Civil Rights Data Collection figures traced to their records and checked before the public sees them
Every CRDC figure traced to its records and checked before the public sees it.
See one case, screen by screen ↓Civil Rights Data Collection (CRDC) Reporting is how a district data team builds its Civil Rights Data Collection: five build agents assemble every discipline, restraint and seclusion, harassment, course access and staffing figure from the student information system, special education and HR. The Plausibility agent checks every figure against the last collection and the source logs, and each school's profile is previewed as the public will see it. Stewards fix their own systems, special education confirms every restraint and seclusion figure, the data director approves each module and the superintendent certifies.
AI in education: a buyer's guide to agentic AI for colleges, universities and school districts →
Updated
The Civil Rights Data Collection is the mandatory survey every public school district answers for the Office for Civil Rights, school by school: enrollment and course access, discipline, restraint and seclusion, harassment or bullying, and staffing, each broken out by race, sex, disability and English learner status. It is published school by school, and advocates, reporters, researchers and OCR's own investigators read it that way. The next collection, on 2025–26 data, is collected during the 2026–27 school year.
Today a district data office builds its CRDC reporting from extracts of the student information system, a separate behavior incident system, the special education system and HR, joined by student ID in spreadsheets against a federal window. Nothing flags a zero, so a restraint logged in one system and missing from the federal extract becomes a reported zero. Special education looks at a total, not record by record, and leaders usually see a school's disparities for the first time in a news story or an OCR data request. Civil Rights Data Collection (CRDC) Reporting builds every figure with one agent per module, checks every figure against the last collection and the source logs, and shows each school's profile as the public will read it before anyone certifies.
Estimated split of a district's hours, by hand and with the agents; our arithmetic, kept modest. The 'before' column totals about 100 hours, the federal burden estimate divided by the number of responses. The 'after' column is mostly people's decisions: stewards' fixes, special education's confirmations and the two sign-offs.
A build agent per module, a Plausibility agent on every figure; stewards fix, special education confirms, the superintendent certifies.
Wednesday, October 21, 2026, at Marlow Bay Unified School District. Between 5:00 and 5:46 a.m. five build agents built 111,642 figures for 62 schools from 48,212 student records, 2,318 incidents and 3,912 staff, and the Plausibility agent left 20 notes for a person. Here is what happens to one of them, screen by screen, in the working solution.
The preview, marked not published, lays out Corrie Hill Middle School's figures in five sections: discipline, restraint and seclusion, harassment or bullying, course access and staffing. The school has 884 students, 15% with disabilities, 12% Black students and 22% English learners; 47 students had out-of-school suspensions. Students subjected to physical restraint reads 0, with a plain note beside it.
Looks implausible: 2023–24 reported 9; the behavior incident system holds 14 restraint reports never coded in the student information system.
The agent shows its seven steps and its sources: 0 in the student information system, 14 restraint reports in the behavior incident system, 9 in 2023–24. It proposes 14 students and 14 instances, 11 served under IDEA and 3 without disabilities, at confidence 0.97 against a 0.90 threshold, and hands the figure to the Director of Special Education.
Corrie Hill’s restraint response was never mapped when the school moved to the new system on Jan 12.
The 14 reports run from Jan 22 to Jun 4, 2026, BIS-CH-26-0114 to 0441, with student IDs masked: 11 with an IEP in effect on the incident date and 3 with no IEP or 504 plan, each with a blank code in the student information system. The interface map shows Physical restraint not mapped for Corrie Hill, against RS-P at the four other pilot schools. A submission note is drafted, and four signers are named.
Imani sends the 14 reports to Dr. Ferreira, due in three school days. His own view lists five restraint and seclusion figures and 64 records, each with grade and status on the date. He confirms all 14 with an attestation, or he could return a figure with a reason code. At Westgate Middle the federal definition applies: a prescribed bus vest is not mechanical restraint, so the proposed figure is 0.
Corrie Hill now shows 14 students and 14 instances of physical restraint: 79% served under IDEA, against 15% of students. The Submission agent drafts the note for the warning that restraint changed by more than 50% from 2023–24, and Imani can edit it before it is saved.
Corrected from 0: 14 reports confirmed by Dr. Leon Ferreira; 11 students served under IDEA; a note is added to the submission.
Tomás sees only his own fixes, each with numbered steps: code the 14 Corrie Hill records RS-P and add the missing map row, due Oct 28. When he marks it fixed, the agent re-reads the student information system, finds 14 of 14 and rebuilds the figure. Marlow Bay High's referrals to law enforcement, 9 in the school resource officer's log, are overdue since Oct 20 and already escalated.
With the state's significant disproportionality method, the agent shows Black students are 12% of Corrie Hill and 32% of the 47 suspended out of school: 15 of 106, a rate of 14.2% and a risk ratio of 3.4, above the state threshold of 3.0. Students with disabilities are at 2.9. Seven schools are above 3.0. Marcus marks it reviewed.
A review never changes a figure.
Every open note sits in one queue with what the agent found, who acts and the due date: discipline 7, restraint and seclusion 6, harassment or bullying 4, staffing 3. Westgate Middle's duplicate expulsions go from 6 to 4, Harborview's two early pickups coded as preschool suspensions from 2 to 0, and teacher absences from 41% to 24% once professional development leave is excluded. Explanations can be accepted in bulk, each with a reason code.
Course access, 46,500 figures with no open notes, is approved at once with an attestation. The others follow as their notes close. When the HR system did not answer at 6:00 a.m., staffing was built from yesterday's copy, and the screen says so with a Try again now button.
With five of five modules approved, special education's confirmations done and 0 edit-check errors, Dr. Vance signs the certification statement. The Submission agent schedules the upload for Dec 1 at 8:00 a.m. with 12 warnings, each with its note, and the OCR evidence pack MB-CRDC-2025-26-EVIDENCE.zip is sealed. The person who approved the modules cannot certify.
The same collection, seen by the people who build it, fix it, confirm it, review it and certify it: what their fall looked like, and what it looks like now.
A build agent per module, a Plausibility agent on every figure; stewards fix, special education confirms, the superintendent certifies.
Builds suspension, expulsion, referral and arrest counts for every school by race, sex, disability and English learner status, from the student information system, checked against the behavior incident system.
Counts students and instances of physical restraint, mechanical restraint and seclusion by the federal definitions, with each student's IDEA and Section 504 status on the incident date.
Builds harassment and bullying allegations, students reported and students disciplined, by basis, from the student services system and the student information system.
Builds course access figures, from Algebra I, geometry and calculus to AP, dual enrollment and gifted and talented, from the master schedule and the course code map.
Builds teacher and support staff FTE, experience and absences from HR assignments and leave.
Compares every figure with 2023–24 and the source logs, always checks a zero against the logs, finds the cause of a figure that looks wrong, proposes the corrected figure and writes the plain note.
Works out each group's rate and risk ratio by school and module with the state's significant disproportionality method, so leaders see what the public will see.
Builds the federal files, runs the published edit checks, drafts a note for every warning from the closed notes, and uploads at the window's opening.
Approve and certify. The agents propose; a named person decides.
The data director can ask the assistant about any figure and get a cited answer that opens the right screen.
Why does Corrie Hill show zero restraints?
Because the student information system has no restraint code on any Corrie Hill record. The school moved to a new behavior incident system on Jan 12, 2026, and its “Physical restraint” response was never mapped. The incident system holds 14 restraint reports for the year: 11 students served under IDEA, 3 without disabilities. 2023–24 reported 9. The Plausibility agent proposes 14; Dr. Leon Ferreira confirms.
Which figures dropped to zero since 2023–24?
Four, and each has records behind it: physical restraint at Corrie Hill Middle (9 in 2023–24; 14 reports), referrals to law enforcement at Marlow Bay High (11; 9 in the officer’s log), race harassment allegations at Westgate Middle (7; 4 reports) and calculus at Seacliff High (31; 27, fixed Oct 16).
Show risk ratios for suspensions at Corrie Hill
Black students are 12% of Corrie Hill and 32% of the 47 students suspended out of school: a risk ratio of 3.4, above the state’s 3.0. Students with disabilities: 2.9. The figure is correct as built; this is what the public will see. The Director of Student Services reviews it; a review never changes a number.
Build the OCR evidence pack
Every figure with its source records, the notes and who closed them, the special education confirmations, the module approvals and the certification, with the agents’ versions and the access log. Student IDs are masked and the key stays with the district.
12 screens from the working solution, on its sample data. Pick one to see it large.
The administrator sets the thresholds, cell sizes and clocks every figure is checked against, and every change is versioned.
| Setting | Default | Choose from |
|---|---|---|
| Flag a change from the last collection | 50% | From 20% to 100%; the Plausibility agent explains or asks about any figure that moves more |
| Always compare a zero with the source logs | Always on | A 0 where the logs hold records is always flagged |
| Protect small counts in shared copies | Under 10 students | From 5 to 20; board, cabinet and principal copies show “fewer than” |
| Risk ratio threshold | 3.0 | The state's significant disproportionality method, 34 CFR 300.647; minimum cell 10 and n-size 30 |
| Steward fixes due | 5 school days | From the agent's ask |
| Special education confirms within | 3 school days | Every restraint and seclusion figure |
| Escalate an overdue fix | 1 school day overdue | To the Director of Student Services, then the Deputy Superintendent |
| Keep each submission package | 6 years | 6, 8 or 10 years |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
Sources: GAO-19-551R, K-12 Education, June 2019 (about 70% of districts reported zero restraint and seclusion in 2015–16) · Federal Register 2025-15038, August 7, 2025 (17,717 responses and 1,762,790 burden hours for the 2025–26 and 2027–28 collections) · 34 CFR 300.647 (risk ratio method) · "target" = design goal, measured in the live solution · "demo" = seen in the working solution, on its sample data · "typical" = published figures · "estimated" = our arithmetic. The federal window and the state threshold are settings for a fictional 2025–26 collection. Marlow Bay Unified School District, its schools and the people named on this page are fictional: characters and sample data in the working solution.
Civil Rights Data Collection (CRDC) Reporting is how a district data team builds its Civil Rights Data Collection: five build agents assemble every discipline, restraint and seclusion, harassment, course access and staffing figure from the student information system, special education and HR. The Plausibility agent checks every figure against the last collection and the source logs, and each school's profile is previewed as the public will see it. Stewards fix their own systems, special education confirms every restraint and seclusion figure, the data director approves each module and the superintendent certifies.
It is the Office for Civil Rights' mandatory survey of every public school district and school, collected every two years under Title VI, Title IX, Section 504 and ADA Title II. It asks for school-level data on enrollment and course access, discipline, restraint and seclusion, harassment or bullying and staffing, by race, sex, disability and English learner status, and the data is published school by school.
The Plausibility agent always compares a zero with the source logs and the last collection. When a school reports 0 restraints but the behavior incident system holds reports, it finds the cause, such as a response code that was never mapped, and proposes the corrected figure. The Director of Special Education then confirms every restraint and seclusion figure record by record, with each student's IEP or 504 status on the incident date.
The Disparity agent works out each group's rate and risk ratio with the state's significant disproportionality method, its threshold, minimum cell size and n-size, so leaders see what advocates and OCR will compute from the published file. The Director of Student Services marks it reviewed; a review never changes a figure.
No. The agents read the district's systems and never change a source record or a figure. A wrong record goes to the steward who owns it, such as the student information system coordinator, and the agent re-reads the system after the fix. Agents never approve, confirm, certify or upload anything the superintendent has not certified.
Yes. Every accepted explanation and every returned figure needs a reason code, special education confirms with an attestation, the Director of Data and Assessment approves each module, and only the superintendent certifies. The person who approves a module cannot certify, and every action is on the audit trail.
Student-level records are for the data team only, principals see only their own school's figures, counts under 10 are protected in shared copies, and every view of a figure's records is logged. The OCR evidence pack masks student IDs, and the key stays with the district.
The Agentic Solution Engine builds and deploys it from your requirements and documents, such as your source-system extracts, code maps and your last certified collection, and it goes live once every quality gate has passed. We will walk you through it on one of your own schools first.
No uncited sentence reaches the accreditor, and likely findings are answered first.
1-min demoIPEDS ReportingEvery survey cell rebuilt from source, every variance explained before lock.
1-min demoFaculty CredentialingEvery section cleared before the term starts, with every credential tied to its proof.
1-min demoWe'll build one of your schools' Civil Rights Data Collection profiles from your own systems and show you what the public will see.