VotePredictor
← All pollsters

RKM Research and Communications Inc.

Graded against the actual result across 21 races (from 34 polls, through 2016).

Races polled
21
Polls
34
Avg miss
6.78 pts
Most recent
2016

Every race (21)

Each race RKM Research and Communications Inc. 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
2016 US PresidentD+3.2D+2.21.0
2004 NE-1 HouseR+4.0R+11.27.2
2004 NE-2 HouseR+30.0R+24.95.1
2004 NE-3 HouseR+78.0R+76.91.1
2004 NE PresidentR+29.0R+33.24.2
2002 MA GovernorD+1.0R+4.85.8
2002 ME GovernorD+10.0D+5.74.3
2002 ME-2 HouseD+1.0D+4.03.0
2002 ME SenateR+28.0R+16.911.1
2000 MA PresidentD+24.0D+27.03.0
2000 ME PresidentEVEND+5.15.1
2000 NE PresidentR+25.0R+29.04.0
2000 NH PresidentD+1.0R+1.32.3
2000 MA SenateD+49.0D+56.97.9
2000 ME SenateR+51.0R+37.913.1
2000 NE SenateD+12.0D+2.29.8
1998 MA GovernorR+9.0R+3.45.6
1998 NE GovernorR+14.0R+7.96.1
1998 NH GovernorD+41.0D+35.25.8
1998 NH-2 HouseR+15.0R+8.46.6
1998 NH SenateR+45.0R+39.65.4

Accuracy by time to election

24686–9 wk3–6 wk1–3 wk≤1 wkavg miss (pts)
RKM Research and Communications Inc.All 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
1–3 wk175.48+0.4182%
3–6 wk47.05+1.3175%
6–9 wk109.65+3.5090%

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)
199875.5R+3.9
2000108.1D+2.3
2002108.9D+3.1
200444.4D+1.3
201631.3D+1.2

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.