50-state check · 2026-09-01
The 50-state check.
As shipped: about 10 pp on 1,521 real polls, 50 states. Calibrated benchmark: 7.47 pp on 404 non-electoral questions. Strict held-out: 9.97 pp. Open-ended quotes and focus-group dialogue are qualitative and are not covered by these numbers. Election-margin questions are published but not claimed. How we measure →
Issue 4 is 225 questions we published before anyone knows the answers. This is the Census-weighted panel you actually get in the product. Two hundred simulated people answer each question. Fifty states. Washington, D.C. is held out of scoring. No extra calibration. A single state is coverage, not a headline.
As it ships
About 10 points off
As the product ships (200 people per question, no extra calibration) we are typically about 10 points off the real poll.
Elections
Not claimed
We do not claim election questions. A dummy that always picks the first option beats us there.
A narrower lab test
7.47%
A narrower April test, on 5 places with 10,000 people and extra calibration, missed by 7.47 points. That is real. It is not what a standard study in the product runs.
Issue 4 is 225 questions we published before anyone knows the answers.
Per-state coverage
Coverage map of option-share error by state. Colors mark MAE buckets. DC is not scored.
Per-state MAE · coverage, not a ranking
Click a state. Cells are coverage with intervals, not headlines.
- <7
- 7–9
- 9–12
- >12
- Not scored
Every state, with a range
Each state has only a few dozen questions. Read the range, not the point. D.C. is held out.
| Range | |||||
|---|---|---|---|---|---|
| AlaskaAK | 25 | 10.29 | 45.8 | 37.5 | 27.9%–64.9% |
| AlabamaAL | 28 | 10.41 | 67.9 | 60.7 | 49.3%–82.1% |
| ArkansasAR | 25 | 8.66 | 66.7 | 75 | 46.7%–82% |
| ArizonaAZ | 44 | 12.14 | 61.4 | 54.5 | 46.6%–74.3% |
| CaliforniaCA | 29 | 11.58 | 75.9 | 62.1 | 57.9%–87.8% |
| ColoradoCO | 34 | 13.82 | 73.5 | 70.6 | 56.9%–85.4% |
| ConnecticutCT | 23 | 11.32 | 73.9 | 65.2 | 53.5%–87.5% |
| District of ColumbiaDC | - | - | - | - | - |
| DelawareDE | 26 | 7.90 | 88.5 | 76.9 | 71%–96% |
| FloridaFL | 33 | 8.36 | 68.8 | 75 | 51.4%–82% |
| GeorgiaGA | 30 | 8.40 | 73.3 | 76.7 | 55.6%–85.8% |
| HawaiiHI | 23 | 8.93 | 91.3 | 73.9 | 73.2%–97.6% |
| IowaIA | 35 | 12.23 | 57.1 | 62.9 | 40.9%–72% |
| IdahoID | 27 | 11.19 | 59.3 | 66.7 | 40.7%–75.5% |
| IllinoisIL | 22 | 12.76 | 66.7 | 52.4 | 45.4%–82.8% |
| IndianaIN | 23 | 8.43 | 78.3 | 78.3 | 58.1%–90.3% |
| KansasKS | 24 | 12.80 | 73.9 | 82.6 | 53.5%–87.5% |
| KentuckyKY | 27 | 12.46 | 55.6 | 59.3 | 37.3%–72.4% |
| LouisianaLA | 23 | 10.23 | 77.3 | 81.8 | 56.6%–89.9% |
| MassachusettsMA | 37 | 8.98 | 75 | 58.3 | 58.9%–86.2% |
| MarylandMD | 24 | 7.75 | 91.7 | 58.3 | 74.2%–97.7% |
| MaineME | 30 | 7.55 | 53.3 | 46.7 | 36.1%–69.8% |
| MichiganMI | 41 | 8.80 | 73.2 | 58.5 | 58.1%–84.3% |
| MinnesotaMN | 25 | 8.12 | 79.2 | 54.2 | 59.5%–90.8% |
| MissouriMO | 28 | 8.40 | 71.4 | 57.1 | 52.9%–84.7% |
| MississippiMS | 26 | 15.79 | 61.5 | 73.1 | 42.5%–77.6% |
| MontanaMT | 33 | 10.59 | 75.8 | 87.9 | 59%–87.2% |
| North CarolinaNC | 57 | 7.84 | 60 | 43.6 | 46.8%–71.9% |
| North DakotaND | 26 | 12.85 | 80.8 | 76.9 | 62.1%–91.5% |
| NebraskaNE | 23 | 14.89 | 52.2 | 60.9 | 33%–70.8% |
| New HampshireNH | 24 | 9.03 | 60.9 | 34.8 | 40.8%–77.8% |
| New JerseyNJ | 24 | 8.22 | 69.6 | 47.8 | 49.1%–84.4% |
| New MexicoNM | 31 | 8.59 | 70 | 66.7 | 52.1%–83.3% |
| NevadaNV | 31 | 8.51 | 73.3 | 46.7 | 55.6%–85.8% |
| New YorkNY | 28 | 9.17 | 82.1 | 64.3 | 64.4%–92.1% |
| OhioOH | 42 | 8.92 | 53.7 | 56.1 | 38.7%–67.9% |
| OklahomaOK | 34 | 13.49 | 70.6 | 76.5 | 53.8%–83.2% |
| OregonOR | 24 | 12.77 | 66.7 | 41.7 | 46.7%–82% |
| PennsylvaniaPA | 43 | 7.77 | 73.8 | 50 | 58.9%–84.7% |
| Rhode IslandRI | 31 | 8.10 | 77.4 | 51.6 | 60.2%–88.6% |
| South CarolinaSC | 24 | 7.91 | 79.2 | 37.5 | 59.5%–90.8% |
| South DakotaSD | 23 | 13.93 | 78.3 | 60.9 | 58.1%–90.3% |
| TennesseeTN | 39 | 9.78 | 81.6 | 63.2 | 66.6%–90.8% |
| TexasTX | 48 | 8.12 | 56.5 | 45.7 | 42.2%–69.8% |
| UtahUT | 23 | 10.15 | 82.6 | 78.3 | 62.9%–93% |
| VirginiaVA | 55 | 9.04 | 77.8 | 75.9 | 65.1%–86.8% |
| VermontVT | 30 | 7.61 | 86.2 | 51.7 | 69.4%–94.5% |
| WashingtonWA | 32 | 8.54 | 78.1 | 62.5 | 61.2%–89% |
| WisconsinWI | 31 | 9.01 | 64.5 | 74.2 | 46.9%–78.9% |
| West VirginiaWV | 27 | 11.36 | 77.8 | 77.8 | 59.2%–89.4% |
| WyomingWY | 26 | 15.20 | 61.5 | 61.5 | 42.5%–77.6% |
Methods
Scoring rules were fixed before results.
Each question is answered by 200 simulated people. Each person says how likely each answer is. A small safety rule keeps no option at zero.
Answer order is shuffled per question. We also score a dummy that always picks the first listed option, so a published rate can be read against a mindless rule on the same items.
D.C. is held out of scoring.
A narrower April test, on 5 places with 10,000 people and extra calibration, missed by 7.47 points. That is real. It is not what a standard study in the product runs. Details are on how we measure.