Abt Associates
23 polls · 12 races · 2004–2017TI / Roper
Average error
3.64
1.91 better than the 5.55 average.
House effect
D+1.84
Systematically overstates Democrats by this much.
Accuracy rank
—
Too few polls to rank.
Methods used: Live Phone, NA
Accuracy by state
Only states where this pollster has at least five matched polls. A house effect that flips sign between states is usually noise; one that holds across many is worth paying attention to.
| State | Polls | Error | House effect |
|---|---|---|---|
| US | 13 | 3.56 | D+0.55 |
| Georgia | 7 | 3.38 | D+2.97 |
How this is measured
Every poll is compared to the actual result of the race it was polling. Error is the average absolute miss in points. The house effect is the average signed miss, positive when the poll overstated Democrats — computed only over clean Democrat-versus-Republican matchups, since the sign is meaningless otherwise. 1998–2023 comes from FiveThirtyEight's raw_polls.csv, used under CC BY 4.0 and not endorsed by them; 2024 is our own reconstruction, matching Wayback-recovered 538 poll files to results we compute from MEDSL's county presidential returns and statewide Senate file. Full caveats are on the accuracy overview.
TI / Roper is FiveThirtyEight's aapor_roper field, which 538 defines as a member of the AAPOR Transparency Initiative or a contributor of polls to the Roper Center archive. It is not AAPOR membership. 538 stamps the same value on every one of this pollster's polls, so it reflects the firm's status as of 538's last file — through 2023 — rather than its status at the time of any particular poll.