New York's 21st congressional district
U.S. House of Representatives
14days to polls
The call
200 synthetic voters, set beside the poll average.
SyntheticD+18.6Blake Gendebien
Poll average · 6 pollsR+2.5
Gap+21.1synthetic − polls, points
SyntheticPoll average
Anthony Constantino
32.0 synthetic44.2 polls
Blake Gendebien
46.7 synthetic42.0 polls
Christopher Schmidt
8.6 synthetic4.8 polls
Against the polls
- Share error
- 6.1 pts per option
- Crosstab error
- 5.7 pts
- Sample
- 200 synthetic voters
By group
Each cell is synthetic / Siena Research Institute (Times Union / WPTZ / WAMC), shares within the group.
| Group | n | R | D | I | Error |
|---|---|---|---|---|---|
| Democrats | 105 | 16.1 / 10.0 | 64.3 / 85.0 | 8.4 / 3.0 | 10.3 |
| Republicans | 54 | 63.6 / 65.0 | 17.9 / 21.0 | 7.4 / 4.0 | 2.2 |
| independent / other | 41 | 30.8 / 30.3 | 39.5 / 51.5 | 11.0 / 9.1 | 6.0 |
| men | 93 | 32.6 / 49.0 | 46.4 / 39.0 | 8.9 / 4.0 | 8.2 |
| women | 107 | 31.5 / 34.3 | 47.0 / 52.5 | 8.4 / 5.1 | 4.1 |
| age 18-34 | 47 | 29.5 / 30.0 | 46.4 / 56.0 | 9.6 / 8.0 | 5.1 |
| age 35-54 | 80 | 32.9 / 44.0 | 47.8 / 39.0 | 8.2 / 7.0 | 5.5 |
| age 55+ | 73 | 32.6 / 42.0 | 45.7 / 47.0 | 8.6 / 4.0 | 5.3 |
| Saratoga / Warren / Washington | 42 | 28.8 / 42.0 | 49.6 / 49.0 | 8.7 / 5.0 | 6.6 |
| Herkimer / Fulton / Montgomery / Oneida / Schoharie | 61 | 32.6 / 41.0 | 47.5 / 43.0 | 8.3 / 7.0 | 4.2 |
| Clinton / Essex / Franklin / Hamilton / Jefferson / Lewis / St. Lawrence | 97 | 33.1 / 41.6 | 45.0 / 45.5 | 8.8 / 4.0 | 4.5 |
| income under $50K | 60 | 32.6 / 46.5 | 43.4 / 38.4 | 9.1 / 7.1 | 7.0 |
| income $50K-$100K | 41 | 31.9 / 42.0 | 48.4 / 49.0 | 8.4 / 3.0 | 5.4 |
| income $100K+ | 90 | 32.5 / 40.0 | 48.0 / 48.0 | 8.2 / 4.0 | 3.8 |
| union household | 24 | 23.7 / 28.0 | 53.4 / 62.0 | 8.5 / 5.0 | 6.5 |
| non-union household | 176 | 33.2 / 47.5 | 45.8 / 38.6 | 8.6 / 5.0 | 7.2 |
The government
Donald Trump's job approval as President
Approve · synthetic31.3%64.7% disapprove
Approve · polls45.0%55.0% disapprove
Net gap−23.4synthetic − polls
Latest polls
Newest first.
| Pollster | Fieldwork to | Sample | D | R | I |
|---|---|---|---|---|---|
| Siena Research Institute (Times Union / WPTZ / WAMC) | OCT 06, 2026 | 496 (LV) | 46.0 | 42.0 | 5.0 |
| Impact Research (D) | OCT 04, 2026 | 500 (LV) | 45.0 | 48.0 | — |
| Public Policy Polling (D) | SEP 10, 2026 | 557 (RV) | 43.0 | 48.0 | — |
| Impact Research (D) | SEP 02, 2026 | 500 (LV) | 44.0 | 47.0 | — |
| J.L. Partners | AUG 17, 2026 | 500 (LV) | 40.0 | 50.0 | — |
| SurveyUSA | AUG 08, 2026 | 612 (LV) | 43.0 | 39.0 | — |
| Impact Research (D) | MAY 31, 2026 | 500 (LV) | 44.0 | 45.0 | — |
← United States midterm elections, 2026
How the call is made. For every race we build a synthetic electorate that matches the real one, at least 200 voters strong, and ask it how it will vote before polling day. The call is set beside the published polls as they come in, and scored against the official count once it is in.