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

TL;DR

Abdul El-Sayed has a 62.1% chance of winning. The race centers on roughly El-Sayed +1.5, not the three- to four-point lead suggested by a simple poll average, because Michigan Senate polling has recently leaned Democratic and El-Sayed runs behind the state’s gubernatorial Democrat. Mike Rogers wins about 38% of the time through a normal polling miss, stronger older-voter turnout, or defections among moderate Democrats.

Context

El-Sayed won the August 4 Democratic primary with 744,952 votes, against 731,160 for Haley Stevens and 61,516 for Mallory McMorrow. His margin over Stevens was 13,792 votes, or 0.90 percentage points of the Democratic vote. Rogers received 883,328 votes in an uncontested Republican primary (Michigan Department of State). (michigan.gov)

Eight pollsters measured the general election after the primary. None completed interviews later than August 28. That leaves the polling blind to El-Sayed’s August 29 apology over his response to the synagogue attack and the subsequent public pledge by more than 400 Michigan Jewish voters to support Rogers (Associated Press, Jewish Telegraphic Agency).

Evidence

The historical base rate favors Democrats, but less than the winning streak suggests. Michigan Democrats have won every regular Senate election since 2000. In this Class 2 seat, their margins fell from 22.72 points in 2002 and 28.82 in 2008 to 13.29 in 2014 and 1.68 in 2020. Rogers then lost Michigan’s other Senate seat to Elissa Slotkin by just 19,006 votes, or 0.34 points, in 2024 (official Michigan results). (michigan.gov)

Michigan’s final Senate polls have recently been too favorable to Democrats. The four-cycle history is mixed, however, and too small to justify mechanically subtracting several points.

The average Democratic overstatement was about two points. I shrink that to a 1.2-point expected correction because there are only four observations, 2014 missed in the opposite direction, and polling methods change. Two-month-out history is even less stable: the 2014 race moved sharply toward the Democrat, while the 2018, 2020 and 2024 races moved toward the Republican. I reflect that instability mainly by widening the forecast distribution rather than assuming another six-point Rogers surge. (realclearpolitics.com)

The post-primary polling is below. I use one electorate definition per poll, preferring likely voters when available.

The simple mean is D+3.25. My recency- and quality-weighted average is D+2.9 after down-weighting partisan sponsorship, Abacus’s small likely-voter subset and concerns about Susquehanna’s education mix. The broader RealClearPolling average was D+2.3 (RealClearPolling).

The national environment keeps El-Sayed ahead after the polling correction. Silver Bulletin’s September 2 generic-ballot average was D+6.6, almost unchanged from D+6.5 at the start of August (Silver Bulletin). (natesilver.net) Four clean paired Michigan polls also show the state leaning Democratic but El-Sayed underperforming Jocelyn Benson: Fox had Benson nine points ahead of El-Sayed’s margin, AARP four, MSU four and EPIC-MRA zero. The mean gap was 4.25 points, placing the Senate race around D+1.5 when the governor’s race is used as an in-state partisan benchmark (Fox, AARP, MSU, EPIC-MRA).

The crosstabs explain that penalty. AARP found Rogers leading voters aged 50 and over by 52%–43%; 86% of that group described themselves as extremely motivated, versus 71% of voters under 50. Fox found El-Sayed behind among all registered voters but ahead by four among definite voters, seven among the extremely motivated and eleven among the extremely interested. MSU found 37% saying a candidate being Muslim made them less likely to offer support; I treat this as a risk indicator, not as 37% of voters being unavailable to him (AARP, Fox, MSU).

Money favors Rogers. Through July 15, El-Sayed had raised $14.51 million and held $2.55 million, while Rogers had raised $10.90 million and held $4.47 million (FEC: El-Sayed, FEC: Rogers). Post-primary FEC expenditure notices through September 3 recorded about $10.27 million opposing El-Sayed or supporting Rogers, against $4.33 million supporting El-Sayed or opposing Rogers—a Republican-side advantage of roughly 2.4 to one (El-Sayed expenditure filings, Rogers expenditure filings). These notices include media-placement estimates, so they are not identical to advertisements already seen by voters.

My poll projection starts at D+2.9, subtracts 1.2 points for expected Michigan polling bias and 0.2 for still-unmeasured candidate and controversy risk. That yields D+1.5. The paired-race and national fundamentals independently land near the same margin. I use a five-point forecast standard deviation to cover correlated polling error, turnout-model failure and two months of campaign movement:

$$P(\text{El-Sayed win})=\Phi\left(\frac{1.5}{5.0}\right)=61.8\%.$$

Here, $\Phi$ is the standard normal cumulative distribution. The resulting 80% margin interval runs from about Rogers +5 to El-Sayed +8.

I checked markets only after fixing that estimate. Polymarket priced a Democratic victory at 64.5% on September 3, while PredictIt’s 59% Democratic and 42% Republican contracts normalize to 58.4% (Polymarket, PredictIt). Their disagreement brackets my independent estimate. Cook, Sabato and Inside Elections continued to call the race a toss-up rather than a clear Democratic race (Cook Political Report, Sabato’s Crystal Ball, Inside Elections). The market check raises the final number slightly from 61.8% to 62.1%.

What's non-obvious

The obvious pro-El-Sayed argument is that he leads seven of eight post-primary polls. That overstates the evidence. Several of the larger Democratic leads come from sponsored, small-subset or compositionally questionable surveys, while the strongest bipartisan polls range from Rogers +4 to El-Sayed +1. Michigan’s recent final Senate polls also overstated Democrats by roughly two to four points in 2018, 2020 and 2024.

The obvious pro-Rogers argument also goes too far. Fox’s R+4 registered-voter result reverses among its highest-turnout respondents, Democrats have the national enthusiasm advantage, and El-Sayed survived a primary advertising deficit of almost ten to one (Axios). Republican spending is a real threat, but El-Sayed has already shown that paid attacks do not translate mechanically into votes.

Uncertainties

• No reputable poll completed fieldwork after August 28. The effect of the synagogue apology, the Piker dispute and organized Jewish defections is therefore unmeasured (Associated Press, Jewish Telegraphic Agency).
• The primary polls missed El-Sayed’s margin over Stevens by roughly twelve points on average, but primary and general-election turnout errors are not interchangeable (Washington Post). (rollcall.com)
• Candidate financial reports are stale through July 15. Current cash balances and August fundraising will not be known until later filings (FEC).
• I found no directly matching Metaculus question. Early-vote data also had not begun providing a useful general-election turnout signal by the cutoff.