NonpartisanIndependent, data-first coverage of US elections — polls, forecasts, and prediction markets, refreshed daily.Updated daily.What's new·Methodology·Send feedback
Debrief · 2008 cycle
The 2008 Debrief · post-mortem · permanent archive

How did the polling industry do this cycle?

The 2008 industry MAE landed at 4.3pt — graded across 95 historical races with a known outcome — and 44 polls were analyzed across 11 pollsters. Signed bias +0.86pt: pollsters again systematically over-predicted Republicans.

INDUSTRY MAE
4.3pt
vs midterm avg
SIGNED BIAS
+0.86pt
R-leaning
POLLS · POLLSTERS
44
11 firms
RACES SCORED
95
w/ known outcome
CYCLE200220042006200820102012201420162018202020222024
Industry mean absolute error · 2002–2008
20024.0pt20044.5pt20065.5pt20084.3pt20105.2pt20124.7pt20144.8pt20166.8pt20184.5pt20206.0pt20224.6pt20244.1ptMAE · 2-WAY MARGIN
Per-pollster scorecard · 2008 · sorted by MAE
Signed bias · ±7.2pt
MAE · 0–7.2pt
surveyusa
+2.81pt
4.6pt
survey-usa
+3.19pt
5.8pt
research-2000
−6.70pt
6.7pt
Biggest-miss races · 2008
RaceExpectedActualMissDirection
us-senate-2008-meR+13.50R+22.75-9.3ptR-margin underestimate
us-senate-2008-akD+8.00D+1.25-6.8ptD-margin overestimate
us-senate-2008-vaD+26.00D+31.305.3ptD-margin underestimate
us-governor-2008-waD+2.00D+6.484.5ptD-margin underestimate
us-senate-2008-iaD+21.00D+25.404.4ptD-margin underestimate
us-senate-2008-orD+5.50D+3.35-2.1ptD-margin overestimate
us-senate-2008-ncD+6.50D+8.472.0ptD-margin underestimate
us-governor-2008-wvD+46.00D+44.08-1.9ptD-margin overestimate