VotePredictor
← All pollsters

Opinion Dynamics

Graded against the actual result across 18 races (from 37 polls, through 2012).

Races polled
18
Polls
37
Avg miss
3.06 pts
Most recent
2012

Head-to-head vs the VotePredictor model

The fair, apples-to-apples test: on the 14 races Opinion Dynamics actually polled, how its final poll's margin compared to what the VotePredictor model predicted for those same races.

ModelAvg miss (pts)Called right
Opinion Dynamics4.0171%
VotePredictor1.9686%

VotePredictor aggregates all the pollsters, so it's expected to beat any single one on margin — that's the value of averaging. The honest comparison among forecasters is on the combined board.

Every race (18)

Each race Opinion Dynamics polled, scored on its final poll — the call right before the vote — against the actual Dem−Rep result. Click a race for its full detail.

RaceTheir callResultMissCall
2012 MA SenateD+4.0D+7.53.5
2008 US PresidentD+7.0D+7.40.4
2004 FL PresidentD+5.0R+5.010.0
2004 IA PresidentR+4.0R+0.73.3
2004 MI PresidentD+2.0D+3.41.4
2004 OH PresidentR+3.0R+2.10.9
2004 PA PresidentD+3.0D+2.50.5
2004 US PresidentD+2.0R+2.44.4
2004 WI PresidentR+3.0D+0.43.4
2004 FL SenateD+6.0R+1.17.1
2004 IA SenateR+47.0R+42.34.7
2004 OH SenateR+30.0R+27.72.3
2004 PA SenateR+22.0R+10.611.4
2004 WI SenateD+14.0D+11.22.8
2000 IL PresidentR+3.0D+12.015.0
2000 MI PresidentD+3.0D+5.12.1
2000 PA PresidentD+3.0D+4.21.2
2000 US PresidentEVEND+0.50.5

Accuracy by time to election

24686–9 wk3–6 wk1–3 wk≤1 wkavg miss (pts)
Opinion DynamicsAll pollsters (field average)

Lower is better. Time to election runs right (election week) to left (~2 months out).

By the numbers

Time to electionPollsAvg missvs fieldCalled right
≤1 wk123.35-0.9358%
1–3 wk93.86-1.2178%
3–6 wk132.07-3.67100%

vs field is this pollster's average miss minus all pollsters' at the same lead time — green beats the field, redtrails it. Our historical polls reach ~2 months out; earlier polling isn't in the record.

Track record by cycle — getting better?

YearPollsAvg missLean (house effect)
200073.3R+3.2
2004233.0R+0.5
200863.1R+2.6

Do we credit a pollster for fixing its bias? Each cycle, the model re-estimates every pollster's lean from all its earlier polls (walk-forward) and subtracts it before using the poll. We tested weighting recent cycles more — it doesn't help: a pollster's lean in one cycle barely predicts the next (correlation 0.28), so the swings above are mostly noise, and averaging over more history beats chasing the latest cycle. The all-time estimate we use came out within ~0.5% of the best option.