How the map is built

Source data

2020 CIP / 2018 SOC Crosswalk, National Center for Education Statistics and Bureau of Labor Statistics. Downloaded 2026-09-06 from nces.ed.gov, SHA-256 ba3d59a191b9d977a5c457a66b9348c4f2f7963aafacf72c0b80113b46bf0ab8. The file is pinned locally and the site is rebuilt from that copy, so a page never depends on a government host being up.

Wage, typical entry education, projected growth and annual openings come from a second official source: BLS Employment Projections, Table 1.2, Table 1.2 Occupational projections, 2025–35, and worker characteristics, 2025 (Employment in thousands). Wages are 2025 medians; growth is the 2025-35 projection. Working-condition figures are O*NET 30.3 mean incumbent ratings, the same release every other O*NET-derived dataset on this site uses.

Four sources, four vintages

Nothing on these pages is “2020 data” or “2025 data”. Four independent sources are joined, each carries its own vintage, and each supports a different kind of statement.

The crosswalk being keyed to CIP 2020 and SOC 2018 does not make the wage figures old: those are 2025medians joined to the same SOC codes. It is the current official crosswalk — NCES publishes no newer edition — and CIP editions are revised roughly once a decade.

What the crosswalk actually says

The crosswalk's own description of its main sheet is one sentence: “This file crosswalks 2020 CIP Codes to 2018 SOC Codes in ascending order by CIP Code.” It contains 5917 relationships between 2142 instructional programs and 688 occupations. It is a many-to-many relation and nothing more: it carries no relationship strength, no employment share, no ordering, and no statement about qualification.

That sentence is the ceiling on what this map may claim, and the design follows from it. A relationship means a field of study is related to a line of work. It does not mean the degree qualifies anyone for the occupation, that graduates usually enter it, or that any given share of them do. No page here says otherwise, and an automated check fails the build if one ever does.

NCES also ships a “NO MATCH” bookkeeping row under CIP 99.9999 carrying 180 occupations. It is not an instructional program and is excluded; leaving it in would have contaminated every aggregate.

Grouping programs into fields of study

Nobody searches for CIP 42.2801. They search for “psychology degree”. CIP's grain is far finer than search language — psychology alone is 31 six-digit programs — so publishing one page per code would put dozens of near-identical lists in front of a single question and let them compete with each other.

So the unit of publication is a field of study: the phrase a person types, bound to the CIP programs it covers. 116 fields are published, drawing on 1513 program bindings in total; 106of them group more than one program. Each binding is declared by hand, never inferred, and every page prints its member CIP codes with their official titles. Where a field's evidence sits only at the broader family level, the page is titled for the family rather than for one member program.

A field is published only if the joined evidence can fill a page: at least 4 related occupations, at least 3 of them with a published BLS wage, and at least 2 distinct typical entry-education levels among them. That is an editorial rule about which pages are worth publishing, not a statistical threshold, and it says nothing about the fields it excludes.

Two fields that resolved to an identical occupation set would be the same answer under two headings, so only one is published and the other is dropped rather than shipped as a near-duplicate.

Joining the occupation facts

Crosswalk occupations are 6-digit SOC 2018 codes. Each is matched to the BLS line item with that code; occupations with no line item are counted on the page and left out of the table rather than shown with blank columns or a borrowed group figure. Nothing is estimated and no value is carried across from a related occupation.

Typical entry education is BLS's description of the education most workers in an occupation hold when they enter it. It is a modal observation, not a requirement, and it is deliberately not combined with the crosswalk relationship into a single verdict. Occupations are split into “at or below a bachelor's degree” and “above a bachelor's degree” because that is the distinction a person holding a bachelor's is actually weighing. It carries no information about licences, certifications or experience, which BLS records separately and this map does not show.

What the map cannot tell you

It cannot tell you whether you would be hired. A crosswalk describes fields and occupations in general; one employer's posting can overturn everything on these pages. That is what the Job Fit Analyzer is for.

It cannot tell you what graduates of a field actually earn. Every wage here belongs to an occupation, not to a degree. Two people with the same degree in different occupations on the same page are separated by far more than the degree.

It cannot rank fields of study. The explorer orders alphabetically by default for that reason, and the sort options reorder published facts rather than scoring anything.

It is a snapshot of CIP 2020, SOC 2018 and BLS 2025. The version appears on every page, and a new crosswalk edition would be a version migration rather than a silent overwrite.

Reproducing the map

Take the crosswalk at the URL above, drop CIP 99.9999, group the programs exactly as each page prints them, take the union of their related SOC codes, and join BLS Employment Projections, Table 1.2. Every list on this site is that operation and nothing else. Version cip2020-soc2018-2026-09-06, generated 2026-09-06.