Skip to content

Judge admission by residency and test policy - #86

Merged
DanialBeg merged 1 commit into
mainfrom
feature/admission-data
Oct 27, 2026
Merged

DanialBeg merged 1 commit into
mainfrom
feature/admission-data

Conversation

@DanialBeg

Copy link
Copy Markdown
Member

What changes

Discover's labels now account for where the student lives and whether the school reads test scores. These were the two follow-ups left out of #85.

In-state and out-of-state admit rates (24 selective publics). Scorecard publishes one rate that blends residents with everyone else. These schools now carry both, and the estimate uses the one that applies to the student. Without a home state it falls back to the blended rate.

  • Sources: the school's own 2025-26 Common Data Set (section C1, split by home address); the UC Information Center for the six UC campuses; Cal Poly's institutional research dashboard; and SCHEV, Virginia's higher-education council, for Virginia and William & Mary. Every row in the migration carries its admit and applicant counts, its URL and its entering class (Fall 2025).
  • International applicants: Cal Poly, Virginia, William & Mary and Maryland count them as out-of-state, and the rows say so.
  • No split published: Michigan and Auburn keep their blended rate.
  • Checks: all 48 rates match their counts. I checked three source files myself (UNC, Georgia Tech, Washington) and their figures match exactly.
  • Examples:
    • UNC: 37% in-state, 10% out-of-state.
    • UT Austin: 36% in-state, 8% out-of-state.
    • UC Davis: 37% in-state, 63% out-of-state; out-of-state is higher at four of the six UC campuses.

Test policy (all schools). From Scorecard's admissions.test_requirements: 93 schools require tests, 1,064 make them optional, and 781 don't consider them. Every four-year school with an admit rate has a policy.

  • Test-blind schools ignore the student's scores.
  • Test-optional schools: a score that would lower the student's chance goes unsent, so it's judged as if it weren't there. The one exception is a student with no GPA on file, where the score is all we know about them. A score that won't be sent no longer marks a school as a long shot.
  • Schools that require tests count the score either way.

School popup. It now shows "In-state / out-of-state" when the school has both rates, and "SAT/ACT: Required / Optional / Not considered".

Database

20261027000000_college_admission_inputs adds five columns to colleges and fills the residency rows. It also recreates match_colleges so it returns the three new columns; the function body is otherwise unchanged.

Already applied to production, so the client's new column lists work as soon as this merges:

  • The deployed function matches the file exactly (same hash).
  • The migration is recorded under the repo's version number.
  • Test policy was loaded through the REST API under a temporary update permission gated by a one-off key, which I dropped afterwards. Only the public read policy remains.
  • From now on scripts/ingest-colleges writes test_policy on every run.

Checks

  • New tests: residency rates, required / optional / test-blind policies, the no-GPA exception, and the long-shot change.
  • Live data: across all 1,586 schools, for in-state, out-of-state and unknown students, a higher SAT, GPA, or SAT with a GPA never gets a worse label.
  • Lint, tests (720) and build all pass.
  • In the browser: I opened UNC's popup in the local dev copy (which uses the production database). It shows 37% / 10% and SAT/ACT: Optional, and UNC is a Reach for a Michigan student.

Not in this PR

  • Michigan's split: it publishes none in its Common Data Set or class profile.
  • More schools: the C1 residency table is now in most schools' 2025-26 Common Data Sets, so more could be added the same way.
  • The share of applicants who submit scores: left out because there's no principled way to use it without outcome data.

Selective publics admit residents and everyone else at very different
rates, and Scorecard publishes only the blend. 24 schools now carry both
rates, read off their own 2025-26 Common Data Sets (or the UC Information
Center, Cal Poly's dashboard and SCHEV), and the estimate uses the
student's own. Every school's test policy comes from Scorecard: a
test-blind school ignores scores, and at a test-optional one a score that
would hurt goes unsent, unless there is no GPA to judge the student by.
The school popup shows both rates and the policy.
@vercel

vercel Bot commented Oct 27, 2026 •

Copy link
Copy Markdown

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated
webapp Ready Ready Preview Oct 27, 2026 7:29am UTC

@cloudflare-workers-and-pages

Copy link
Copy Markdown

Deploying timeline-prototype with  Cloudflare Pages  Cloudflare Pages

Latest commit: 7380894
Status: ✅  Deploy successful!
Preview URL: https://159348e0.timeline-prototype.pages.dev
Branch Preview URL: https://feature-admission-data.timeline-prototype.pages.dev

View logs

@DanialBeg
DanialBeg merged commit 1337244 into main Oct 27, 2026
6 checks passed

This branch was successfully deployed

1 active deployment
Preview — 73808942 Deployed Oct 27, 2026 by vercel[bot]
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant