 million brought its reported Michigan commitment to $30 million (Axios). Money signals a serious race, not a hidden Rogers advantage.

Michigan polling error has no stable sign. The final RCP averages overstated the Democratic Senate margin by about 1.8 points in 2018, 3.7 points in 2020 and 2.0 points in 2024. But the final RCP average understated Whitmer’s 2022 gubernatorial margin by roughly 9.5 points (RealClearPolitics). I treat this as a reason for wide uncertainty, not a mechanical Republican correction.

My fundamentals estimate is El-Sayed +3.5 after reducing a much stronger generic-D projection for his observed candidate penalty. Combining 75% polling at +2.7 and 25% fundamentals at +3.5 produces an expected margin of about El-Sayed +2.92. A 4.9-point forecast standard deviation, covering correlated polling error, turnout error and the remaining campaign, converts that to 72.4%.

Only after fixing that estimate did I check markets. Polymarket priced a Democratic Michigan win at 71.5% with about $577,000 in volume (Polymarket), while PredictIt’s Democratic contract had a 70-cent bid and 71-cent ask (PredictIt). Those prices are close enough that I made no market-driven adjustment. The expert ratings tell the same broad story: Cook and Inside Elections rate the race a toss-up, while Sabato moved it to Lean Democratic on September 22 (Cook; Inside Elections; Sabato’s Crystal Ball).

What's non-obvious

The large Benson–El-Sayed polling gap should cap enthusiasm, but it should not be subtracted from the Senate polls a second time. The direct ballot tests already include voters’ concerns about El-Sayed’s ideology, religion, Israel positions and coalition. The gap is why the fundamentals prior is only D+3.5 rather than something near D+7; using it again to shift a D+2.7 polling average toward a tie double-counts the same evidence.

The cutoff also matters. The NYT/Siena D+5 release was apparently public shortly before the client’s 12:14 UTC cutoff, while the nearly simultaneous co/efficient poll showed a tie (Siena; co/efficient). Adding both moves the weighted polling center only about three-tenths of a point. The race did not suddenly break open, but forecasts omitting Siena are slightly stale.

Uncertainties

The main uncertainty is correlated polling and turnout error, not sampling error. El-Sayed relies more on young, Black and college-educated voters, while Rogers is stronger with older and non-college voters (Emerson crosstabs). A different likely-electorate model can move the race several points. My approximate 90% interval for the final margin runs from Rogers +5 to El-Sayed +11.

The Siena landing page did not expose an exact posting timestamp or full Michigan sample details, so I included it without allowing it to dominate. Candidate finance reports are more than two months stale. There is also no useful early-vote result signal yet: absentee ballots become available 40 days before the election, and Michigan’s statewide in-person early-voting period does not begin until October 24 (Michigan Department of State). The most valuable missing evidence is another round of high-quality likely-voter polling after the Harris visit and current advertising surge, followed by October debate polling and validated absentee-return benchmarks.