Aggregate Data
Archive Statistics
3,803 verbatim alerts · 4,392 alert messages · 539 institutions · 1969–2026
Two patterns dominate this archive. First, roughly 29% of cases occur in the August–October return-to-campus window, a seasonal cluster driven by move-in incidents, opening-week events, and fall sports. Second, verbatim confirmation climbs sharply after 2018: among alerts dated 2018 and later, 87% are word-for-word confirmed as Twitter/X archive recovery and university alert-page indexing finally make exact transmitted text recoverable.
The charts below cover incident type, year, state, institution type, resolution, and Clery category. Above all, they show how much of the archive we have recovered word-for-word. This archive is about the exact wording and the sequence of the messages, not response-time analytics; counts here reflect what is documented and findable, not real-world frequency.
About these numbers
This archive is a limited, AI-assembled sample of campus emergency alerts, not a complete or randomly drawn census. Every chart below describes what has been documented here so far, which is shaped by what sources are publicly findable and what has been added to date. Please read these as patterns within this archive, not as statistically representative measures of US campuses overall, and avoid drawing population-level conclusions from them. Spot something wrong or missing? You can help correct or expand it.
Looking for more than counts? The Dataset page is the archive’s citable landing: versioned downloads, the archive-level citation and codebook, and the collection’s scope and limits.
Tool · Message structure
Message Element Analyzer
Go beyond the counts: see which of the six warning-message elements (source, hazard, location, guidance, time, impact) each verbatim first alert actually includes.
50 states · 4 territories
avg messages in a sequence
1969–2026
Geographic Coverage
Cases Across All 55 Jurisdictions
50 U.S. states, Washington D.C., and 4 territories, every corner of American higher education.
How to read this: Tile shading shows the cases documented here per state, which tracks collection effort and findability, not per-capita risk or any state's true number of incidents.
55 states & territories documented · stronger blue = more cases · tap any tile to see its cases
| State/Territory | Cases |
|---|---|
| PA | 250 |
| NC | 148 |
| CA | 137 |
| NY | 134 |
| SC | 104 |
| TX | 98 |
| IL | 77 |
| MN | 77 |
| MA | 75 |
| VA | 72 |
| FL | 68 |
| RI | 63 |
| OH | 53 |
| LA | 53 |
| IN | 45 |
| AZ | 40 |
| MI | 38 |
| GA | 38 |
| DC | 36 |
| NJ | 29 |
| WA | 28 |
| MS | 27 |
| CO | 26 |
| IA | 23 |
| MD | 21 |
| AL | 21 |
| KY | 20 |
| OR | 19 |
| OK | 18 |
| WI | 17 |
| TN | 17 |
| MO | 16 |
| NE | 13 |
| NH | 12 |
| NM | 11 |
| CT | 11 |
| ME | 10 |
| NV | 10 |
| HI | 10 |
| UT | 9 |
| ID | 8 |
| AR | 7 |
| MT | 6 |
| VT | 5 |
| KS | 4 |
| PR | 3 |
| DE | 3 |
| WV | 3 |
| MP | 2 |
| SD | 2 |
| AK | 2 |
| ND | 2 |
| VI | 2 |
| WY | 2 |
| GU | 1 |
Incident Breakdown
What Triggers Campus Alerts
Distribution across 40+ incident types: from active shooters to weather emergencies, gas leaks, and public-health crises.
How to read this: This is the mix of incidents we happen to have documented, not how often each type actually occurs on campuses.
Type Distribution
By Incident Type
- shooting231shooting231
- robbery193robbery193
- bomb threat114bomb threat114
- burglary94burglary94
- severe weather92severe weather92
- sexual assault87sexual assault87
- police activity83police activity83
- armed person77armed person77
- other71other71
- swatting68swatting68
What to notice: shooting-adjacent reports dominate the record. The cases that rewrote alert language live in the Active Shooters chapter.
| Institution | Cases |
|---|---|
| shooting | 231 |
| robbery | 193 |
| bomb threat | 114 |
| burglary | 94 |
| severe weather | 92 |
| sexual assault | 87 |
| police activity | 83 |
| armed person | 77 |
| other | 71 |
| swatting | 68 |
Year Trend
Cases by Year
What to notice: coverage is densest where original alert text still survives online. Browse the most recent cases to read today's alert language, or read the five eras that shaped this curve.
| Label | Count |
|---|---|
| 2007 | 4 |
| 2008 | 1 |
| 2009 | 1 |
| 2010 | 1 |
| 2011 | 1 |
| 2012 | 5 |
| 2013 | 10 |
| 2014 | 12 |
| 2015 | 11 |
| 2016 | 33 |
| 2017 | 43 |
| 2018 | 51 |
| 2019 | 70 |
| 2020 | 53 |
| 2021 | 63 |
| 2022 | 223 |
| 2023 | 245 |
| 2024 | 478 |
| 2025 | 434 |
| 2026 | 284 |
Eras of campus alerting
Five moments explain most of the shape above: two federal laws, one massacre, and two waves of threats that never materialized.
The Clery Act
Campus crime disclosure becomes federal law: colleges must publish crime statistics and warn their communities about ongoing threats. Every timely warning in this archive traces its legal DNA here.
Browse the 1990s →Virginia Tech and the ENS era
Thirty-two people are killed on April 16, 2007; the first campuswide email arrives more than two hours after the first shootings. The massacre makes mass-notification systems all but universal within two years.
Read the case →HEOA
The Higher Education Opportunity Act writes emergency notification into federal law: immediate notification upon confirmation of a significant emergency, publicly disclosed procedures, and an annual test.
Browse the policy archive →The HBCU bomb-threat wave
Dozens of historically Black colleges and universities receive coordinated bomb threats in January and February 2022, many timed to the first day of Black History Month.
See the 2022 HBCU cases →The swatting era
False active-shooter calls hit campuses at scale, forcing full lockdown responses to threats that turn out not to exist. The archive marks these honestly: resolved as hoax or unfounded, never as non-events.
Browse hoax and unfounded cases →Institution Lens
Who Files the Most Alerts
Institution type, Clery classification, and incident resolution: three views of the same archive from different regulatory angles.
How to read this: This reflects which institutions and categories are represented in the archive so far; it over-represents some types (such as large universities) and under-represents others.
Institution Type
By Institution Type
- public r11,053public r11,053
- private r1434private r1434
- community college126community college126
- public r293public r293
- public masters88public masters88
- hbcu72hbcu72
- private liberal arts49private liberal arts49
- private r241private r241
- private masters27private masters27
- public bachelors13public bachelors13
- private bachelors12private bachelors12
- territory7territory7
- technical college6technical college6
- for profit2for profit2
- tribal college1tribal college1
- public1public1
- military1military1
What to notice: large publics are over-represented because their alert archives are public and indexed. Browse by institution to find the community colleges, HBCUs, and tribal colleges the archive actively balances toward.
| Institution | Cases |
|---|---|
| public r1 | 1,053 |
| private r1 | 434 |
| community college | 126 |
| public r2 | 93 |
| public masters | 88 |
| hbcu | 72 |
| private liberal arts | 49 |
| private r2 | 41 |
| private masters | 27 |
| public bachelors | 13 |
| private bachelors | 12 |
| territory | 7 |
| technical college | 6 |
| for profit | 2 |
| tribal college | 1 |
| public | 1 |
| military | 1 |
Clery Classification
By Clery Category
What to notice: timely warnings are a genre of their own, composed hours after a crime rather than seconds into one. See the timely-warning chapter.
| Label | Count |
|---|---|
| emergency notification | 1,300 |
| timely warning | 498 |
| advisory | 190 |
| missing student | 14 |
| community notification | 12 |
| test | 12 |
Outcome
By Resolution
What to notice: many cases documented here as active threats resolved as hoax or unfounded. Read the hoax and unfounded cases for the language of standing down.
| Label | Count |
|---|---|
| confirmed threat | 670 |
| under investigation | 485 |
| all clear | 204 |
| unfounded | 160 |
| contained | 156 |
| confirmed hoax | 134 |
| ongoing | 107 |
| resolved | 28 |
| false alarm | 20 |
| unknown | 15 |
| resolved safely | 4 |
| suspect apprehended | 3 |
| hoax | 3 |
| weather cleared | 3 |
| fatality | 2 |
| no impact confirmed | 2 |
| voluntary end | 2 |
| suspect not apprehended | 2 |
| suspect fled | 1 |
| suspect neutralized | 1 |
Leaders & Patterns
States & Institutions in Focus
Top states by documentation volume, and the 12 most-represented institutions in the archive.
How to read this: 'Top' and 'most' here mean most-documented in this archive, which tracks how findable and well-covered a place is as much as any real-world frequency.
State Volume
Top States by Case Count
- PA250PA250
- NC148NC148
- CA137CA137
- NY134NY134
- SC104SC104
- TX98TX98
- IL77IL77
- MN77MN77
- MA75MA75
- VA72VA72
- FL68FL68
- RI63RI63
- OH53OH53
- LA53LA53
- IN45IN45
What to notice: state volume tracks collection effort as much as risk. The institutions index shows exactly what is documented where.
| Institution | Cases |
|---|---|
| PA | 250 |
| NC | 148 |
| CA | 137 |
| NY | 134 |
| SC | 104 |
| TX | 98 |
| IL | 77 |
| MN | 77 |
| MA | 75 |
| VA | 72 |
| FL | 68 |
| RI | 63 |
| OH | 53 |
| LA | 53 |
| IN | 45 |
Institution Volume
Most-Represented Institutions
- Temple · PA208Temple208
- UNC · NC108UNC108
- Columbia · NY81Columbia81
- Clemson · SC79Clemson79
- UMN · MN62UMN62
- Brown · RI58Brown58
- Illinois · IL43Illinois43
- USC · CA37USC37
- MIT · MA33MIT33
- Tulane · LA33Tulane33
- OSU · OH26OSU26
- South Carolina · SC21South Carolina21
What to notice: repeat appearances usually mean a well-indexed public alert archive, exactly the sources that make word-for-word recovery possible.
| Institution | State | Cases |
|---|---|---|
| Temple | PA | 208 |
| UNC | NC | 108 |
| Columbia | NY | 81 |
| Clemson | SC | 79 |
| UMN | MN | 62 |
| Brown | RI | 58 |
| Illinois | IL | 43 |
| USC | CA | 37 |
| MIT | MA | 33 |
| Tulane | LA | 33 |
| OSU | OH | 26 |
| South Carolina | SC | 21 |
Long-Range View
The archive over time
Documented cases span decades. Coverage is densest for recent years, where alert text is easiest to recover.
How to read this: This reflects how findable and well-documented incidents are by era, not real-world frequency.
Long-Range View
Cases by Decade
What to notice: coverage thins before the SMS era. The chapter narratives trace how alert language evolved decade by decade.
| Label | Count |
|---|---|
| 1960s | 1 |
| 1970s | 1 |
| 1990s | 1 |
| 2000s | 6 |
| 2010s | 237 |
| 2020s | 1,780 |
Archive Authenticity
Verbatim Recovery Over Time
Each column is the total alert messages for that incident year, split into verbatim-confirmed (exact transmitted text) and reconstructed (paraphrased from secondary coverage).
How to read this: This tracks our own progress recovering exact alert wording; it is a measure of the archive itself, not a claim about campus alerts in general.
Authenticity Trend
Verbatim Recovery by Year
What to notice: word-for-word recovery is this archive's core work. Browse every verbatim-confirmed case to read the exact transmitted text.
- 2011: 2 of 5 alert messages verbatim-confirmed (40%).
- 2012: 9 of 17 alert messages verbatim-confirmed (53%).
- 2013: 23 of 39 alert messages verbatim-confirmed (59%).
- 2014: 22 of 34 alert messages verbatim-confirmed (65%).
- 2015: 32 of 40 alert messages verbatim-confirmed (80%).
- 2016: 48 of 59 alert messages verbatim-confirmed (81%).
- 2017: 136 of 157 alert messages verbatim-confirmed (87%).
- 2018: 127 of 156 alert messages verbatim-confirmed (81%).
- 2019: 120 of 135 alert messages verbatim-confirmed (89%).
- 2020: 94 of 112 alert messages verbatim-confirmed (84%).
- 2021: 117 of 133 alert messages verbatim-confirmed (88%).
- 2022: 339 of 385 alert messages verbatim-confirmed (88%).
- 2023: 444 of 517 alert messages verbatim-confirmed (86%).
- 2024: 920 of 1076 alert messages verbatim-confirmed (86%).
- 2025: 820 of 923 alert messages verbatim-confirmed (89%).
- 2026: 536 of 585 alert messages verbatim-confirmed (92%).
The verbatim share climbs sharply in recent years as Twitter/X archive recovery and university alert-page indexing make exact wording recoverable. The 2018 inflection point tracks the widespread adoption of social-media alert channels.
Methodology Note
All figures derive from the live data/cases/ directory and recompute at build time. Verbatim confirmation requires a direct source URL; reconstructed alerts are clearly marked. Where a source documents the interval from incident to first alert, that timing is recorded on the case and included in exports. The descriptive counts on this page are plain build-time aggregation; the archive’s cross-corpus analytics layer is owned by Claude Fable 5.