Lake Research Partners/The Tarrance Group
Graded against the actual result across 8 races (from 42 polls, through 2016).
Every race (8)
Each race Lake Research Partners/The 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 |
|---|---|---|---|---|
| 2016 US President | D+8.0 | D+2.2 | 5.8 | ✓ |
| 2012 US President | EVEN | D+3.9 | 3.9 | ✗ |
| 2008 US President | D+5.0 | D+7.4 | 2.4 | ✓ |
| 2004 US President | R+6.0 | R+2.4 | 3.6 | ✓ |
| 2000 MI President | D+1.0 | D+5.1 | 4.1 | ✓ |
| 2000 PA President | D+5.0 | D+4.2 | 0.8 | ✓ |
| 2000 US President | R+5.0 | D+0.5 | 5.5 | ✗ |
| 2000 WI President | D+1.0 | D+0.2 | 0.8 | ✓ |
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 wk | 8 | 4.35 | +0.11 | 50% |
| 1–3 wk | 10 | 3.06 | -2.00 | 80% |
| 3–6 wk | 16 | 3.19 | -2.51 | 63% |
| 6–9 wk | 8 | 4.67 | -1.47 | 25% |
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 | 15 | 3.8 | R+3.6 |
| 2004 | 8 | 2.2 | R+2.2 |
| 2008 | 10 | 4.9 | R+4.8 |
| 2012 | 8 | 3.0 | R+2.9 |
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.