The 2020 Debrief · post-mortem · permanent archive
How did the polling industry do this cycle?
The 2020 industry MAE landed at 6.0pt — graded across 516 historical races with a known outcome — and 215 polls were analyzed across 92 pollsters. Signed bias −5.74pt: pollsters again systematically over-predicted Democrats.
INDUSTRY MAE
6.0pt
vs midterm avg
SIGNED BIAS
−5.74pt
D-leaning
POLLS · POLLSTERS
215
92 firms
RACES SCORED
516
w/ known outcome
Industry mean absolute error · 2002–2020
PoliAgg model · self-grade · 2020
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
11.24pp
poliagg-v16 ensemble vs final margin
Call accuracy
94.2%
fraction with right winner called
Races graded
464
outcomes used in the grade
Per-pollster scorecard · 2020 · sorted by MAE
Signed bias · ±8.7pt
MAE · 0–8.7pt
Biggest-miss races · 2020
| Race | Expected | Actual | Miss | Direction |
|---|---|---|---|---|
| us-senate-2020-me | D+3.00 | R+8.59 | -11.6pt | D-overestimate |
| us-senate-2020-ky | R+10.00 | R+19.53 | -9.5pt | R-margin underestimate |
| us-senate-2020-mt | R+1.00 | R+10.02 | -9.0pt | R-margin underestimate |
| us-senate-2020-al | R+12.00 | R+20.36 | -8.4pt | R-margin underestimate |
| us-senate-2020-sc | R+2.00 | R+10.27 | -8.3pt | R-margin underestimate |
| us-senate-2020-nj | D+24.00 | D+16.31 | -7.7pt | D-margin overestimate |
| us-senate-2020-ia | D+1.00 | R+6.59 | -7.6pt | D-overestimate |
| us-senate-2020-il | D+23.50 | D+16.06 | -7.4pt | D-margin overestimate |