How we built the quiz

Last updated October 7, 2026 · Quiz version 2026-10-06

What the quiz measures

The quiz follows social cognitive career theory, which links career and study choices to three things: what people are interested in, how confident they feel, and what they expect a career to give them.

• Interests (48 questions). Each question describes an activity, such as "Write code for an app or website," and asks how much you'd like it, from Strongly dislike to Strongly like. This is the same style and scale as the O*NET® Interest Profiler. Each question was written for one of the specific interest areas in the O*NET 31.0 database, such as Life Science, Finance or Design. The quiz uses 36 of its 41 areas. Four of the five it leaves out describe work that usually doesn't require a bachelor's degree (manual labor, machine operation, personal service and culinary arts). The fifth, religious activities, was also left out to keep the quiz short, so Religious Studies matches rely on that major's other interest areas. Twelve areas that many majors depend on, such as Engineering, Math & Statistics, Teaching and Health Care, get a second question that describes a different side of the area, and the quiz averages the two. In our tests, this kept more majors in their students' top five when one question didn't fit how a person thinks about the field. Every area also feeds one or two of Holland's six RIASEC types, so each type score combines 8 to 12 questions.
• Confidence (6 questions). How confident you feel about math, science, writing, technology, art and leading people.
• Career goals (8 questions). How important pay, job growth, helping people, creative work, working outdoors, leading others, regular hours and skipping grad school are to you.

Where each major's profile comes from

No government dataset rates college majors directly, so each major's profile is built from the jobs it leads to:

1. Major to jobs. The National Center for Education Statistics CIP 2020 to SOC 2018 crosswalk links each major to the occupations it prepares students for.
2. Jobs to ratings. The O*NET 31.0 database (August 2026) rates about 900 occupations on the 41 interest areas, the six RIASEC types, the knowledge it needs, its work activities and its work conditions.
3. Pay and growth. The U.S. Bureau of Labor Statistics Employment Projections for 2025-35 supply each occupation's 2025 median pay, projected job growth and typical entry education.
4. Averaging. A major's profile is the average of its jobs, weighted by the square root of each job's employment, so large occupations count more without drowning out the rest.

Some crosswalk links would distort a major's profile, so a few rules apply. Jobs that usually need a bachelor's degree or more (O*NET Job Zones 4 and 5) are preferred. Two catch-all codes, "Managers, All Other" and "Life, Physical, and Social Science Technicians, All Other," are skipped. Jobs that the Bureau of Labor Statistics says need five or more years of experience in a related job, mostly senior managers and executives, are left out because graduates don't start there, unless a major has no other jobs. Elementary and high school teaching counts only for the Education majors, because it otherwise swamps fields like Biology and English. College professor codes are used only when a major has fewer than two other jobs.

Majors with hand-picked jobs

For eight majors the crosswalk lists no jobs or only a narrow few, so we added jobs graduates commonly go on to. We list them here so the choice is open:

• Health Sciences: medical and health services managers, physician assistants, occupational therapists, physical therapists, health information technologists
• Kinesiology & Exercise Science: physical therapists, athletic trainers
• English: editors, technical writers, writers and authors
• History: archivists, curators, museum technicians
• Philosophy: lawyers
• Animal Science: veterinarians
• Criminal Justice: police officers, detectives, probation officers
• Criminology: probation officers, detectives, forensic science technicians

Two more choices to note. The crosswalk's general Health Sciences program code (51.0000) links to no jobs, so Health Sciences uses the closest code, Health and Wellness, General (51.0001), plus the jobs above. Pre-Law & Legal Studies combines two program codes, Legal Studies (22.0000) and Law (22.0101), and Religious Studies combines Religion/Religious Studies (38.0201) and Theological Studies (39.0601).

How your match score is calculated

1. Interest fit. Your interest answers form a profile across the 36 areas, using the average where an area has two questions. First, each answer is pulled 20% toward its related interest areas, using how interests go together across the 891 occupations in the O*NET database (for example, Engineering travels with Mechanics, Physical Science and Math). That way one answer that doesn't quite fit the way a question was worded can't erase a whole area. Each major's profile gets the same treatment. We then compute the correlation (r) between your profile and the major's. Correlation compares the shape of the two profiles, not their level, which is how the O*NET Interest Profiler matches people to careers. The base score is 50 + 40 × r.
2. Career goals, up to ±6 points. Only goals you rate Very important or Essential count. A major gains points when its jobs score above the quiz average on those goals and loses points when they score below.
3. Confidence, from −8 to +3 points. This applies only to subjects a major leans on heavily. Low confidence there lowers the score a little and adds a heads-up; high confidence raises it a little. A low answer never removes a major.

The highest possible score is 99. Scores are rounded, so two majors can show the same percentage; the one with the closer interest fit is listed first.

Majors whose correlation with your profile is negative are labeled "Not a match" instead of getting a percentage, a rule also used by the O*NET Interest Profiler. "Also worth a look" shows up to three strong matches from fields different from your top three, one per field.

Confidence and career goals count for less than interests on purpose. Research finds that confidence and interests are closely linked, and confidence differs by gender from subject to subject. For example, a meta-analysis of 68,429 students found boys report more math and computer confidence and girls more language arts confidence.

How we tested it

We ran made-up students with clear-cut interests through the quiz to check that obvious cases come out as expected. A student who loves coding and math got Computer Engineering, Data Science and Computer Science. A student who loves helping people got Special Education, Social Work and the teaching majors.

We also tested stability, since real people answer with some randomness. We added random answer noise to eight made-up students, 20 times each, and counted how many of their top five majors stayed in the top five.

The six-type method also produced implausible matches, so the quiz uses the 36-area method.

Next, we checked that every major can be found. For each major, we answered the quiz the way its own profile would, mapping the O*NET 1-to-7 ratings onto the 1-to-5 answer scale. 80 of the 86 majors came out first and all 86 came out in the top three. In a harsher version, where many activities get "Strongly like" as some students answer, 81 of 86 stayed in their own top five, and the rest landed just outside it behind closely related majors. That check led us to leave out the senior-manager jobs described above, which had been pushing computing majors down.

Then we tested what happens when answers vary, because two people interested in the same field don't answer the same way. For each of the 86 majors we simulated 50 students who are interested in it but answer imperfectly: one key activity "doesn't click" and gets a 3 instead of a 5, and about a fifth of the other answers move up or down a notch. The exact major landed in the top five for 4,298 of those 4,300 simulated students, and the #1 result was in the right family (the same field, or a major with nearly the same profile) for 4,289 of them. Under a harsher version, with two answers that don't click, more random variation and some students who avoid or overuse the extremes, the figures were 97% and 95%. This test is why the quiz pulls each answer toward its related areas: without that step, a single mismatched answer knocked more majors out of their own top five.

Last, we wrote 22 realistic students with mixed likes, dislikes, confidence levels and career goals, such as a pre-med student, a future nurse, a theater kid, a future lawyer and a chemical engineer, and listed the majors a counselor would expect for each. Each one took the quiz 10 times with random answer noise. In 213 of those 220 runs (97%), the top match was one the counselor would expect and at least two of the top five were too. Building this test is also how we found and fixed weak profiles for Finance and Criminology.

A real chemical engineer who took the quiz was matched with Mathematics. That happened because the engineering question at the time described a prosthetic hand and a bridge, which didn't fit chemical engineering, and the single answer to that question carried the whole engineering signal. We rewrote the question to cover more branches of engineering, reworded the physical-science question so it isn't only about lab experiments, and added the smoothing step described above. Later we added a second engineering question about making factories, power plants and water systems run better, so engineering interest no longer hangs on one question.

Fair wording

Interest questionnaires can show large gender gaps, and how the questions are worded affects how large those gaps are. We reviewed the original 36 activities and rewrote five that leaned on stereotyped settings, for example replacing "make an engine work better" with a design problem. The engineering question now reads "Use math and science to design something that works, like a bridge, a battery or a cleaner fuel," which covers civil, electrical, chemical and mechanical engineering without the engine.

The research behind it

Limits

• The link between interest fit and college success is real but small. Grades, study habits, cost, the specific college and many other things matter too.
• Each major's profile describes the jobs it leads to, not the students in it. A large German study found that fit with fellow students predicted staying in a program better than fit with the target job.
• 24 of the 36 interest areas rest on one question each; the other 12 have two. The broader six-type scores are more reliable than any single area.
• Some majors lead to almost the same jobs, so their profiles are nearly identical: Biology and Neuroscience, Computer Science and Information Technology, Foreign Languages and Linguistics, and others. The quiz can't tell those apart, so it labels them on your results and suggests comparing their courses.
• Religious activities isn't one of the 36 areas, so Religious Studies matches are less precise.
• Teen interests are still changing, so results can shift over a year or two.
• The quiz is for exploration. It is not a substitute for talking with a school counselor or academic advisor.

What's next

People who finish the quiz can choose to share their answers anonymously, along with their major. Once enough people have, we'll check each scale's reliability against the O*NET standard and start building major profiles from the interests of real students in each major. See the privacy policy for what is and isn't collected.

Data credits

This page includes information from the O*NET 31.0 Database by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA). Used under the CC BY 4.0 license. The College Major Quiz has modified all or some of this information. USDOL/ETA has not approved, endorsed, or tested these modifications. O*NET® is a trademark of USDOL/ETA. Pay and growth data come from the U.S. Bureau of Labor Statistics; the major-to-job links come from the National Center for Education Statistics. Full references are listed on the quiz page.