Tarrance Group
Graded against the actual result across 63 races (from 76 polls, through 2024).
Head-to-head vs the VotePredictor model
The fair, apples-to-apples test: on the 49 races Tarrance Group actually polled, how its final poll's margin compared to what the VotePredictor model predicted for those same races.
| Model | Avg miss (pts) | Called right |
|---|---|---|
| Tarrance Group | 6.13 | 71% |
| VotePredictor | 5.58 | 84% |
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 (63)
Each race Tarrance 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.
| Race | Their call | Result | Miss | Call |
|---|---|---|---|---|
| 2024 MI Senate | R+2.0 | D+0.3 | 2.3 | ✗ |
| 2022 IA Governor | R+17.0 | R+18.5 | 1.5 | ✓ |
| 2022 NY-25 House | D+4.0 | D+11.3 | 7.3 | ✓ |
| 2022 IA Senate | R+11.0 | R+12.2 | 1.2 | ✓ |
| 2020 IL-17 House | D+5.0 | D+4.1 | 0.9 | ✓ |
| 2020 NM-2 House | R+1.0 | R+7.4 | 6.4 | ✓ |
| 2020 PA-10 House | R+4.0 | R+6.6 | 2.6 | ✓ |
| 2020 MI Senate | D+1.3 | D+1.7 | 0.4 | ✓ |
| 2018 CO-6 House | D+2.0 | D+11.2 | 9.2 | ✓ |
| 2018 NM-2 House | R+4.0 | D+1.9 | 5.9 | ✗ |
| 2018 TX-31 House | R+21.0 | R+2.9 | 18.1 | ✓ |
| 2018 MI Senate | D+7.0 | D+6.5 | 0.5 | ✓ |
| 2018 WV Senate | D+4.0 | D+3.3 | 0.7 | ✓ |
| 2016 CA-24 House | R+3.0 | D+6.8 | 9.8 | ✗ |
| 2016 IA-3 House | R+15.0 | R+13.7 | 1.3 | ✓ |
| 2016 NV-3 House | R+5.0 | D+1.3 | 6.3 | ✗ |
| 2016 VA President | D+9.0 | D+5.3 | 3.7 | ✓ |
| 2014 AZ Governor | R+7.0 | R+11.8 | 4.8 | ✓ |
| 2014 HI Governor | D+3.0 | D+12.4 | 9.4 | ✓ |
| 2014 IA-3 House | R+2.0 | R+10.5 | 8.5 | ✓ |
| 2014 KS-2 House | R+12.0 | R+18.4 | 6.4 | ✓ |
| 2014 MA-6 House | R+1.0 | D+13.8 | 14.8 | ✗ |
| 2014 MN-7 House | R+1.0 | D+8.5 | 9.5 | ✗ |
| 2014 VA-10 House | R+12.0 | R+16.1 | 4.1 | ✓ |
| 2012 CA-52 House | R+4.0 | D+2.4 | 6.4 | ✗ |
| 2012 CA-9 House | R+1.0 | D+11.1 | 12.1 | ✗ |
| 2012 NV-4 House | R+10.0 | D+8.0 | 18.0 | ✗ |
| 2012 PA-4 House | R+22.0 | R+25.3 | 3.3 | ✓ |
| 2012 AZ Senate | R+6.0 | R+3.0 | 3.0 | ✓ |
| 2010 CA Governor | EVEN | D+12.9 | 12.9 | ✗ |
| 2010 IA-2 House | R+1.0 | D+5.1 | 6.1 | ✗ |
| 2010 IL-14 House | R+6.0 | R+6.3 | 0.3 | ✓ |
| 2010 IL-17 House | R+1.0 | R+9.6 | 8.6 | ✓ |
| 2010 KY-6 House | R+1.0 | D+0.3 | 1.3 | ✗ |
| 2010 MS-1 House | R+11.0 | R+14.5 | 3.5 | ✓ |
| 2010 MS-4 House | R+2.0 | R+5.1 | 3.1 | ✓ |
| 2010 NM-2 House | R+9.0 | R+10.8 | 1.8 | ✓ |
| 2010 PA-10 House | R+5.0 | R+10.4 | 5.4 | ✓ |
| 2010 TN-8 House | R+15.0 | R+20.2 | 5.2 | ✓ |
| 2010 CA Senate | EVEN | D+10.0 | 10.0 | ✗ |
| 2008 CA-50 House | R+13.0 | R+5.1 | 7.9 | ✓ |
| 2008 FL-16 House | R+26.0 | R+20.2 | 5.8 | ✓ |
| 2008 IA-4 House | R+22.0 | R+21.1 | 0.9 | ✓ |
| 2008 MN-1 House | D+18.0 | D+29.6 | 11.6 | ✓ |
| 2008 OH-7 House | R+24.0 | R+16.4 | 7.6 | ✓ |
| 2006 NJ-7 House | R+15.0 | R+1.5 | 13.5 | ✓ |
| 2004 LA-3 House | R+7.0 | D+0.5 | 7.5 | ✗ |
| 2004 CO Senate | D+2.0 | D+4.8 | 2.8 | ✓ |
| 2004 WI Senate | D+5.0 | D+11.2 | 6.2 | ✓ |
| 2002 CA Governor | D+3.0 | D+4.9 | 1.9 | ✓ |
| 2002 CO-7 House | R+2.0 | R+0.1 | 1.9 | ✓ |
| 2002 FL-5 House | D+2.0 | R+1.7 | 3.7 | ✗ |
| 2002 GA-11 House | D+4.0 | R+3.3 | 7.3 | ✗ |
| 2002 GA-12 House | R+15.0 | R+10.4 | 4.6 | ✓ |
| 2002 NH-2 House | R+15.0 | R+16.0 | 1.0 | ✓ |
| 2002 TX-11 House | D+8.0 | D+4.4 | 3.6 | ✓ |
| 2002 UT-2 House | D+1.0 | D+0.7 | 0.3 | ✓ |
| 2002 CO Senate | R+6.0 | R+4.9 | 1.1 | ✓ |
| 2000 FL President | R+4.0 | EVEN | 4.0 | ✓ |
| 2000 MN Senate | D+1.0 | D+5.5 | 4.5 | ✓ |
| 2000 NJ Senate | R+4.0 | D+3.0 | 7.0 | ✗ |
| 1998 FL Governor | R+12.0 | R+10.5 | 1.5 | ✓ |
| 1998 AR Senate | D+8.0 | D+12.8 | 4.8 | ✓ |
Accuracy by time to election
Lower is better. Time to election runs right (election week) to left (~2 months out).
By the numbers
| Time to election | Polls | Avg miss | vs field | Called right |
|---|---|---|---|---|
| 1–3 wk | 24 | 4.64 | -0.43 | 75% |
| 3–6 wk | 35 | 7.59 | +1.85 | 69% |
| 6–9 wk | 14 | 7.77 | +1.62 | 64% |
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?
| Year | Polls | Avg miss | Lean (house effect) |
|---|---|---|---|
| 2000 | 3 | 5.2 | R+5.2 |
| 2002 | 10 | 3.6 | D+1.7 |
| 2004 | 3 | 5.5 | R+5.5 |
| 2008 | 7 | 15.2 | D+5.5 |
| 2010 | 16 | 5.9 | D+1.6 |
| 2012 | 5 | 8.6 | R+7.2 |
| 2014 | 10 | 7.7 | R+1.8 |
| 2016 | 5 | 5.5 | R+4.0 |
| 2018 | 5 | 6.9 | R+6.4 |
| 2020 | 5 | 2.1 | D+2.0 |
| 2022 | 3 | 3.3 | R+1.5 |
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.