NYC Open Data, NYPD Complaint Data Historic (dataset qgea-i56i), https://data.cityofnewyork.us/Public-Safety/NYPD-Complaint-Data-Historic/qgea-i56i, downloaded Sept. 22, 2026, does some disaggregation of felony assault complaints (most complaints remain unclassified). Felony assaults are offense key code 106. There are also some NYPD internal subcodes. Assaults on police/peace officers are given a subcode of 106 (yes, that’s the same number as the umbrella code).
That subcode is what we used to approximate NYPD’s use of “assaults on police officers,” even though the code also covers firefighters, EMTs and some others. For assaults on other government workers, we used internal subcodes 102 (school safety agent), 103 (traffic agent), and 108 (other public service employees). We do not know if NYPD used a different definition when it referenced “other government workers” in its press release.
Open Data does not break out domestic incidents. For that, we used NYPD’s Domestic Violence Reports, “Complaint and Radio Run Statistics” monthly and quarterly spreadsheets, https://www.nyc.gov/site/nypd/stats/reports-analysis/domestic-violence.page, downloaded Sept. 22, 2026 (2019 data are on the archive page). We aggregated the monthly data as an annual 2019 report was not posted and we wanted to compare apples with apples.
As mentioned in the main text of the article, CompStat’s year-end totals, Citywide Seven Major Felony Offenses 2000–2025, https://www.nyc.gov/site/nypd/stats/crime-statistics/historical.page, freeze annual data for a year as of mid-January of the following year, whereas Open Data’s information continues to be updated. As such, the two sources differ by 200 complaints in 2019 (Open Data is higher). The difference between the two sources in 2025 complaints is, as of the download date, trivial (just three complaints).
The way we get to “all other” (the remainder) is by taking a total and subtracting out the specified subcategories. This is done separately using the CompStat and Open Data annual totals. Because the specified subcategories are fixed and identical under both methods, all of the variation as between CompStat and Open Data annual totals is attributed to remainder. One visible manifestation is that the CompStat and Open Data “all other” rows report different percentages in Table 1.
Open Data’s counts of complaints are of complaints reported in a particular year, not necessarily assaults that occurred in that year.
In checking NYPD’s claims as measured from 2020, we used the same methodology, starting from 2020 instead of 2019, with the Open Data file being downloaded on Oct. 1, 2026, and the 2020 domestic violence files on Sept. 30, 2026.