Benenson Strategy Group
Graded against the actual result across 34 races (from 40 polls, through 2014).
Head-to-head vs the VotePredictor model
The fair, apples-to-apples test: on the 34 races Benenson Strategy 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 |
|---|---|---|
| Benenson Strategy Group | 6.46 | 62% |
| VotePredictor | 4.73 | 91% |
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 (34)
Each race Benenson Strategy 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 |
|---|---|---|---|---|
| 2014 CO Senate | D+3.0 | R+1.9 | 4.9 | ✗ |
| 2012 IN Governor | R+6.0 | R+2.9 | 3.1 | ✓ |
| 2012 FL-26 House | D+11.0 | D+10.6 | 0.4 | ✓ |
| 2012 GA-12 House | D+3.0 | D+7.4 | 4.4 | ✓ |
| 2012 IA-3 House | EVEN | R+8.6 | 8.6 | ✗ |
| 2012 IL-12 House | D+8.0 | D+8.9 | 0.9 | ✓ |
| 2012 MI-1 House | D+3.0 | R+0.5 | 3.5 | ✗ |
| 2012 NV-3 House | R+5.0 | R+7.5 | 2.5 | ✓ |
| 2012 RI-1 House | D+11.0 | D+12.1 | 1.1 | ✓ |
| 2012 VA-2 House | R+9.0 | R+7.7 | 1.3 | ✓ |
| 2012 HI Senate | D+18.0 | D+25.2 | 7.2 | ✓ |
| 2010 OH Governor | D+1.0 | R+2.0 | 3.0 | ✗ |
| 2010 AZ-5 House | D+7.0 | R+8.8 | 15.8 | ✗ |
| 2010 IL-14 House | D+10.0 | R+6.3 | 16.3 | ✗ |
| 2010 VA-5 House | R+1.0 | R+3.8 | 2.8 | ✓ |
| 2010 WA-9 House | D+19.0 | D+9.7 | 9.3 | ✓ |
| 2010 KY Senate | R+3.0 | R+11.5 | 8.5 | ✓ |
| 2009 NY-20 House | R+7.0 | D+0.9 | 7.9 | ✗ |
| 2008 FL-8 House | D+11.0 | D+4.0 | 7.0 | ✓ |
| 2008 KY-2 House | D+6.0 | R+5.1 | 11.1 | ✗ |
| 2008 MI-7 House | D+8.0 | D+2.3 | 5.7 | ✓ |
| 2008 NJ-7 House | D+9.0 | R+8.0 | 17.0 | ✗ |
| 2008 NY-29 House | D+5.0 | D+5.0 | 0.0 | ✓ |
| 2008 OH-15 House | D+9.0 | D+0.8 | 8.2 | ✓ |
| 2008 TX-22 House | EVEN | R+7.1 | 7.1 | ✗ |
| 2008 VA-5 House | R+8.0 | D+0.2 | 8.2 | ✗ |
| 2008 OR Senate | D+2.0 | D+3.4 | 1.4 | ✓ |
| 2006 TN Governor | D+40.0 | D+38.9 | 1.1 | ✓ |
| 2006 IA-3 House | D+20.0 | D+5.4 | 14.6 | ✓ |
| 2006 IL-8 House | D+19.0 | D+6.9 | 12.1 | ✓ |
| 2006 NY-25 House | R+2.0 | D+4.1 | 6.1 | ✗ |
| 2006 PA-7 House | D+7.0 | D+12.8 | 5.8 | ✓ |
| 2006 TN Senate | D+5.0 | R+2.7 | 7.7 | ✗ |
| 2005 VA Governor | D+1.0 | D+5.7 | 4.7 | ✓ |
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 | 11 | 7.14 | +2.07 | 55% |
| 3–6 wk | 16 | 7.23 | +1.49 | 56% |
| 6–9 wk | 13 | 5.21 | -0.94 | 77% |
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) |
|---|---|---|---|
| 2006 | 8 | 8.0 | D+3.0 |
| 2008 | 10 | 7.2 | D+5.3 |
| 2010 | 8 | 8.4 | D+8.4 |
| 2012 | 11 | 3.8 | R+1.0 |
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.