Methodology
This page documents how metrics are defined and how to interpret them. When multiple sources exist, the most transparent and reproducible source is preferred.
Financial metrics
- Salary at graduation (average, USD).
- Salary five years after graduation (average, USD).
- Tuition (in-state and out-of-state where applicable).
- Break-even (years) uses a simple estimate:breakEvenYears = tuitionPerYear / salaryAtGraduation(Does not include living costs, taxes, discounting, or opportunity cost.)
Intangibles
- Completion rate and dropout rate come from institution-level College Scorecard outcome fields. They are shown only in the university-first flow as school-level context and are not displayed in the major-first flow because they are not major-specific.
- Likelihood of employment in-field is shown only in the major-first flow. It is sourced from NCES Digest Table 505.50, using the published percentage of full-time employees in jobs closely related to their field of study.
- Each major is mapped into an NCES broad field using its CIP code. This makes the displayed value a category-level estimate rather than a fully major-specific placement rate.
- AI exposure is shown only in the major-first flow. It is now computed from linked occupations:Major CIP code -> NCES/BLS CIP-SOC crosswalk -> AIOE occupation scores -> equal-weight meanOccupation AIOE scores are normalized to 0..100 before aggregation. If an exact 6-digit CIP match is unavailable, the method falls back to the closest usable CIP family so every major still receives a source-backed estimate.
- AI exposure is derived from the official NCES/BLS CIP-to-SOC crosswalk and Felten, Raj, and Seamans' AIOE occupation scores. Each major is matched to its linked occupations, occupation scores are normalized to 0..100, and the major's score is the equal-weight mean across matched occupations.
In-field employment methodology
- The source is NCES Digest Table 505.50, using the 2015-16 bachelor's degree recipient cohort measured in 2017.
- The app uses the published NCES statistic for the percent of full-time employees working in a job closely related to their field of study.
- Each major is mapped into one NCES broad field using its CIP code, then assigned that field's published percentage.
- The main broad fields include engineering, biological and physical sciences, mathematics and computer science, social sciences and history, humanities, health professions, business and management, education, psychology, and public affairs and social services.
- When a major does not cleanly fit one of those published NCES buckets, the app falls back to an `Other` category or a manual override.
- This value should be interpreted as a field-level estimate, not as a precise placement rate for one exact major at one exact school.
AI exposure methodology
- The app starts with the major's CIP code, maps it to occupations using the official NCES/BLS CIP 2020 to SOC 2018 crosswalk, and then looks up each occupation's AIOE score.
- AIOE is an AI exposure index, not a literal probability of job loss. The displayed value should be interpreted as relative exposure within the dataset, not as a forecasted percent chance of displacement.
- Occupation scores are min-max normalized to a `0..100` scale and then averaged with equal weight across the matched occupations for that major.
- Matching uses a fallback hierarchy: exact CIP match first, then a canonicalized CIP match for legacy/irregular major codes, then a broader CIP-family fallback when a narrower official match is unavailable.
- Confidence labels reflect that hierarchy: `High` for exact matches, `Medium` for canonicalized matches, and `Low` for broad fallback matches.
Missing data will be shown as “Data not available” rather than filled or guessed.
Data sources
Sources used here include College Scorecard for tuition and salary data, the NCES/BLS CIP 2020 to SOC 2018 crosswalk for major-to-occupation mapping, NCES Digest Table 505.50 for broad-field in-field employment likelihood, and Felten, Raj, and Seamans' AIOE appendix for occupation-level AI exposure.