← Back to the House
Forecast · r64-20260730-236b8cdd42f0 · as of 2026-07-30

The House · MI-04

Projected resultModel share · middle 80% of simulations · chance to win
DDiop Harris162,874 votes
47.5
RBill Huizengaincumbent seeking reelection172,703 votes
50.3
Other7,549 votes
2.2
remainder

Projected turnout: 343,126 votes · model estimate, not counted votes.

Displayed shares include the model's Other-vote allowance and therefore add to 100%. Interval endpoints are the model's 10th and 90th percentiles, not a polling margin of error.

SURE — · not yet rateable

How reliable is the evidence behind this race's number — not who wins. A safe seat with no polls can score low while its forecast is easy.

No public poll of this race has been fielded yet. SURE rates poll evidence, so it has nothing to rate here — this is an absence of evidence, not a judgement that the evidence is weak.

Limited evidence: zero polls. The fitted race-error distribution widens automatically; no manual score penalty or race override is applied.

Effective poll evidence0.00
Race-specific error SD4.0 points
Historically calibrated 80% half-range±5.7
Model treatmentprovisional widen-only chamber fallback
Fundamentals Environment Polls Money (FEC) Roster settled

SURE v3 — historically calibrated evidence-conditioned uncertainty · generated 2026-07-27. SURE rates the EVIDENCE behind a race's number, not the outcome: a safe seat with no polls can score low while its forecast is easy. Never a second win probability. It changes only idiosyncratic uncertainty, never the center. Senate marginal intervals are historically calibrated under the stated 0-120 day assumptions; House/Governor conditioning is provisional and widen-only until chamber-specific history passes the same gates.

MI-04 · Michigan
Rating band
Striped = seat changes partyDistrict locator

County map

County view
Projected margin
R+40R+20EvenD+20D+40
Striped = seat changes partyProjected county outcome

Counties colored by the projected margin — the 2024 result shifted by the state's projected swing, summing exactly to the forecast above. Hover a county for its numbers.

County by county

CountyMarginDiop HarrisBill HuizengaTotal
KalamazooD+21.865,90441,886110,215
OttawaR+20.333,82251,53687,278
AlleganR+24.420,90434,77656,933
Van BurenR+12.013,98517,91032,612
CalhounR+2.015,05415,68631,431
BerrienD+8.413,09411,02124,657

Projected — each county's share of THIS district only (split counties count just the votes inside the lines), summing to the forecast above.

State of the race

Call the Map currently has Huizenga ahead of Harris by R+2.9, with a 69 in 100 chance to win. The seat is currently held by a Republican. That result would keep the seat with Republicans. The current 2-poll average is D+1.3.

CTM forecast
Lean R hold · 69 in 100
Seat
Rep.-held · projected hold
Forecast range
R+10.6 to D+4.8
Polling
D+1.3 · 2 polls
Daily forecast

How the race is moving

69%R win probability today

Since the Jul 21 edition: margin 3.8 pts toward D · win probability +20 pts D

Estimated two-party vote shareLine = estimate · tint = 80% of simulations
D 48.5%● R 51.5%
Win probabilityProbability in each tracked daily edition
D 31%● R 69%

Tracked editions only. The pale segment marks the time remaining to Election Day; it is not fabricated future data. Hover, tap, or focus a chart and use the arrow keys for each edition's exact values.

How the margin was built

Published Democratic margin−2.929

6 contributions, summing to the forecast R+2.9. Every model adjustment is published here — including candidate-quality shifts. Read how the whole margin is built →

Comparable evidenceFloor 0.05 · precision discount 0.1

TWIN neighbours

TWIN means Typology-Weighted Inference from Neighbors, Call the Map’s comparable-unit matching engine for any kind of race. It borrows poll residuals only—never centers, terrain, or outcomes. Similarity is demographic and behavioural, not geographic, and each contribution decays with similarity distance. Borrowed evidence always has lower precision than a race’s own polling.

Borrowed term: ±0.000 points. This race has its own usable polling, so TWIN does not return borrowed evidence to it.

Signals at a glance

CTM Partisan Lean Index
R+5.9
Historical match quality
89/100 — How closely this race matches its eight historical analogues; not a win probability.

The Race Feed

1 source post
  1. Source post on XPolling source · MI-04
    Open on X ↗

    @jamesd0wns · Public Policy Polling · 2025-11-21

More coverage and field notes →

Recent polling

GBAO (D)2026-06-01n=500D+3×0.22
Ragnar Research Partners (R)2026-03-14n=400R+6×0.05

2 polls · weighted average D+1 · Full polling for this race →

Poll data via VoteHub, used under CC BY 4.0.

Money & Ads

Money in this race

Explore all spending →
Total supporting$0
Total opposing$368.2K
FEC transactions9

Who does this money help? These are independent expenditures — spending by groups that are not the campaigns and may not coordinate with them. Each dollar is reported either supporting one candidate or opposing another, and both can help the same person, so the table below nets them out per candidate. It lists everyone with reported activity in this cycle's FEC filings — which can include primary candidates who are no longer on the ballot.

CandidateSupporting themOpposing themNet for them
Bill Huizenga RPrimary field$368.2K−$368.2K
FEC committee activity

Direct PAC support

$204.5K · 71 transactions
  • Sean McCannNot on the ballot$204.2K
  • Diop Harris IINot on the ballot$250
FEC independent expenditures

Top outside spenders

Support and opposition reported separately

Official FEC independent expenditures. Cycle totals may include primary spending. Direct PAC totals include governed committee contributions and in-kind contributions. Campaign-video embeds are metadata-only editorial context from exact, reviewed channels; inclusion does not establish paid placement. Outside spending and campaign videos have zero forecast weight; direct PAC activity is reporting only.

Comparable past results

On the same ballot in Michigan