Ciruli Associates
Graded against the actual result across 15 races (from 33 polls, through 2016).
Every race (15)
Each race Ciruli Associates 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 CO President | EVEN | D+4.9 | 4.9 | ✗ |
| 2016 CO Senate | D+8.0 | D+5.7 | 2.3 | ✓ |
| 2008 CO President | D+1.0 | D+9.0 | 8.0 | ✓ |
| 2008 CO Senate | D+7.0 | D+10.3 | 3.3 | ✓ |
| 2006 CO Governor | D+19.0 | D+16.8 | 2.2 | ✓ |
| 2004 CO-4 House | R+17.0 | R+6.3 | 10.7 | ✓ |
| 2004 CO President | R+6.0 | R+4.7 | 1.3 | ✓ |
| 2004 CO Senate | D+4.0 | D+4.8 | 0.8 | ✓ |
| 2002 CO Governor | R+36.0 | R+29.0 | 7.0 | ✓ |
| 2002 CO-4 House | R+17.0 | R+13.3 | 3.7 | ✓ |
| 2002 CO-7 House | R+2.0 | R+0.1 | 1.9 | ✓ |
| 2002 CO Senate | D+4.0 | R+4.9 | 8.9 | ✗ |
| 2000 CO President | R+9.0 | R+8.4 | 0.6 | ✓ |
| 1998 CO Governor | R+4.0 | R+0.6 | 3.4 | ✓ |
| 1998 CO Senate | R+29.0 | R+27.5 | 1.5 | ✓ |
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 | 5 | 5.18 | +0.90 | 60% |
| 1–3 wk | 15 | 3.08 | -1.99 | 93% |
| 3–6 wk | 11 | 5.29 | -0.45 | 91% |
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) |
|---|---|---|---|
| 1998 | 6 | 3.5 | R+1.1 |
| 2000 | 4 | 0.9 | R+0.9 |
| 2002 | 12 | 5.5 | R+2.0 |
| 2004 | 5 | 4.8 | R+4.8 |
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.