Osage Research: polls, partners and record

§1Whom it works for

Why this matters: A firm that also works for one party's campaigns can still publish accurate polls, but readers deserve to know.

Works for Republican candidates or groups, by its federal filings. From 2023 through 2026, 6 federal committees reporting Republican affiliation paid it, and none reporting Democratic^1,2,3,4.

Federal campaigns and committees that paid it

From 2023 through 2026, 6 federal committees reported paying Osage Research $385,790 for polling, surveys or research^1,2,3,4. By the party each committee reports to the FEC: 0 Democratic, 6 Republican, and 0 none (most PACs and super PACs report none).

The largest payers, by amount paid for polling, surveys or research. Only federal committees report to the FEC; state and local campaigns, work paid through another consultant, and polling billed as consulting are not here. Matched by the payee name on the filing: OSAGE RESEARCH LLC.
Committee	Kind	Party it reports	Paid for polling	Between
SFA FUND, INC	PAC	Republican	$157,100	July 24, 2023 – January 24, 2024
FRIENDS OF DAVE MCCORMICK	Senate campaign	Republican	$156,947	July 25, 2024 – January 30, 2025
JAMES HAYES FOR CONGRESS	House campaign	Republican	$22,750	July 9, 2024 – October 3, 2024
MOONEY FOR SENATE, INC.	Senate campaign	Republican	$17,060	April 10, 2023 – April 25, 2024
AL FOR MONTANA	House campaign	Republican	$17,023	May 1, 2026
BYRON DONALDS FOR CONGRESS	House campaign	Republican	$14,910	August 27, 2024

§2Record against results

Why this matters: The one test that doesn't depend on other polls: how close a firm's final polls came to what voters did.

From FiveThirtyEight's final public data^5,6: general-election polls between a Democrat and a Republican in the last three weeks before each election, compared with the certified result. FiveThirtyEight rated each name polls were published under separately. Across all 8,957 such polls, the average miss was D+0.7, so the relative figure takes out the misses every pollster shared that year.

Error is the poll's D−R margin minus the result's; D+ means it showed Democrats doing better than they did.
Published as	Polls	Elections	Average miss	Relative to other polls	Typical size of miss	538's last rating
Osage Research	0	—	—	—	—	1.0 of 3 · 1% partisan work

ABC News closed FiveThirtyEight in March 2025; its data stops at the 2022 elections. 2026 results will be added once states certify them.