Pollster scorecard · ranked 61 of 69
How accurate is Google Consumer Surveys?
Google Consumer Surveys' 12 polls in the final three weeks of 2016 races missed the final margin by 8.5 points on average, ranking 61st of 69 pollsters on the ElectionsUSA scorecard. On average its polls leaned 1.8 points toward Democrats, and 100% had the eventual winner ahead.
- Average error
- 8.5
- points on the final margin; all pollsters 5.8
- Adjusted error
- 6.99
- 61st of 69 on the scorecard
- Partisan bias
- D+1.8
- average signed miss
- Right call
- 100%
- of polls had the winner ahead
- Rated polls
- 12
- final three weeks, 2016
- 538 grade
- B/C
- final (2023)
Main survey method in the rated polls: Online.
By election year
| Year | Polls | Avg error | Bias | Right call |
|---|---|---|---|---|
| 2016 | 12 | 8.5 | D+1.8 | 100% |
By race
| Race | Polls | Avg error | Bias | Right call |
|---|---|---|---|---|
| President | 7 | 7.9 | D+7.4 | 100% |
| Senate | 4 | 11.1 | R+7.1 | 100% |
| House | 1 | 1.8 | R+1.8 | 100% |
How these numbers work
Accuracy covers polls taken in the final three weeks of races from 2016 to 2022, from FiveThirtyEight's open archive of polls matched to certified results (CC BY 4.0). Average error is the mean absolute difference between the poll's margin and the actual margin. The adjusted error shrinks it toward the field average (k = 15 polls) so a short hot streak can't outrank a long record, and the ranking uses it. Bias is the average signed miss: D+ means the polls were more favourable to Democrats than the result, R+ more favourable to Republicans.
Full method on the scorecard and methodology pages. Something wrong? Report it through the corrections page.