College ROI Dataset — lifetime return on investment of U.S. colleges and majors

The AI metric measures exposure, not displacement.</p><p><strong>Vintage disclosure:</strong> ROI earnings derive from FREOPP's 2021 publication (the newest with a public download); costs are IPEDS 2023-24 Final; deflators BEA 2023.</p><p>Canonical repository: github.com/thomasthinks/college-roi-data · Interactive tables: le-teen.com/rankings</p><p><strong>Full mirror set:</strong> Zenodo (hub, DOI 10.5281/zenodo.21351602) · GitHub · Hugging Face · Kaggle · OSF · figshare · OpenML.

Also a free keyless JSON API at <a href="le-teen.com/api">le-teen.com/api</a> (640 endpoints) and an MCP server for LLM tooling (<code>college-roi-mcp</code> on npm).</p><p><strong>Changes in v2 (2026-08-13) — data correction.</strong> In <code>roi-by-major-category.csv</code>, the column <code>pct_never_breakeven</code> was carrying the share of graduates whose 30-year lifetime ROI is below zero, rather than the share FREOPP flags with its 999 “never breaks even” sentinel.

Five of nineteen rows are corrected: Agriculture and Natural Resources 15.8→15.5, Communications and Journalism 16.5→16.4, Mathematics and Statistics 3.2→2.9, Computer and Information Sciences 2.5→2.3, Economics 2.1→1.9. The remaining fourteen are identical under either definition. A new column <code>pct_negative_lifetime_roi</code> now carries the below-zero share, so both figures are available and no existing column was renamed or removed.

The two are not interchangeable: never-breaks-even is a strict subset of below-zero-ROI, and the graduates in the gap do break even, in their late fifties, past the 30-year window. No other table is affected.</p><p><strong>Changes in v4 (2026-09-24) — four columns added to <code>out-of-state-penalty.csv</code>, no value changed.</strong> The file still holds all 455 public universities, and its first eight columns are byte-identical to v3 on every row.

Ten of the 455 are graduate/professional or health-science campuses that do not admit first-time freshmen; LE TEEN leaves them out of the named rankings it publishes, and until now nothing in this file said so, which put University of Maryland, Baltimore at #7 in the out-of-state ranking with no label. Four columns are appended: <code>unitid</code> (the IPEDS join key to <code>institutions.csv</code>), <code>admits_freshmen</code>, <code>excluded_from_named_rankings_reason</code> (empty or <code>no-freshman-admission</code>) and <code>rank_freshman_admitting</code>, the 1..445 rank le-teen.com publishes. <code>rank</code> still runs 1..455 over every row of this file.

The exclusion is editorial, not a data correction: the ten rows stay, with their values unchanged. This matches Zenodo version 1.1.0. No other table is affected.</p>