Call
the Map
FORECAST54 DAYS OUT
← Pollster performanceFilter the Full Poll Archive →
Pollster file

Ipsos

Historical accuracy is measured against certified results. Current-cycle records come from the canonical, deduplicated archive and do not by themselves change the sealed forecast.

Historical polls
106
Mean absolute error
4.21percentage points
Historical bias
D+2.08
Represented cycles
132002–2024
Current archive
138canonical records
History span
Oct 16, 2002through Nov 3, 2024
Identity

Known names and relationships

Known aliases
Reuters/Ipsos
Organization type
Single canonical organization
Cycles represented
2002 · 2003 · 2004 · 2006 · 2008 · 2010 · 2012 · 2014 · 2016 · 2018 · 2020 · 2022 · 2024

106 eligible historical polls across 13 represented cycles passed the published history identity checks. Historical bias is descriptive and is kept separate from sponsor or client affiliation.

Canonical 2025–2026 archive

Current-cycle records

Archive source date 2026-09-09 · bundled copy. Archive records may be newer than the sealed forecast; inclusion here does not mean the forecast uses them.

12 shown · 138 total

Disapprove +13

Approve 42% · Disapprove 55%

Field dates
Apr 18–Apr 25, 2025
Sample
1,992 RV
Sponsor / client
ABC News · The Washington Post
Partisan sponsorship
No partisan sponsorship flag
Disapprove +16

Approve 39% · Disapprove 55%

Field dates
Apr 18–Apr 25, 2025
Sample
2,464 A
Sponsor / client
ABC News · The Washington Post
Partisan sponsorship
No partisan sponsorship flag
Disapprove +10

Approve 43% · Disapprove 53%

Field dates
Mar 31–Apr 2, 2025
Sample
1,486 A
Sponsor / client
Not reported in archive
Partisan sponsorship
No partisan sponsorship flag
See this pollster in the Full Poll Archive →
Certified-result history

Eligible historical observations

6 shown · 106 total

CycleRaceTypeField endSamplePollResultError
20022002_House-G_USHouse-G2002-11-02505R+3.0R+5.0D+2.0
20022002_House-G_USHouse-G2002-10-31600R+2.0R+5.0D+3.0
20022002_House-G_USHouse-G2002-10-306000R+5.0D+5.0
20022002_House-G_USHouse-G2002-10-28600D+1.0R+5.0D+6.0
20022002_House-G_USHouse-G2002-10-26600D+2.0R+5.0D+7.0
20022002_House-G_USHouse-G2002-10-166000R+5.0D+5.0

Method · historical final-three-week D/R general-election rows plus verified historical supplement; min n=5

Definition · bias_D is poll D-R margin minus actual D-R margin; positive overstates Democrats.

A missing sample, sponsor or source is labeled unavailable; it is never reconstructed.

Poll data via VoteHub, used under CC BY 4.0. Poll data from The New York Times, used under CC BY 4.0.