# Questions to ask of NYC Subway Crime Rates and Ridership, 1998–2026

These questions are starting points for exploring the files named under each. The files can describe patterns and comparisons, but do not by themselves establish why a change happened; the card's "Before you use these numbers" (README.md) sets out the limits. The first five also appear on the card.

## 1. Compared with 2019, how have recorded felony rates changed per paid ride and per estimated entry?

The MTA counts paid rides. About 4% of subway entries in early 2019 were unpaid, and about 10% in early 2025. If entries and recorded complaints stayed the same, a higher unpaid share would mean fewer paid rides and so a higher rate per paid ride. Compare the monthly table's rates per paid ride and per estimated entry, then use `subway_ridership_monthly.csv` to move the unpaid share one percentage point above or below the MTA's estimate: the same complaints, two denominators, and a choice about which one to report.

Files: `subway_crime_monthly.csv`, `subway_ridership_monthly.csv`, `fare_evasion_quarterly.csv`.

## 2. Which offences fell with ridership in 2020, and which did not?

In 2020 the subway carried 38% of its 2019 rides. Grand larcenies fell by half, closer to the fall in riders, but felony assaults stayed at 96% of their 2019 number and robberies at 105%. The daily and monthly files let you measure, offence by offence, how closely recorded complaints track ridership, and ask what the offences that held steady have in common.

Files: `subway_crime_annual.csv`, `subway_crime_monthly.csv`, `subway_crime_daily.csv`.

## 3. Which station complexes have the highest major felony complaint rates per entry: the busy hubs, or quieter stations?

Counts and rates rank stations differently. In 2025 the Bronx's station complexes recorded almost three times Manhattan's rate of major felony complaints per entry. The station table pairs matched complaints with MTA station entries for each complex and year. Compare the rankings by count and by rate, then recalculate them without the complaints matched less surely, using `station_match` in the record file.

Files: `subway_stations_annual.csv`, `subway_station_crosswalk.csv`, `subway_complaint_records.csv`.

## 4. Are major felony complaint rates higher at weekends than on weekdays?

From 2023 to 2025 the rate of major felony complaints per ride was about twice as high at weekends as on weekdays. Use the daily file to compare this gap across months and offences. The hourly file shows when complaints began, but there are no hourly ridership figures here, so it cannot show whether rates per ride differ by hour.

Files: `subway_crime_daily.csv`, `subway_crime_by_hour.csv`.

## 5. Theft of services: what do complaints, arrests and the MTA's survey each measure?

Complaints of theft of services, the offence that covers fare evasion, went from 54 in 2019 to 13,009 in 2025. Arrests for it in the Transit jurisdiction went from 2,712 to 12,475 over the same years, yet they were 22,776 in 2014, when complaints numbered 50. The MTA's own survey has the unpaid share rising from about 4% to 10%. Three records of one behaviour move three different ways. Set side by side month by month, they show where and when they diverge; they cannot by themselves divide that divergence between behaviour and record-keeping.

Files: `transit_arrests_monthly.csv`, `subway_crime_monthly.csv`, `fare_evasion_quarterly.csv`, `city_context_monthly.csv`.

## 6. How have transit-system felonies per paid subway ride changed since the 1990s?

Transit-system major felonies per million paid subway rides fell from 5.0 in fiscal 1998 to 1.7 in fiscal 2026. The crime count covers the whole transit system and the denominator subway rides, so compare within this series, not with the subway-premises complaint rate. How does the rise after the pandemic compare with the earlier decline?

Files: `transit_major_felonies_fiscal.csv`, `subway_ridership_monthly.csv`.

## 7. How has the subway's share of the city's recorded major felonies changed?

The subway's share of the city's six major felonies was 2.8% in 2019 and 2.0% in 2025. The context file counts complaints across the whole city, the subway included, the way the subway files count them, so the two can be set against each other month by month.

Files: `city_context_monthly.csv`, `subway_crime_monthly.csv`.

## 8. Did borough differences in major felony complaint rates widen or narrow after 2020?

Group the station complexes by borough, add up their complaints and their entries, then divide; compare the years before and after 2020 and find the stations that moved most. Rates themselves are never added. A complex served by several lines has one count and one denominator, so the files give rates by complex and by borough, not by line.

Files: `subway_stations_annual.csv`, `subway_station_crosswalk.csv`.

## 9. Are recorded assault rates per ride higher on hot days?

Heat is often said to raise violence. The daily file has each day's complaints, rides and Central Park temperature, so the question can be asked per ride rather than per day, with the season held apart from the heat.

Files: `subway_crime_daily.csv`.

## 10. How do rain and snow relate to ridership and recorded complaints per ride?

With daily rides, complaints and precipitation side by side, you can see whether a wet day carries fewer riders, and whether its complaints fall with them, more or less.

Files: `subway_crime_daily.csv`.

## 11. How has the recorded victim profile changed by age group and sex?

The victim file counts complaints by the victim's age group and sex for every offence and year. It also shows how many complaints have no person as victim: offences against the State, such as fare evasion, and against businesses. Without ridership by age and sex, the counts describe who appears in the records, not who is most at risk.

Files: `subway_victims_annual.csv`.

## 12. How did complaint counts by offence code change around changes in the law?

New York widened felony assault to more transit workers in 2022 and 2024, and redefined rape in a law signed in January 2024. The internal codes separate assaults on police and peace officers and on other public service employees, so the codes can be compared before and after each change. They cannot isolate the assaults newly made felonies: newly covered workers can also sit in the general felony-assault code (`pd_code` 109). And until the rape law's effective date is settled, the rape series should not be split at it.

Files: `subway_crime_monthly.csv`, `subway_complaints_by_code_monthly.csv`, `events.csv`.

## 13. Is pickpocketing disappearing from the records?

Grand larceny's internal codes separate pickpocketing, bags opened in a crowd, thefts from sleeping riders and stolen phones and cards. Their mix has shifted since 2014, and the code table follows each one month by month.

Files: `subway_complaints_by_code_monthly.csv`.

## 14. How has the ratio of Transit arrests to recorded complaints changed?

Arrests and complaints are separate records: an arrest is not linked to a complaint, so the ratio is not a clearance rate. Compare arrest and complaint counts by month and offence to see where they move together and where they part.

Files: `transit_arrests_monthly.csv`, `subway_crime_monthly.csv`.

## 15. What changed in recorded complaints and arrests around the policing announcements of 2022, 2024 and 2025?

events.csv dates each announcement. The monthly, daily and arrest files show what moved afterwards. More officers can raise recorded counts as well as lower crime, and other changes fall in the same weeks, so a before-and-after comparison describes, but does not measure an effect.

Files: `events.csv`, `subway_crime_monthly.csv`, `subway_crime_daily.csv`, `transit_arrests_monthly.csv`.

## 16. How do the shelter census and subway complaints move together?

The shelter census nearly doubled between January 2022 and January 2024 and is often brought into arguments about the subway. The context file has its monthly average beside the subway's counts, so the association can be examined. These are city-wide totals, not linked records: they say nothing about what any person in a shelter did.

Files: `city_context_monthly.csv`, `subway_crime_monthly.csv`.

## 17. How did ridership and complaint rates change around the start of congestion pricing?

The congestion toll began on 5 January 2025, and a policing announcement followed eleven days later. Daily rides and complaints before and after, with 2024 as a comparison, describe what changed; they cannot separate the two.

Files: `subway_ridership_monthly.csv`, `subway_crime_daily.csv`, `subway_crime_monthly.csv`, `events.csv`.

## 18. Did the unpaid share move the same way around each fare increase?

The base fare rose to $2.90 in August 2023 and to $3 in January 2026, and the MTA's quarterly estimates run on either side of both. Compare the direction and size of the changes, then test whether the answer depends on the quarters chosen. This describes patterns around the increases, not their effects.

Files: `fare_evasion_quarterly.csv`, `events.csv`.

## 19. How do reporting delays vary by offence and year?

Most complaints are reported the day they happen, but not all. The report-delay file shows, offence by offence, how long the rest took to arrive among the complaints already in the files. The newest years have had less time for the slowest reports, and the file cannot see reports still to come, so it describes delays rather than predicting a month's final count.

Files: `subway_report_delay_annual.csv`, `subway_crime_monthly.csv`.

## 20. How much of the seasonal pattern before 2020 comes from the model?

Monthly rides before March 2020 are modelled from yearly totals. The MTA's station counts, summed, run from February 2017 and show where the model is off: in every November from 2017 to 2019 the counted rides are 3% to 4.5% above the modelled ones. How much does a seasonal analysis change when the counted series is used instead?

Files: `subway_ridership_monthly.csv`, `subway_crime_monthly.csv`.

## 21. How do recorded complaints per ride compare on buses and on the subway?

The context file counts complaints whose premises is a New York City Transit bus, and the MTA's daily bus ridership, from September 2020. The two may not cover exactly the same buses, bus stops and terminals are counted apart, and the subway count covers its premises; compare the rates with those scopes in mind rather than as a verdict on which mode is safer.

Files: `city_context_monthly.csv`, `subway_crime_monthly.csv`.

## 22. What happened to recorded overnight complaints when the subway closed at night?

From 6 May 2020 the subway closed from 1 to 5 a.m.; a shorter closure, 2 to 4 a.m., was announced on 15 February 2021, and 24-hour service returned on 17 May 2021. The share of complaints that began between 1:00 and 4:59 a.m. was 11% in 2019, 4% from June 2020 to January 2021 and 12% in 2022. These are shares of complaints, not rates per ride (there are no hourly rides here). Did the number in those hours fall, and did other hours or stations change? The record file has hour and station together.

Files: `subway_crime_by_hour.csv`, `subway_complaint_records.csv`, `events.csv`.

## 23. Can you tell a felony from a misdemeanour by when and where it happened?

The record file has a row per complaint and more than a dozen categorical columns. Predicting the law category from the time, day, station, borough and transit district is a real classification exercise with uneven classes, and a lesson in leakage: `offence`, `key_code`, `pd_code`, `pd_description` and `is_major_six` give the answer away and must be left out. Decide whether violations are a third class or set aside, hold out the latest months for testing, and keep identical rows: they are separate complaints. It is an exercise in method, not a way to predict anyone's behaviour.

Files: `subway_complaint_records.csv`.

## 24. When are suspect descriptions blank, and when recorded as unknown?

The suspect's age group is blank in the NYPD's file for 33,830 complaints and recorded as UNKNOWN for 29,911 more. Compare blanks, recorded unknowns and filled descriptions by offence, year and station. These comparisons can reveal systematic differences in which descriptions are blank or recorded as unknown; they cannot establish what the missing descriptions would have been, or which way of filling the gaps is right.

Files: `subway_complaint_records.csv`, `missing_values.csv`.

## 25. How wide are the recorded windows in which offences took place?

Most complaints record when the offence began and when it ended; 16,297 record no end date. The interval may describe uncertainty about when the offence happened rather than how long it lasted, and some intervals run for years. Compare window widths by offence and year, and study the complaints without an end date apart from those with one.

Files: `subway_complaint_records.csv`, `missing_values.csv`.

Credit as:

    SparkyData (2026). NYC Subway Crime Rates and Ridership, 1998–2026 (version 2.0). SparkyData additions CC BY 4.0; government-source data under publisher terms; code MIT.
    https://sparkydata.dev/data/nyc-subway-crime-and-ridership