Case Study — Alliance Against Seclusion & Restraint

The complete federal record on who gets restrained and secluded in American schools.

Prior Signal transformed two years of raw U.S. Department of Education CRDC data into an accessible, credible, interactive public record — for the nation's leading restraint-and-seclusion advocacy organization.

2 school years 98,010 schools All 50 states + DC WCAG 2.1 AA No external dependencies
Open the data experience →

01 — The Brief

What AASR needed.

The Alliance Against Seclusion & Restraint is the country's most visible advocacy organization working to end the physical restraint and seclusion of students in schools. Every two years, the U.S. Department of Education releases the Civil Rights Data Collection — the most comprehensive federal census of school discipline practices. AASR needed a data experience that could turn this raw federal record into an authoritative, citable public reference: one their advocates, congressional contacts, and journalists could use.

  • Translate raw OCR CRDC CSV files (2020–21, 2021–22) into a verifiable public record — honest about suppression and limitations.
  • Lead with disability disproportionality: AASR's central advocacy frame, supported by relative-risk and concentration metrics.
  • No frameworks, no CDN dependencies — the artifact must survive server shutdowns and be shareable as a single file forever.
  • Trauma-informed editorial voice: facts without sensationalism, credibility without coldness.

02 — The Approach

Structure. Verify. Build. Design.

Prior Signal's method for high-stakes data work: every output is ground-truthed against the official record before any design work begins. In an advocacy context, a data error isn't an inconvenience — it's a liability.

Phase 01

Structure

Python pipeline parses and normalizes both CRDC vintages. All 14 headline figures are computed independently and held for cross-validation.

Phase 02

Verify

Each computed figure is asserted against OCR's official "A First Look" publication. Any mismatch aborts the pipeline. Zero tolerance for data error in an advocacy context.

Phase 03

Build

Verified data is inlined into a single HTML file. Vanilla JS only — no runtime network requests, no third-party scripts, no build toolchain required to run it.

Phase 04

Design

WCAG 2.1 AA throughout. Every chart has an accessible data table fallback. Okabe-Ito colorblind-safe palette. Reduced-motion gate on all animations.

03 — The Experience

Explore the data.

The full interactive tool, embedded below. Switch between school years and toggle between student counts and incident totals. Click any state to drill into its district-level record.

Embedded: the complete interactive data experience. All data sourced from the U.S. Department of Education Office for Civil Rights, Civil Rights Data Collection, 2020–21 and 2021–22.

04 — Key Findings

What the data shows.

These figures are computed directly from the CRDC record and verified against OCR's official publications. Each represents a distinct dimension of the restraint-and-seclusion crisis — disability disproportionality, racial disparity, and escalating intensity over time.

77.3%

of restrained children have disabilities

Students with disabilities represent just 14% of total enrollment — yet account for more than three-quarters of every child physically restrained in American schools.

5.5×

relative restraint risk for disabled students

Controlling for enrollment share, a student with a disability is 5.5 times more likely to be physically restrained than a non-disabled peer. The gap has not narrowed in two decades.

40%

of mechanically restrained children are Black

Black students represent 15% of enrollment but 40% of children subjected to mechanical restraint — devices such as straps, cuffs, or prone-restraint boards. The racial disparity is most severe in this highest-risk category.

3.3×

average incidents per affected child

The CRDC captures both children affected and total incidents. The ratio reveals that restraint is not a one-time event — each affected child was restrained an average of 3.3 times in a single school year.

+127%

rise in mechanical restraint since 2017–18

Mechanical restraint — the most dangerous category — grew from 3,619 children in 2017–18 to 8,219 in 2021–22. Physical restraint broadly has declined modestly; mechanical restraint has more than doubled.

13%

of schools suppressed all R&S data

OCR suppresses figures for small student groups to protect privacy — but suppression rates are themselves revealing. One in eight schools in the dataset reported no usable R&S data, making the true scale unknowable from the public record alone.

05 — Deliverables

What Prior Signal produced.

Every deliverable is designed to outlast the engagement — portable, self-verifying, and legible to a non-technical audience.

01

Interactive two-year data experience

Single self-contained HTML file (~820KB). All data inlined. No server required. Works offline.

02

Python data pipeline

14-figure ground-truth validation against OCR's "A First Look" reports. Pipeline aborts on any mismatch.

03

Cleaned, annotated datasets

JSON + CSV outputs for both school years. Suppression codes preserved and annotated for downstream use.

04

Static archive pages

Per-year snapshot pages for long-term citability. Stable URLs, no JavaScript dependency.

05

Social share card suite

12 PNG crops at standard OG dimensions — optimized for Twitter/X, LinkedIn, and Facebook preview contexts.

06

Methodology documentation

Full methodology note covering source files, suppression logic, derived metrics, and reserve-code analysis.

07

WCAG 2.1 AA audit — 20/20

Axe-core automated pass plus manual keyboard and screen-reader testing. Full accessibility score documented.

06 — Impact

Why it matters.

A citable, credible reference

AASR's advocates, congressional contacts, and journalists now have a single, verifiable source for the federal restraint-and-seclusion record — built to the evidentiary standard required in legislative and legal contexts.

Disability disproportionality, made visible

The experience leads with relative-risk metrics rather than raw counts — translating the scale of the crisis into the analytical frame most useful to disability-rights advocates and policymakers.

Suppression made legible

Rather than hiding OCR's data gaps in footnotes, the experience surfaces suppression rates as a first-class finding — making the known unknowns part of the story, not an asterisk at the bottom.

Keeping All Students Safe Act

The data experience links directly to KASSA — the federal legislation AASR is championing — with a find-your-representative CTA that turns data engagement into direct legislative action.

This is Prior Signal's work.

Evidence-first research and data products for organizations that need to be right. If you have a dataset that the public needs to understand, let's talk.