CBS News/The New York Times
Graded against the actual result across 22 races (from 40 polls, through 2016).
Head-to-head vs VotePredictor Elections
The fair, apples-to-apples test: on the 11 races CBS News/The New York Times actually polled, how its final poll's margin compared to what VotePredictor Elections predicted for those same races.
| Model | Avg miss (pts) | Called right |
|---|---|---|
| CBS News/The New York Times | 4.53 | 100% |
| VotePredictor Elections | 3.23 | 100% |
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 (22)
Each race CBS News/The New York Times 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+3.0 | D+2.2 | 0.8 | ✓ |
| 2012 US President | D+1.0 | D+3.9 | 2.9 | ✓ |
| 2010 OH Governor | R+1.0 | R+2.0 | 1.0 | ✓ |
| 2010 OH Senate | R+11.0 | R+17.4 | 6.4 | ✓ |
| 2008 US President | D+13.0 | D+7.4 | 5.6 | ✓ |
| 2006 NY Governor | D+50.0 | D+38.6 | 11.4 | ✓ |
| 2006 OH Governor | D+25.0 | D+23.9 | 1.1 | ✓ |
| 2006 NJ Senate | D+1.0 | D+16.0 | 15.0 | ✓ |
| 2006 NY Senate | D+30.0 | D+34.4 | 4.4 | ✓ |
| 2006 OH Senate | D+13.0 | D+12.3 | 0.7 | ✓ |
| 2004 US President | R+2.0 | R+2.4 | 0.4 | ✓ |
| 2002 NJ Senate | D+13.0 | D+9.9 | 3.1 | ✓ |
| 2001 NJ Governor | D+13.0 | D+14.8 | 1.8 | ✓ |
| 2000 FL President | D+5.0 | EVEN | 5.0 | ✗ |
| 2000 MI President | D+6.0 | D+5.1 | 0.9 | ✓ |
| 2000 NY President | D+20.0 | D+24.5 | 4.5 | ✓ |
| 2000 US President | D+1.0 | D+0.5 | 0.5 | ✓ |
| 2000 FL Senate | D+9.0 | D+4.9 | 4.1 | ✓ |
| 2000 MI Senate | R+11.0 | D+1.6 | 12.6 | ✗ |
| 2000 NY Senate | D+8.0 | D+11.9 | 3.9 | ✓ |
| 1998 NY Governor | R+35.0 | R+20.1 | 14.9 | ✓ |
| 1998 NY Senate | EVEN | D+9.9 | 9.9 | ✗ |
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 | 4 | 2.23 | -2.01 | 50% |
| 1–3 wk | 16 | 5.42 | +0.36 | 81% |
| 3–6 wk | 14 | 4.68 | -1.02 | 93% |
| 6–9 wk | 6 | 2.59 | -3.55 | 83% |
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 | 4 | 9.9 | R+9.9 |
| 2000 | 13 | 3.9 | R+1.2 |
| 2004 | 3 | 2.5 | R+1.2 |
| 2006 | 5 | 6.5 | R+1.2 |
| 2008 | 6 | 4.2 | D+2.9 |
| 2016 | 3 | 1.6 | D+0.1 |
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.