← New Jersey Transit Score

How the updated Transit Score works

2024 ACS household estimates · 2023 workplace jobs · 2050 land-use scenarios

This analysis revisits the land-use question behind NJ TRANSIT’s 2000 Transit Score: where can the concentration of homes and jobs support transit? It uses a revised two-factor index for 2,175 New Jersey census tracts, with modeled service available as a separate comparison.

Two land-use inputs, one transparent index

The current index uses occupied households from the 2024 ACS five-year estimates and workplace jobs from 2023 Census LODES. Both densities divide by tract land area, excluding water.

50 × log₂(1 + occupied households per land acre ÷ 3)
+ 50 × log₂(1 + workplace jobs per land acre ÷ 4)

The household and employment components have equal weights. The reference densities are 3 households and 4 jobs per land acre; a tract at both reference values scores 100. The logarithmic transformation limits the influence of exceptionally high density. These are explicit modeling choices, rather than coefficients fitted to observed ridership.

The formula reproduces every current score in the published extract. The score is continuous and has no fixed upper limit. It describes the concentration of origins and destinations; it is not a percentage or a forecast of transit riders. Vehicle ownership, observed transit use, and scheduled service remain separate measures.

Contemporary land-use screening bands
BandCurrent index
LowBelow 50
Emerging50 to below 100
Supportive100 to below 150
Strong150 to below 200
Very Strong200 or above

These bands communicate the index; they do not establish minimum service levels or determine which transit project to build. All stored current band labels agree with these boundaries.

2050 views and service comparisons

The trend and concentrated-growth views show alternative 2050 land-use scenarios on the current index’s scale. Their values describe development patterns, not future transit schedules or committed projects.

The optional rank difference subtracts weekday-service percentile from land-use percentile. A positive value means a tract’s land-use rank exceeds its recorded service rank. This is a screen for investigation: neither sign determines appropriate service levels. Walking assumptions, geography, operator coverage, and the chosen service measure affect the result.

Updating the framework, changing the method

NJ TRANSIT’s October 2000 report scored 1,950 tracts using forecasts for 2020. The inputs covered population, households, employment, and households with zero or one car. It connected five score categories with preliminary service and investment criteria.

The revised index retains the statewide land-use screening purpose while changing the equation. Household and job density now form the core score; population and vehicle availability are not additional score components. Its logarithmic scale and contemporary bands differ from the original additive formula and thresholds.

Consequently, the old and new scores are not a numerical time series. Comparing the original forecast with observed outcomes would require the original tract scores, a geographic crosswalk, and the same formula applied consistently. This page does not claim that retrospective test.

The printed original formula on page 20 divides its residential subtotal by three. The worked examples on pages 21–22 omit that division. The historical record is preserved with that discrepancy; the current index uses the explicitly stated two-factor equation above.

Reading map colors

Current and 2050 land-use views use a common scale so a color keeps the same meaning across scenarios. The five contemporary screening bands above belong to the revised index and cannot be substituted for the original 2000 categories.

Service views retain their own units. Weekday and Saturday departures use the same boundaries of 5, 20, and 80. Coverage uses 25%, 50%, and 75%; useful hours use 3, 9, and 15. Changing the county does not rescale the values. A missing value remains distinct from a measured zero.

Walking reach, frequency, and duration

The documented walking thresholds are 1,312 feet (about ¼ mile) for bus and 2,625 feet (about ½ mile) for rail, light rail, rapid transit, and ferry. The longer-walk view uses 2,625 feet for bus and 3,937 feet (about ¾ mile) for other modes. Distances follow a pedestrian network.

The comparison dates are Tuesday, August 18, and Saturday, August 15, 2026, from 6 a.m. to midnight. The extract does not establish whether those dates represent other seasons or Sundays.

Reachable departures
The resident-weighted mean number of unique vehicle trips reachable from origin blocks. A vehicle trip is counted once per origin block even when several stops are reachable. This measures service opportunities near residents’ homes. It does not establish a particular route’s frequency or destinations reachable.
Recorded 15-, 30-, and 60-minute coverage
The resident share satisfying the model’s frequency classification. The precise qualifying-hour and route-grouping rules have not been reproduced for this published extract. A coverage percentage must not be read as the share with that frequency throughout the day.
Recorded useful hours
The resident-weighted average number of weekday hours classified as meeting a frequency threshold. Pair this with coverage: broad coverage can last for a short part of the day. Qualifying hours may be separated; the average does not describe an unbroken service span or every resident’s experience.
Rail coverage
The resident share with at least one reachable weekday rail, light-rail, or rapid-transit trip under the walking model.
Saturday change
Absolute change equals Saturday minus weekday departures. Percentage change divides that difference by the weekday value. A zero weekday baseline has no defined percentage change and is shown separately.

Walking reach does not establish step-free access. Elevator availability, sidewalk condition, fares, reliability, capacity, transfers, and journey times fall outside this extract.

People and tracts answer different questions

Saturday has fewer recorded reachable departures in 1,654 of 2,172 measured tracts: 76.2%. There are 151 tracts with a positive weekday value and a zero Saturday value. Three additional tracts have no service estimate.

The resident-weighted average falls from 61.22 weekday departures to 39.81 Saturday departures, a 35.0% reduction. This gives more weight to places where more people live. It is an average of modeled opportunities, rather than a count of statewide vehicle trips. Individual changes vary.

Resident-weighted average =
sum(tract population × tract resident-weighted departures)
÷ sum(tract population)

Both days use the same 2,172 tracts and a population denominator of 9,343,809. Weights are documented as 2020 block population scaled to 2024 ACS tract totals. Uniform scaling cannot locate population shifts within a tract after 2020.

Download statewide and county comparisons (CSV). A count of people living in tracts with lower Saturday averages does not establish how many individuals lose access.

What the source record establishes

The revised land-use equation and its current band labels reconcile to the published tract values. The downloadable audit checks those calculations, service-summary arithmetic, and data integrity. A complete input manifest and executable run producing these exact service estimates have not been established; individual walking paths and schedules still require source validation.

InputDocumented basisUnresolved for this extract
SchedulesAugust 2026 GTFS schedules; NJ TRANSIT is a named source.Exact feed hashes and complete included/excluded operator list.
Population and households2024 ACS tract values and 2020 Census blocks.Matched input files and uncertainty estimates.
Workplace jobs2023 LODES jobs. The exported density field retains the legacy name job_density_2024.Matched input file and original land-area calculation.
Walking networkOpenStreetMap and documented walking limits.Exact snapshot and each tract’s connected population share.

Reading a zero

Zero is a recorded output. It can reflect limited service, walking-network limitations, or incomplete operator coverage; this extract cannot separate those causes. A blank means no estimate was supplied. Source validation remains unresolved for each tract.

PATH, PATCO, ferries, private operators, and local shuttles need to be checked against the original run before claiming complete statewide coverage. A geographic line overlay does not establish that its schedules entered the calculation.

Source-coverage register · Version and evidence manifest

Geographic context and place search

These reference layers help locate a tract. They do not alter the service estimates or establish which operators entered the August 2026 model.

Municipalities: 564 municipal areas from the 2024 Census county-subdivision boundaries. The displayed outlines are simplified by 75 feet. A tract appears in a municipality’s selector when at least 1% of its polygon area overlaps that municipality or its interior representative point lies inside it. This geographic lookup does not calculate municipal population or service totals.

Major roads: 2,859 features from the 2024 Census primary and secondary road file for New Jersey, simplified by 60 feet for display. This reference layer does not depict walking paths or establish pedestrian access.

Rail-station landmarks: 234 points from the New Jersey Department of Community Affairs passenger-station inventory, whose metadata is dated December 22, 2021. Names and points serve as landmarks. They do not confirm current station operation, accessible entrances, or August 2026 service. The source may omit stations added after its date.

Reproduce the published comparisons

The audit package contains the unchanged tract extract, a Python script using only the standard library, audit results, county summaries, and source-coverage records. Run python3 audit-data.py beside tracts.geojson to recreate the summaries and 14 integrity checks.

Download the data and audit package · Audit results · Tract GeoJSON · Audit script

Full source-to-output reproduction requires a matched feed inventory, pedestrian-network snapshot, origin connections, and metric implementation. A later recovered analysis has different outputs and is not proof that this extract has been reproduced.

Primary source documentation

NJ TRANSIT schedule feeds · GTFS specification · ACS estimates and uncertainty · Census LODES · OpenStreetMap

State and county context uses 2024 Census cartographic boundaries. Municipality labels supply geographic context and do not replace tract boundaries. For travel today, consult current operator schedules.

Methods and corrections policy