VotePredictor
← All pollsters

Garin-Hart-Yang Research Group

Graded against the actual result across 87 races (from 101 polls, through 2022).

Races polled
87
Polls
101
Avg miss
6.03 pts
Most recent
2022

Head-to-head vs the VotePredictor model

The fair, apples-to-apples test: on the 73 races Garin-Hart-Yang Research Group actually polled, how its final poll's margin compared to what the VotePredictor model predicted for those same races.

ModelAvg miss (pts)Called right
Garin-Hart-Yang Research Group6.1873%
VotePredictor3.9481%

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 (86)

Each race Garin-Hart-Yang Research Group 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
2022 AZ SenateD+4.0D+4.90.9
2022 GA SenateD+3.0D+1.02.0
2022 NV SenateD+1.0D+0.80.2
2022 PA SenateD+2.0D+4.92.9
2022 WI SenateR+2.0R+1.01.0
2020 MO GovernorR+2.0R+16.414.4
2020 OH-10 HouseR+4.0R+16.712.7
2020 VA-10 HouseD+22.0D+13.18.9
2020 AZ PresidentD+11.0D+0.310.7
2020 FL PresidentD+6.0R+3.49.4
2020 GA PresidentD+7.0D+0.26.8
2020 IA PresidentR+2.0R+8.26.2
2020 MI PresidentD+5.0D+2.82.2
2020 MO PresidentR+2.0R+15.413.4
2020 NC PresidentD+2.0R+1.33.3
2020 OH PresidentR+2.0R+8.06.0
2020 PA PresidentD+11.0D+1.29.8
2020 TX PresidentR+2.0R+5.63.6
2020 US PresidentD+8.0D+4.43.6
2020 WI PresidentD+7.0D+0.66.4
2020 AZ SenateD+12.0D+2.39.7
2020 GA SenateD+5.0R+1.86.8
2020 IA SenateEVENR+6.66.6
2020 MI SenateD+8.0D+1.76.3
2020 NC SenateD+13.0R+1.714.7
2018 GA GovernorD+6.0R+1.47.4
2018 IL-6 HouseD+5.0D+7.22.2
2018 KY-6 HouseD+7.0R+3.210.2
2018 VA-2 HouseD+8.0D+2.25.8
2018 TN SenateD+2.0R+10.812.8
2017 MT-1 HouseR+6.0R+5.60.4
2016 WV GovernorD+13.0D+6.86.2
2016 IN-9 HouseEVENR+13.713.7
2016 KS-3 HouseR+4.0R+10.76.7
2016 NJ-5 HouseD+7.0D+4.42.6
2016 NV PresidentD+3.0D+2.40.6
2016 WV PresidentR+32.0R+42.110.1
2016 NV SenateR+1.0D+2.43.4
2014 CA-7 HouseD+4.0D+0.83.2
2014 CT-5 HouseD+16.0D+6.59.5
2014 MA-6 HouseD+11.0D+13.82.8
2014 NJ-5 HouseR+5.0R+12.17.1
2014 IA SenateEVENR+8.38.3
2014 NC SenateD+3.0R+1.64.6
2012 WI GovernorR+2.0R+6.84.8
2012 CT-5 HouseD+4.0D+3.10.9
2012 FL-18 HouseD+9.0D+0.68.4
2012 IL-11 HouseD+4.0D+17.113.1
2012 MI-1 HouseD+9.0R+0.59.5
2012 MT SenateD+1.0D+3.72.7
2012 PA SenateD+13.0D+9.13.9
2012 VA SenateD+4.0D+5.91.9
2010 MD-1 HouseD+6.0R+12.118.1
2010 ND-1 HouseD+2.0R+9.811.8
2010 WI-7 HouseR+1.0R+7.76.7
2010 MO SenateR+4.0R+13.69.6
2010 PA SenateD+3.0R+2.05.0
2008 NC GovernorD+6.0D+3.42.6
2008 MD-1 HouseD+2.0D+0.81.2
2008 NC PresidentR+3.0D+0.33.3
2008 CO SenateD+12.0D+10.31.7
2008 KY SenateR+2.0R+5.93.9
2008 NC SenateR+2.0D+8.510.5
2006 MD GovernorD+6.0D+6.50.5
2006 PA-6 HouseD+3.0R+1.34.3
2006 MD SenateD+12.0D+10.02.0
2006 VA SenateD+4.0D+0.43.6
2004 UT GovernorEVENR+16.416.4
2004 NV-3 HouseR+3.0R+14.111.1
2004 MN PresidentD+7.0D+3.53.5
2004 NM PresidentD+4.0R+0.84.8
2004 NV PresidentR+1.0R+2.61.6
2004 KY SenateR+1.0R+1.30.3
2002 FL GovernorR+3.0R+12.89.8
2002 IL GovernorD+16.0D+7.18.9
2002 TN GovernorD+8.0D+3.14.9
2002 IN-8 HouseR+9.0R+5.33.7
2002 CO SenateR+2.0R+4.92.9
2002 NC SenateR+3.0R+8.65.6
2002 SC SenateR+8.0R+10.22.2
2002 TN SenateR+7.0R+9.92.9
2001 VA GovernorD+10.0D+5.14.9
2000 US PresidentD+3.0D+0.52.5
1998 ID-1 HouseD+2.0R+10.512.5
1998 ID-2 HouseD+6.0R+7.813.8
1998 NY SenateD+4.0D+14.210.2

Accuracy by time to election

24686–9 wk3–6 wk1–3 wk≤1 wkavg miss (pts)
Garin-Hart-Yang Research GroupAll 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 wk234.42-0.6587%
3–6 wk496.45+0.7171%
6–9 wk296.61+0.4672%

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)
1998312.2D+5.4
2002115.6D+4.9
200485.7D+4.0
200663.1D+2.4
200873.7R+0.2
2010510.3D+10.3
201295.7D+1.7
201475.6D+3.7
201676.2D+5.2
201867.1D+5.0
2020237.6D+7.5
202251.4R+0.5

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.