NYC ballot · Nov 3, 2026 · Unofficial election night results
Prop 3 support vs. the protected bike network
New York City passed Proposal 3, a charter amendment to speed up street safety projects, on Nov 3, 2026. This map compares support in each of the city's ~4,000 election districts with the protected bike lanes nearby and the share of households without a car, to show where support outruns the infrastructure.
This page was built with Claude, an AI model made by Anthropic. The data processing, analysis and code were AI-generated under human direction. Figures come from the public sources listed in Acknowledgements; check them against those sources before citing.
Yes
No
Explore by precinct
Every voting precinct in the city, six ways: the support gap (Yes share ranked against the protected bike network within 0.5 km), opportunity (support gap × car-free share × density), and the four inputs on their own: Yes share, bike infrastructure, car ownership and density. Drag to pan, scroll or pinch to zoom, and hover or tap a precinct to see all of its numbers.
Protected/greenway · painted · shared
Opportunity
Support gap × share of households without a car × density factor: where support outruns the bike network, most households don't drive, and many people live.
This layer weights the gap by how many households depend on something other than a car and by how many people a new lane would serve. The density factor is the square root of a precinct's density divided by the citywide median (about 53,000 residents per square mile). A precinct at the median counts 1×, one at four times the median counts 2×, and one at a quarter of the median counts 0.5×; the square root keeps the densest blocks from running away with the score. A precinct whose Yes share ranks 40 points above its bike score, where 75% of households don't own a car and density is at the median, scores 40 × 0.75 × 1 = 30; the same precinct at four times the median scores 60. Precincts where the bike network ranks at or above support have no shortfall and are shown flat; higher scores mean strong support, little protected infrastructure and many car-free households.
Highest-scoring Assembly Districts
Vote-weighted average score across each district's precincts, then car-free share and number of precincts scoring 30 or more.
How to read it
Density raises Manhattan's share of the 250 highest-scoring precincts from 28 to 51, almost all in Harlem and Upper Manhattan. Midtown and downtown already rank higher for protected lanes than for support, so they show no shortfall regardless of density.
Both inputs are relative. The gap compares precincts' rankings, so a high score means "underbuilt compared with the rest of the city," not an absolute shortage.
Car-free share is a five-year survey estimate spread from census tracts, so single-precinct values are approximate; clusters are more meaningful than individual precincts.
Yes share stands in for demand, but Proposal 3 also covered city property and office leasing.
Bike scoring
How each precinct's bike infrastructure score is built, from the NYC DOT bike route file (current segments only).
Method
Rate each side of the street. Every segment lists a facility for each direction of travel. Each direction gets the weight in the table, and the segment's weight is the average of the two. A two-way protected path scores 1.0 per km; a one-way protected lane scores 0.5.
Draw a 0.5 km circle around each precinct. That's roughly a five-minute walk or two-minute ride. The tight radius rewards a dense network: a precinct only scores well if protected routes run close by, not somewhere in the wider neighborhood.
Add up weighted kilometers of every segment inside the circle. The result is the precinct's score, in quality-weighted km.
What's left out
Shared lanes, sharrows, signed routes and wide parking lanes score zero. Riders share them with cars.
Painted lanes count for 5–10% of a protected lane, because nothing separates them from traffic.
Retired segments and exact duplicate geometries are dropped.
The DOT file doesn't say what kind of protection a lane has, so concrete barriers, parked cars and plastic posts all count as "Protected."
Greenways and on-street protected lanes are weighted the same.
Weights and what they add up to
Facility
Weight
Route km
Share of score
Route km counts each direction as half a route, so a two-way facility counts once. Share of score is each facility's share of the weighted total across the whole city. Protected facilities drive almost all of it.
Reading the support gap
Both measures are ranked across 3,920 precincts with at least 50 votes.
The gap is a precinct's Yes-share percentile minus its bike-score percentile. Purple means support ranks higher than the bike network around it; blue means the bike network ranks higher than the support. The two rankings are only loosely related (rank correlation 0.25). Proposal 3 also covered city property and office leasing, and some voters say Yes to ballot questions by habit, so Yes share is an imperfect stand-in for demand for bike lanes.
Support most ahead of infrastructure
Infrastructure most ahead of support
Assembly Districts, vote-weighted average gap. Neighborhood names are approximate.
Car ownership and support
Share of households with no vehicle available, American Community Survey 2017–2021 five-year estimates, by census tract.
Car-free households track support more closely than anything else on this page. Yes share rises steadily with the share of households that don't own a car, and car ownership tracks Yes share (rank correlation 0.48) much more closely than the bike network does (0.25). The bike network itself follows car ownership (0.57): protected lanes are concentrated where fewer people drive.
Households without a car
Precincts
Yes share
Tract estimates are spread onto precincts by area, so a precinct's figure reflects its surrounding tracts and carries the survey's margin of error. Precincts with fewer than 20 estimated households are left blank.
By Assembly District
Each dot is one of 65 Assembly Districts; dot size reflects votes cast. Tap or hover a dot for its numbers.
Lowest Yes share: AD 48 in Brooklyn. Highest: AD 83 in the Bronx.
All districts
Click a column heading to sort. ADs that cross borough lines are combined and listed under the borough with most of their votes.
AD
Borough
Yes
No
Yes %
Acknowledgements
This map is built entirely from public data. Thanks to the agencies and people who publish it. None of them produced or endorsed this analysis.
Election results
NYC Board of Elections, unofficial election night results for Proposal 3, Nov 3, 2026 (as of Nov 4, 12:42 AM), by Assembly District and election district. Totals will change as remaining ballots are counted; certified results come later.
NYC Department of Transportation, New York City Bike Routes, published through NYC Open Data. Current segments only. The quality weights, 0.5 km radius, gap and opportunity scores are this project's own (see Bike scoring and Opportunity).
Car ownership
U.S. Census Bureau, American Community Survey 2017–2021 five-year estimates (vehicles available by household), as processed for 2020 census tracts by the NYC Department of City Planning for Population FactFinder, including DCP's 2010–2020 tract crosswalk.
Population density
U.S. Census Bureau, American Community Survey 2017–2021 five-year estimates (total population), from the same NYC Department of City Planning Population FactFinder extract, mapped onto precincts the same way as car ownership and divided by each precinct's land area.
Census tract boundaries
NYC Department of City Planning, 2010 Census Tracts (release 20A), from the NYC Health Department's public nycehs/NYC_geography repository. Car ownership was mapped onto these tracts with the crosswalk, then spread onto precincts by area.
Inspiration
The CUNY Graduate Center's Center for Urban Research, whose NYC 2026 election map was the starting point for this project.
Typefaces and tools
Source Serif 4, Public Sans and IBM Plex Mono via Google Fonts. Data processed in Python with shapely, pyproj, pyshp and openpyxl. Analysis and map built with the help of Claude, an AI model by Anthropic.