Data Methodology

FTE is the denominator of the per-student rates: full-time students count fully and part-time students count in part, so schools of different sizes and mixes can be compared. We use the FTE that NCES publishes for each institution and fall, for every year; we do not calculate it.

FTE (DRVEF variable FTE, "Full-time equivalent fall enrollment")

NCES adds the institution's full-time fall enrollment to an estimate of the full-time equivalent of its part-time enrollment: part-time students multiplied by factors that depend on the institution's control and level and the student's level (for example 0.403543 for part-time undergraduates at public 4-year institutions, 0.335737 at public 2-year institutions). It counts students enrolled for credit in the fall, not over the whole year, and it is published for each institution, not each campus.

PCTDEEXC ("Percent of students enrolled exclusively in distance education courses")

Students enrolled only in distance education courses, divided by all students enrolled, as of the institution's official fall reporting date or October 15. Rates use it to estimate on-campus FTE (see Rate Rules).

Source: NCES, DRVEF data dictionary for 2023; see also the IPEDS Glossary.

Earlier versions of this site estimated 2015-2020 FTE with one formula for every school (part-time undergraduates × 0.403543, part-time graduate students × 0.361702). That matched NCES's figure exactly for only about 58% of institutions, and was off by more than 5% for about 830 institutions a year, so it is no longer used.

Rate Rules

Every rule that rates and rate rankings follow, with its threshold. None of them changes a count or a ranking by count. School pages and Compare show the rate of an institution under the floor of rule 2 or with fewer than 20 incidents, with a note; for an institution-year left out by rule 3, 8 or 9 they show the counts and a note saying why, and no rate. The numbers cover 2015-2024; an institution-year is one institution in one year.

1. Denominator: on-campus FTE. On-campus FTE = FTE × (1 − PCTDEEXC / 100), rounded to a whole number: the institution's FTE without NCES's share of students enrolled exclusively in distance education, because Clery figures cover the campus. It is an approximation: PCTDEEXC is a share of fall headcount, not of FTE, and a student who takes some courses online counts as on campus. For 2020 and 2021, when many institutions taught remotely for the pandemic, it uses the institution's fall 2019 share, so those years divide by the same kind of enrollment as the others (the year's own share when NCES has no 2019 figure: 66 institutions in 2020, 102 in 2021). Every institution with an NCES FTE in these years has a PCTDEEXC. This replaces a cut-off that left out institutions with 50% or more of students exclusively online and divided by their whole FTE below it.
2. At least 1,000 on-campus FTE. Rate rankings list an institution only in a year its on-campus FTE (with any FTE added under rules 8 and 9) is at least 1,000, because with fewer students a few incidents move the rate far. In 2024, 1,987 institutions reach it.
3. Campuses whose population is not mostly students. A rate per student says little where many people on campus are not in FTE. Rate rankings leave out an institution in a year IPEDS reports that it has a hospital (HD variable HOSPITAL = 1: patients, visitors and hospital staff; 882 institution-years, 701 of them with at least 1,000 on-campus FTE, 71 in 2024); else one whose Carnegie basic classification in IPEDS is "Special Focus Four-Year: Medical Schools & Centers" (code 25 of C15BASIC, C18BASIC and C21BASIC) or the 2021 category "Special Focus Research Institution" (C21BASIC 27; 20 of its 22 institutions in 2024 were medical schools and centers in the 2018 classification): 335 institution-years, 205 with at least 1,000 on-campus FTE, 21 in 2024; and the institutions of a reviewed list with a sourced reason, now only Gallaudet University for 2015-2024, whose campus also holds the Clerc Center's Kendall Demonstration Elementary School and Model Secondary School for the Deaf, whose pupils are not in its FTE (Laurent Clerc National Deaf Education Center; 10 institution-years). In 2024 the median rate of all criminal offenses per 10,000 on-campus FTE was 80.2 at research universities with very high research activity (Carnegie R1) with a hospital and 34.6 at those without one. Limitation: IPEDS reports HOSPITAL = 1 only when the institution itself runs the hospital. A university whose medical center is a separate legal entity (for example Johns Hopkins University) is not caught by this rule and stays ranked, so its rate may also include incidents involving hospital patients, visitors and staff. Sources: IPEDS HD data dictionary, Carnegie Classification, 2021 Basic Classification.
4. At least 20 incidents for a rank. A rate gets a rank only when it is built on at least 20 incidents. Taking a year's count as a Poisson count, its relative standard error is 1/√count: 22.4% at 20, 31.6% at 10, 50% at 4. CDC WONDER flags rates based on fewer than 20 deaths as unreliable for the same reason (a relative standard error of 23% or more; CDC WONDER). Institutions with 1 to 19 incidents follow the ranked ones as "too few incidents to rank by rate", in order of rate, with the rate and its interval shown. In the 2024 national rate rankings (1,872 institutions), 310 are ranked for all criminal offenses, 84 for violent crimes and 158 for property crimes.
5. Counts only for rare offenses. An offense with fewer than 1,000 incidents a year on average across all campuses has rankings by count only: incest (2.5 a year), negligent manslaughter (6.2), murder (48.8), statutory rape (78.7) and arson (654.9). Across 2015-2024 only 7 institution-years in rate rankings reached 20 incidents of any of them (statutory rape 3, arson 4), so a rate ranking would rank almost no one. The next rarest offense, robbery, averages about 2,200 a year.
6. 95% intervals and rank ranges. Each rate is shown with a 95% interval: the exact Poisson interval of its count (Garwood, 1936), divided by the on-campus FTE and multiplied by 10,000. For a count of 20 the interval of the count is 12.2 to 30.9, for a count of 0 it is 0 to 3.7. It shows how much a count varies by chance from year to year; it does not cover differences in reporting or classification. A ranked institution's rank range runs from 1 plus the number of ranked institutions whose interval lies entirely above its own, to the number of ranked institutions minus the number whose interval lies entirely below it. Ranks inside one range are not clearly different. In 2024 the median ranked institution could hold any of 48 ranks (of 84) for violent crimes and 155 (of 310) for all criminal offenses.
7. Violent and property crimes. "Violent crimes" adds murder (non-negligent manslaughter included), rape, robbery and aggravated assault, the four offenses of the FBI's violent crime (FBI, Crime in the United States 2019: Violent Crime). "Property crimes" adds burglary, motor vehicle theft and arson, the FBI's property crime offenses without larceny-theft, which Clery does not collect (FBI, Property Crime). Fondling, incest, statutory rape and negligent manslaughter are in neither; all 11 offenses are in "All criminal offenses", which is still offered. Rankings open on violent crimes because motor vehicle theft was 37% of all criminal offenses in 2024 (15% in 2015) and so decides much of that total's order.
8. Mergers and renumbered campuses. In a year DOE reported a campus under two institutions (its old and its new ID, from different releases), the institution whose NCES FTE holds the campus's students is the only one of the two with an FTE or, when both have one, the one whose every campus that year is such a linked campus. When the other institution counts the campus and the students' institution counts no campus of its own that year, the other's rate adds the students' institution's on-campus FTE (both institutions' students over both institutions' incidents). An institution gets no rate that year, and is left out of rate rankings with a note on its school page, only when a campus with at least one incident that year has its incidents counted under one institution and its students under another, when it cannot be told which FTE holds the students, or when those students' FTE would be split among several institutions. In a year a merged institution's campuses count under several institutions and one of them has no NCES FTE while counting a campus with incidents that no institution with an FTE also counts, every institution of the group is left out. A campus with no incident that year needs none of this: whichever FTE holds its students, every rate is the same. In all, 91 institution-years are left out by these rules (27 with at least 1,000 on-campus FTE; before this rule was narrowed, 295, 97 of them with at least 1,000 FTE), and 31 are rated with another institution's FTE added.
9. Shared campuses. When DOE publishes the same figures for campuses of two institutions in a year, they count once, under the host campus, while each institution's students are in its own FTE. When the other institution counts no campus of its own that year and its figures count under one host institution, the host's rate adds the other's on-campus FTE (44 institution-years). Otherwise neither rate can be made exact: adding the other institution's FTE would add its students at its other campuses, and not adding it would leave its students on the shared campus out. Then the host institutions and the other institution are left out of rate rankings that year, with a note: 243 institution-years, 209 with at least 1,000 on-campus FTE, 21 in 2024, among them Texas A&M University, Purdue University, Indiana University Bloomington and the University of Maryland, College Park. Most are a program of one institution on another's campus, such as the Indiana University School of Medicine at Bloomington or the University of Maryland Global Campus at College Park.
10. Unusually high years. A rate is marked "Unusually high for this institution; DOE counts crimes in the year reported" when its count is at least 20 and more than 3 times the median of the institution's counts of that category in its other years (a year without one counts as 0), with at least 3 other years. It stays ranked: a year's figure is the crimes reported that year, and one report of many past incidents can raise one year (see How Institutions Count). In the 2024 national rate rankings 21 institutions are marked for all criminal offenses, 10 for violent crimes and 33 for property crimes.

Rules 3, 8 and 9 leave out 1,546 institution-years (1,139 with at least 1,000 on-campus FTE). Rate rankings, state and national, list 2,153 institutions in 2015, 2,106 in 2016, 2,092 in 2017, 2,057 in 2018, 2,025 in 2019, 1,973 in 2020, 1,945 in 2021, 1,822 in 2022, 1,834 in 2023 and 1,874 in 2024 (2,077 in 2024 under the earlier rules).

Every category on this site is a Clery Act category, counted by each institution and reported to the U.S. Department of Education (ED) in its yearly Campus Safety and Security survey. The definitions below are the regulation's, shortened: the crimes in Appendix A to Subpart D of Part 668 (taken from the FBI's Uniform Crime Reporting manuals), and the VAWA offenses and locations in 34 CFR 668.46. Where the DOE data uses another name, it is given in brackets.

Locations (Clery geography)

How institutions count

• Reports, not convictions. A crime is counted when it is reported to a campus security authority or to local police and occurred on the institution's Clery geography (34 CFR 668.46(c)(1)). Only sworn or commissioned law enforcement personnel may determine a report false or baseless ("unfounded"); unfounded reports are left out of the counts and reported separately (668.46(c)(2)(iii)). This site does not show unfounded reports.
• By the year reported, not the year it happened. Statistics are disclosed by "the year in which the crimes were reported" (2016 Handbook, chapter 2). A crime reported in 2019 counts in 2019 even if it happened years earlier, so a year's figure can jump when many people report past abuse at one institution. For example, the national number one campus by count of criminal offenses was Ohio State University in 2018, 2019 and 2021 (mostly fondling) and the University of Michigan in 2020 (fondling and rape), and Michigan State University's 1,012 rapes in 2018 are about 12% of that year's national total. These are the figures the institutions reported; this site does not mark or remove them.
• One offense per incident, with exceptions. When several criminal offenses happen in one incident, only the most serious is counted (the FBI's Hierarchy Rule, 668.46(c)(9)(v)). Arson is always counted, and a rape, fondling, incest or statutory rape in the same incident as a murder is counted along with the murder (668.46(c)(9)(vi)-(vii)). The Hierarchy Rule does not apply to VAWA offenses, so an incident can be counted both as, say, an aggravated assault and as dating violence (2016 Handbook, chapter 3).
• Arrests and referrals count people. The arrest figures count people arrested (the 2016 Handbook defines an arrest as a person processed by arrest, citation or summons). The disciplinary figures count people "not included" in the arrest count who were referred for campus disciplinary action (668.46(c)(1)(ii)), so a person both arrested and referred for the same violation is counted once, as an arrest. A referral counts only for a violation of the law: breaking a campus rule alone (for example, a student of legal drinking age on a dry campus) is not counted (2016 Handbook, chapter 3). These counts are not subject to the Hierarchy Rule and are counted in addition to any criminal offense in the same incident.
• Stalking is counted per year. Stalking that continues into another calendar year is counted in each year in which the course of conduct was reported (668.46(c)(6)).

What this site does not show

DOE also publishes hate crimes, unfounded crime reports and fire statistics; this site does not read or show them. Hate crimes are a breakdown rather than a separate total: the criminal offenses above that were hate crimes (already in those counts, so adding them would count them twice), plus four crimes counted only when they are hate crimes: larceny-theft, simple assault, intimidation and destruction/damage/vandalism of property (668.46(c)(1)(iii)). The site has not added that breakdown; use DOE's data tool for it. Unfounded reports are the reports withheld from the counts.

Totals on this site

Changes in definitions, reporting or circumstances between 2015 and 2024 that affect comparisons over time. Notes marked with years also appear next to charts and rankings that show those years.

Context for every year

What these figures can and cannot support. Read these before comparing schools or years.

What this site corrects, and what it does not

• One DOE release per campus-year. Each campus-year comes from the newest DOE release that reports it, with all of its figures from that release, so a campus missing from the latest release keeps the years an earlier release published.
• Figures reported by two campuses count once. When campuses in the same state report exactly the same figures for a location (or a whole year), totals and rankings count them once, under one host campus. School pages still show each campus's figures as DOE publishes them.
• Known errors in DOE's 2021-2023 release. Its residence-hall disciplinary file repeats the on-campus figures for 2022 and 2023, and its residence-hall arrest file has no 2022 liquor-law column. For campuses whose figures come from that release, 2022 values are taken from the 2020-2022 release and 2023 residence-hall disciplinary referrals are left out (see Comparing Years).
• DOE's 2015 and 2016 files are read correctly. Those SAS files are read with a reader that returns every row; an earlier reader lost 13 Alabama campuses' 2015 and 2016 figures.
• Enrollment is NCES's own FTE for every year, never an estimate.
• Not corrected: the figures themselves. This site does not audit institutions or change what they reported; it publishes DOE's figures with the rules above. An implausible figure here is an institution's report as DOE publishes it: ask the institution, or check its Annual Security Report. See also the list of corrections to this site's data.

Build identifier

This build is b209c8dce7164eaeaa18468a22f2d540ca78b67e6dac356926965e7aa9a38776, built November 14, 2026. The identifier is the SHA-256 of the build record, which lists a checksum of every source file the build read and every data file it published; the footer shows its first 12 characters. Rebuilding from the same source files gives the same figures, but the build date is part of the record, so every run has a new identifier. Cite it with the date you accessed the data.

audit.json

The result of the build's audit (below): overall_status, each check with its details, coverage counts by year, and audited_build.site_record_sha256, the identifier of the build it checked.

campus-years.csv

One row per campus and year the site shows (each year a campus reported to DOE, zeros included): year, unitid (DOE campus ID), base_unitid (the IPEDS institution: the first six digits), campus_name (the campus's own DOE name, or the institution's when DOE gives none), institution_name, city, state, main_campus, doe_release (the DOE file suffix the figures come from, e.g. 222324 for the 2022-2024 release), counts_in_rankings (false when every figure the campus reported that year is counted under another campus), institution_fte (NCES FTE of the whole institution), institution_online_pct (percent of students enrolled exclusively in distance education), online_pct_fall (the fall that share is from), institution_rate_fte (the denominator of the institution's rates: its on-campus FTE, with any FTE added under Rate Rules 8 and 9) and institution_rate_rule (the rule that leaves the institution out of rate rankings that year: hospital, medical or listed for rule 3, merger or family for rule 8, shared for rule 9; empty when none).

counts-YYYY.csv, one per year

Every nonzero count, by campus, location and offense, as the campus reported it: year, unitid, geo (On-campus, Residence halls, Non-campus, Public property), offense, count, counted_under_unitid (set when another campus, or the same campus under its new DOE ID, reported the same figures for that location; totals and rankings count them under that campus).

2015201620172018201920202021202220232024

Reproducing the site's totals

From counts-YYYY.csv: leave out Residence halls (part of on-campus), keep rows with an empty counted_under_unitid, and, for national totals, keep campuses whose state in campus-years.csv is one of the 50 states, DC, PR or VI. The site's JSON files (school pages, rankings) are under /data/ and listed in the build record.

Rebuilding from DOE's and NCES's files

The pipeline downloads the source files and builds every file above. With Python 3.11 or later, in etl/ of the repository:

python -m venv venv
venv\Scripts\activate        # macOS/Linux: source venv/bin/activate
pip install -r requirements.txt
python 01_download_raw_data.py
python 01b_download_ipeds.py
python 02_transform_to_parquet.py
python 02b_transform_ef_enrollment.py
python 02a_transform_enrollment.py
python 03_build_dimensions.py
python 04_create_aggregates.py
python 05_generate_json.py
python 06_validate.py
python 07_comprehensive_audit.py

The README has the details, including the optional step that re-extracts NCES's 2015-2020 FTE.

The 2016 Handbook was rescinded in October 2020 and replaced by the Clery Act Appendix for the FSA Handbook; ED keeps the Handbook online for historical purposes. ED said at the time that no statutory or regulatory requirement for Clery Act reporting had changed. The definitions above are the regulation's, set by the VAWA final rule that took effect on July 1, 2015; eCFR's version history of 34 CFR 668.46 and its Appendix A, which begins in December 2016, shows no substantive amendment since. The Handbook is cited only where it explains how to count.

• U.S. Department of Education, Office of Postsecondary Education. Campus Safety and Security (data and downloads)
• U.S. Department of Education. Campus Security (guidance, Appendix and the 2016 Handbook)
• Electronic Code of Federal Regulations. 34 CFR 668.46, Institutional security policies and crime statistics
• Electronic Code of Federal Regulations. Appendix A to Subpart D of Part 668, Crime Definitions in Accordance With the FBI's UCR Program
• U.S. Department of Education, Office of Postsecondary Education. The Handbook for Campus Safety and Security Reporting, 2016 Edition (rescinded 2020, kept for historical purposes)
• U.S. Department of Education. Clery Act Appendix for FSA Handbook (October 2020), which replaced the 2016 Handbook
• U.S. Department of Education, Federal Student Aid. Rescission of and Replacement for the 2016 Handbook for Campus Safety and Security Reporting (Electronic Announcement, October 9, 2020; updated January 19, 2021)
• Federal Register. Violence Against Women Act final rule, 79 FR 62752 (October 20, 2014), effective July 1, 2015
• U.S. Department of Education, Federal Student Aid. Additional regulatory flexibilities related to Coronavirus (COVID-19) (Electronic Announcement, July 10, 2020)
• National Center for Education Statistics. Integrated Postsecondary Education Data System (IPEDS)
• National Center for Education Statistics. IPEDS DRVEF (Derived Fall Enrollment) data dictionary, 2023 (FTE and PCTDEEXC)
• National Center for Education Statistics. IPEDS Glossary
• National Center for Education Statistics. Fast Facts: Distance learning
• National Center for Education Statistics. IPEDS HD (Directory Information) data dictionary, 2024 (HOSPITAL, C21BASIC)
• Carnegie Classification of Institutions of Higher Education. 2021 Basic Classification
• Gallaudet University. Laurent Clerc National Deaf Education Center
• Centers for Disease Control and Prevention. CDC WONDER, Underlying Cause of Death help: unreliable rates
• Biometrika. Garwood, F. (1936). Fiducial limits for the Poisson distribution. Biometrika 28(3/4), 437-442
• Federal Bureau of Investigation, Uniform Crime Reporting Program. Crime in the United States, 2019: Violent Crime
• Federal Bureau of Investigation, Uniform Crime Reporting Program. Crime in the United States, 2019: Property Crime
• Bureau of Justice Statistics. Rape and Sexual Assault Victimization Among College-Age Females, 1995-2013 (NCJ 248471, December 2014)