| title: redrawn | |
| emoji: 🗺️ | |
| colorFrom: blue | |
| colorTo: gray | |
| sdk: static | |
| pinned: false | |
| license: mit | |
| short_description: Redraw the district lines and the answer changes | |
| # redrawn | |
| **Does the map change the answer?** 1.9 million New York City crashes, one | |
| question — do neighbourhoods where speeding is cited more often have more | |
| injurious crashes? — and a different answer from every map of the city. | |
| Between individual crashes the correlation is **+0.055**. Across census tracts | |
| it is +0.262, across police precincts +0.469, across community districts +0.551, | |
| across boroughs **+0.709**. Contiguous equal-size districts drawn deliberately | |
| span **+0.07 to +0.83**. Not one crash record changes; only the boundaries move. | |
| This page re-runs the analysis in your browser. It holds the per-tract crash | |
| counts and the tract adjacency graph and recomputes the zone rates and the | |
| correlation on every interaction, so nothing you see is a stored answer — and | |
| the "draw a new map" button really does build a fresh contiguous partition of | |
| 2,310 census tracts. | |
| Data: [NYPD Motor Vehicle Collisions](https://data.cityofnewyork.us/Public-Safety/Motor-Vehicle-Collisions-Crashes/h9gi-nx95) | |
| and Department of City Planning boundaries, via NYC Open Data. | |
| Method, controls and the full write-up: <https://github.com/UsmarHaider/redrawn> | |