The 2024 Debrief · post-mortem · permanent archive
How did the polling industry do this cycle?
The 2024 industry MAE landed at 4.1pt — graded across 376 historical races with a known outcome — and 268 polls were analyzed across 86 pollsters. Signed bias −3.18pt: pollsters again systematically over-predicted Democrats.
INDUSTRY MAE
4.1pt
vs midterm avg
SIGNED BIAS
−3.18pt
D-leaning
POLLS · POLLSTERS
268
86 firms
RACES SCORED
376
w/ known outcome
Industry mean absolute error · 2012–2024
PoliAgg model · self-grade · 2024
How did our own production model do this cycle? The model issues one prediction per race; the polling industry above is averaged across many polls per race. Metrics are not directly comparable but published side-by-side for transparency.
Mean error per race
5.18pp
poliagg-v8 ensemble vs final margin
Call accuracy
96.3%
fraction with right winner called
Races graded
376
outcomes used in the grade
Per-pollster scorecard · 2024 · sorted by MAE
Signed bias · ±5.0pt
MAE · 0–5.0pt
Biggest-miss races · 2024
| Race | Expected | Actual | Miss | Direction |
|---|---|---|---|---|
| us-senate-2024-fl | D+4.00 | D+12.78 | -8.8pt | R-margin underestimate |
| us-senate-2024-nj | R+18.00 | R+9.62 | -8.4pt | D-margin underestimate |
| us-senate-2024-tn | D+22.00 | D+29.64 | -7.6pt | R-margin underestimate |
| us-senate-2024-ny | R+25.00 | R+17.53 | -7.5pt | D-margin underestimate |
| us-senate-special-2024-ne | D+18.00 | D+25.16 | -7.2pt | R-margin underestimate |
| us-senate-2024-me | D+24.00 | D+17.42 | 6.6pt | R-margin underestimate |
| us-governor-2024-mo | D+14.00 | D+20.42 | -6.4pt | R-margin underestimate |
| us-governor-2024-nh | D+3.00 | D+9.34 | -6.3pt | R-margin underestimate |