Changes and Corrections

A correction changes figures the site showed; a method change changes how they are counted, ranked or rated. Figures cited from an earlier build can differ from today's: cite the build identifier shown in the footer (see how to cite). The full history of the code and data is in the repository. To report an error, write to [email protected].

1. November 14, 2026Method

   Rates per on-campus student, with intervals; violent and property crimes

   Rates now divide by on-campus FTE: the FTE without NCES's share of students enrolled only in distance education. The 50% online cut-off is gone, and rate rankings list institutions with at least 1,000 on-campus FTE. Grand Canyon University (79% exclusively online in fall 2024), for example, is now listed.

   Rate rankings leave out institutions with a hospital (IPEDS), special-focus medical schools and research institutions (Carnegie), and Gallaudet University, whose campus also holds two schools for deaf children: 1,227 institution-years, among them Ohio State, Michigan and UCLA in 2024. They also leave out 243 institution-years in which DOE reports a campus's figures under two institutions and their students cannot be matched exactly, such as Purdue and Texas A&M in 2024; 44 more are rated with both institutions' FTE.

   Only rates built on at least 20 incidents get a rank; the others are listed as too few to rank by rate, with their rate. Every rate shows a 95% interval and every rank the range of ranks the intervals allow. Murder, negligent manslaughter, incest, statutory rape and arson are ranked by count only.

   New categories: "Violent crimes" and "Property crimes", the offenses of the FBI's violent and property crime (without larceny-theft, which Clery does not collect). Rankings open on violent crimes. A year more than 3 times an institution's usual count is marked as unusually high and stays ranked.

   In 2024 rate rankings list 1,874 institutions (2,077 before). Of the previous top 25 by all criminal offenses, 13 are still in it and 11 are now left out, most of them medical institutions or institutions with a hospital. No count changed.

   Methodology
2. November 14, 2026Method

   Fewer merger years left out of rate rankings

   Rate rankings left out 295 institution-years (97 with at least 1,000 FTE) in which a merger or renumbering might have put a campus's incidents and its students under different institutions. They now leave out only the years where a campus with at least one incident has its incidents under one institution and its students under another, or where it cannot be told which institution's FTE holds them: 91 institution-years (29 with at least 1,000 FTE). Tufts 2015, the University of Arizona 2020 and Northeastern 2023, for example, have rates again.

   31 institution-years are now rated with the FTE of an institution whose campus figures DOE reported under them, such as Tufts 2015 with the School of the Museum of Fine Arts. No count changed.

   Methodology
3. November 14, 2026Correction

   Campus names, renumbered campuses and mergers

   Branch campuses are now titled with their own DOE campus name, with the institution under it (6,705 school pages). Search also finds campuses by city and by names DOE and IPEDS used for them before.

   When DOE moved a campus to a new ID (a renumbering or a merger), its years were split across two pages. 313 such pairs, each reviewed, now link to each other; each page keeps the years DOE reported under its own ID.

   Where both IDs reported the same full year from different DOE releases, those figures were counted twice in state and national totals. They now count once, under the newer ID: 86 campus-years, 661 incidents across 2015-2023. No campus's own figures changed.

   In a year a campus was reported under two institutions, or a merged institution's enrollment covered campuses counted elsewhere, incidents and enrollment may not cover the same campuses. Those institution-years have no rate and are left out of rate rankings; 78 of them were in the rate rankings before (for example Vermont State 2022, the Connecticut State colleges 2022, and Penn State 2019 and 2021).

   Methodology
4. November 14, 2026Site

   Definitions, limitations, citation and downloads

   The Methodology page defines every category and location from the regulation (34 CFR 668.46 and its Appendix A), with sources, and lists the limitations of the data.

   Charts shade 2020 and 2021, when most instruction was remote, and link to the note. School pages say when a residence-hall figure comes from another DOE release because of a known error in DOE's file.

   The pages now say that counts are by the year a crime was reported, not the year it happened. The figure that 80% of crimes go unreported was replaced by what the Bureau of Justice Statistics measured: 80% of rape and sexual assault victimizations of female college students aged 18-24 in 1995-2013 were not reported to police.

   School pages call the top category the most common criminal offense, show year-over-year change in neutral colors, and label enrollment as the whole institution's FTE.

   New: an APA and BibTeX citation with the build identifier, CSV downloads of every figure, and this page. No figure changed.

   Methodology
5. November 12, 2026Method

   Pandemic years judged by the 2019 online share

   Rate rankings leave out institutions where at least half of students study exclusively online. For 2020 and 2021 the test now uses each institution's fall 2019 share, because temporary remote teaching did not make schools online schools. The 2020 rate ranking now lists 2,168 institutions (was 1,588) and 2021 lists 2,167 (was 1,936).

   Methodology
6. November 12, 2026Correction

   Years reported with no incidents shown as 0

   A year a campus reported with no incidents used to look the same as a year it did not report, and campuses that never had an incident had no page. Such years now show 0, every campus that reported has a page, and rankings list campuses with 0 after the others, sharing one rank.

7. November 12, 2026Method

   Rates per institution; mostly online institutions left out

   NCES publishes enrollment per institution, not per campus, so a branch campus's rate divided its own incidents by the whole institution's enrollment. Rates are now per institution: the incidents of all its campuses over its FTE. Rankings by count still list each campus.

   Rate rankings list institutions with at least 1,000 FTE students and under 50% of students exclusively online. School pages still show other institutions' rates, with a note.

   Methodology
8. November 12, 2026Method

   Ties share a rank; criminal offenses by default

   Tied schools now share a rank and the next rank skips (1, 2, 2, 4). Rankings, school pages and Compare open on all criminal offenses; "All categories" (which adds arrests and disciplinary referrals, counts of people) stays selectable.

   Methodology
9. November 10, 2026Correction

   NCES's own enrollment for 2015-2020

   For 2015-2020 the site estimated FTE with one formula for every school. That matched NCES's published FTE exactly for only 58% of institution-years and was off by more than 5% for about 830 a year. Every year now uses NCES's own FTE, which changes rates for those years.

   Methodology
10. November 10, 2026Correction

    Campus-years missing from the newest DOE release restored

    Each year used to come only from the newest DOE release covering it, so a campus missing from that release lost the year. Each campus-year now comes from the newest release that has figures for it. This restores 1,551 campus-years at 1,126 campuses (28,695 incidents, 2015-2023).

    Two known errors in DOE's 2021-2023 release (residence-hall disciplinary referrals and arrests) are worked around; school pages name the campuses affected.

    Methodology
11. November 10, 2026Correction

    Figures two campuses report identically counted once

    Some campuses report exactly the same figures for a year and location as another campus in the same state. State and national totals and rankings now count them once, under one campus (242 location-year blocks), and also when a whole year is copied or the copy is in an older DOE release (141 more blocks). School pages still show the figures as reported and name the other campus.

    Methodology
12. November 10, 2026Correction

    Alabama 2015 and 2016 restored

    A file reader misread the first rows of DOE's 2015 and 2016 files, so 13 Alabama campuses, including the University of Alabama, Auburn and UAB, had no 2015 or 2016 data. Alabama's 2015 total was shown as 1,540; it is 4,646.

13. November 10, 2026Site

    Analytics disclosed; sorting covers the whole list

    The Privacy page said the site had no analytics; collegesafetydata.org uses Cloudflare Web Analytics (cookieless, aggregate), and the page now says so. Sorting a rankings column used to sort only the rows loaded so far; it now sorts the full list.