NonpartisanIndependent, data-first coverage of US elections — polls, forecasts, and prediction markets, refreshed daily.Updated daily.What's new·Methodology·Send feedback
Debrief · 2004 cycle
The 2004 Debrief · post-mortem · permanent archive

How did the polling industry do this cycle?

The 2004 industry MAE landed at 4.5pt — graded across 94 historical races with a known outcome — and 131 polls were analyzed across 35 pollsters. Signed bias +0.15pt: polls landed close to the mean — no systematic over-predict on either side.

INDUSTRY MAE
4.5pt
vs midterm avg
SIGNED BIAS
+0.15pt
R-leaning
POLLS · POLLSTERS
131
35 firms
RACES SCORED
94
w/ known outcome
CYCLE200220042006200820102012201420162018202020222024
Industry mean absolute error · 2002–2004
20024.0pt20044.5pt20065.5pt20084.3pt20105.2pt20124.7pt20144.8pt20166.8pt20184.5pt20206.0pt20224.6pt20244.1ptMAE · 2-WAY MARGIN
Per-pollster scorecard · 2004 · sorted by MAE
Signed bias · ±9.4pt
MAE · 0–9.4pt
research-2000
+0.24pt
1.0pt
tarrance-group
+1.64pt
2.6pt
surveyusa
+0.98pt
3.5pt
mason-dixon
−0.57pt
3.6pt
rasmussen-reports
+0.58pt
3.9pt
strategic-vision
+3.61pt
4.6pt
sooner-poll
−8.93pt
8.9pt
Biggest-miss races · 2004
RaceExpectedActualMissDirection
us-senate-2004-paR+20.00R+10.639.4ptR-margin overestimate
us-senate-2004-okR+3.20R+11.53-8.3ptR-margin underestimate
us-senate-2004-kyR+7.50R+1.326.2ptR-margin overestimate
us-governor-2004-waD+5.00tied-5.0ptD-overestimate
us-senate-2004-sdR+3.00R+1.151.9ptR-margin overestimate
us-senate-2004-wiD+13.00D+11.24-1.8ptD-margin overestimate
us-senate-2004-akR+1.50R+3.03-1.5ptR-margin underestimate
us-senate-2004-waD+11.00D+12.241.2ptD-margin underestimate