University of Connecticut
Graded against the actual result across 25 races (from 42 polls, through 2012).
Head-to-head vs VotePredictor Elections
The fair, apples-to-apples test: on the 13 races University of Connecticut actually polled, how its final poll's margin compared to what VotePredictor Elections predicted for those same races.
| Model | Avg miss (pts) | Called right |
|---|---|---|
| University of Connecticut | 3.81 | 77% |
| VotePredictor Elections | 3.13 | 77% |
VotePredictor Elections 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 (25)
Each race University of Connecticut 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 |
|---|---|---|---|---|
| 2012 CT President | D+14.0 | D+17.3 | 3.3 | ✓ |
| 2012 US President | D+3.0 | D+3.9 | 0.9 | ✓ |
| 2012 CT Senate | D+6.0 | D+12.5 | 6.5 | ✓ |
| 2008 CT-2 House | D+33.0 | D+29.2 | 3.8 | ✓ |
| 2008 CT-4 House | EVEN | D+0.8 | 0.8 | ✗ |
| 2008 CT President | D+25.0 | D+22.4 | 2.6 | ✓ |
| 2006 CT Governor | R+22.0 | R+27.8 | 5.8 | ✓ |
| 2006 CT-2 House | R+2.0 | EVEN | 2.0 | ✗ |
| 2006 CT-4 House | EVEN | R+3.4 | 3.4 | ✗ |
| 2006 CT-5 House | D+4.0 | D+12.9 | 8.9 | ✓ |
| 2006 CT Senate | D+36.0 | D+30.1 | 5.9 | ✓ |
| 2004 CT-2 House | R+3.0 | R+8.4 | 5.4 | ✓ |
| 2004 CT-4 House | R+5.0 | R+4.9 | 0.1 | ✓ |
| 2002 CT Governor | R+15.0 | R+12.2 | 2.8 | ✓ |
| 2002 CT-2 House | R+5.0 | R+8.2 | 3.2 | ✓ |
| 2002 CT-5 House | R+17.0 | R+11.0 | 6.0 | ✓ |
| 2000 CT President | D+16.0 | D+17.5 | 1.5 | ✓ |
| 2000 FL President | EVEN | EVEN | 0.0 | ✗ |
| 2000 NH President | D+10.0 | R+1.3 | 11.3 | ✗ |
| 2000 PA President | D+6.0 | D+4.2 | 1.8 | ✓ |
| 2000 CT Senate | D+33.0 | D+29.0 | 4.0 | ✓ |
| 2000 FL Senate | EVEN | D+4.9 | 4.9 | ✗ |
| 2000 PA Senate | R+19.0 | R+6.9 | 12.1 | ✓ |
| 1998 CT Governor | R+37.0 | R+27.5 | 9.5 | ✓ |
| 1998 CT Senate | D+40.0 | D+32.8 | 7.2 | ✓ |
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 | 7 | 4.02 | -1.05 | 86% |
| 3–6 wk | 22 | 4.69 | -1.05 | 86% |
| 6–9 wk | 10 | 5.93 | -0.22 | 70% |
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 | 7 | 5.1 | R+0.2 |
| 2002 | 11 | 4.8 | R+3.1 |
| 2004 | 5 | 2.5 | D+1.0 |
| 2006 | 6 | 4.6 | D+0.4 |
| 2008 | 5 | 6.0 | R+3.5 |
| 2012 | 6 | 4.0 | R+2.7 |
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.