Methodology and notes

What the number means, how it is built, and the seven things it cannot tell you. Nothing here is hidden on the tool pages; it is only moved.

College Sonar by Optivara

The number, in one sentence

Net gain is roughly what a typical graduate of a program is financially ahead twenty years after finishing, compared with not having gone.

Four things go into it, and three of them are usually left out elsewhere.

  • What the degree pays above the alternative. Not the salary. The salary minus what a high-school graduate in the same state earns. A program paying $60,000 in a state where the comparison is $35,000 is worth $25,000 a year, not $60,000.
  • Four years of what the college actually costs you at your family’s income level, after grants and scholarships. Not the sticker price.
  • Four years of wages you do not earn while studying. This is the big one almost nobody counts: about $139,000, which is larger than the tuition for most families.
  • The chance of actually finishing. The earnings are weighted by the share of students who graduate. The costs are not, because a student who leaves still paid for the years they attended.

Future money is discounted 3% a year, because a dollar in year twenty is worth less than a dollar now.

This is not a formula we invented. The net present value of an earnings premium over a counterfactual, less the direct cost and the earnings given up, is the standard way economists measure the return on education. The fullest public implementation is the Foundation for Research on Equal Opportunity’s, whose return on investment methodology is worth reading if you want the long version.

Where we differ, we are harsher. That model runs over a working lifetime; this one stops at twenty years, and it charges four years of foregone wages against the degree. Both choices push the figure down, so a program that looks negative here can be positive on a lifetime measure. Twenty years is used because it is a span a seventeen-year-old can picture, and because a number that depends on earnings in 2070 is a forecast rather than a measurement.

What a negative number means

A negative figure does not mean the degree failed to recover its tuition. It means the graduate would have been ahead not going. Because four years of foregone wages are subtracted alongside four years of cost, a program lands below zero only when, twenty years later, what it paid above the alternative still has not made up the wages given up to earn it. Roughly half of all program rows on this site are negative at the $48k to $75k income level, so this is the ordinary case and not an exception.

The alternative being measured is a typical high-school graduate in the same state, not the specific person considering the program. Someone who would have out-earned that typical wage is giving up more than the figure assumes, and someone who would have earned less is giving up less. The number is a sound way to rank programs against each other and a rough one for any single person.

Why it is a dollar figure and not a percentage

A percentage return divides earnings by cost, and that rewards cheapness over earnings. The cheaper program wins even when the better-paying one leaves the graduate far better off. One widely circulated list ranks a college third in the nation on a 465% return while that college graduates 8.9% of its students, because the formula divides by a small number and never asks whether anyone finished.

A dollar figure keeps both halves visible, and it can be negative. About half of all programs are negative on this measure. That is the honest answer, not a glitch.

Why we rank programs, not colleges

Seventy percent of the variation in what graduates earn sits inside colleges, between their own programs, not between one college and another. At the median college the best-paying program out-earns the worst by about $48,500. So a ranking of colleges averages away the thing that decides the answer.

There is one exception. At a college whose programs all lead to similar work, a single figure for the college is fair. At a broad university it is not.

Why the selectivity band is not optional

The score rises with selectivity, partly because selective colleges enroll students who would have earned more anyway. If every list were pooled, it would return the hardest college to get into every time. Carnegie Mellon has the highest-paying programs in Pittsburgh, and that is useful only to students Carnegie Mellon admits.

So every list is ranked within a band: colleges admitting under 40%, 40% to 70%, or 70% and over. The bands are deliberately plain descriptions of how many applicants get in, not judgments about quality.

A fourth band, open admission, works differently and the difference is worth stating plainly. The first three are cut from the admit rate, so a college is in one of them because we hold its number. This one is assigned from what a college awards: mostly two-year credentials. It is named for admission rather than called “two-year” because some of these colleges award bachelor’s degrees and some of their programs take four years to finish.

Most of them do admit everyone who applies, and federal data confirms it for 881 of the 927: a college reporting an open admission policy is not asked to file an admissions survey at all, which is why it has no admit rate. But 39 report that they are not open admission, and two of those admit under 15% of applicants. Where we hold a rate for a college we show the rate, and where federal data contradicts the band we do not repeat the band. A band describes a group; it is never a claim about the college you are looking at.

Associate and bachelor’s degrees are ranked separately

An associate degree is charged two years of net price and two years of foregone earnings; a bachelor’s is charged four of each. Foregone earnings are the larger half of the cost of college, so in a single pooled list a two-year degree would outrank almost every bachelor’s on arithmetic alone. That is true and it is the wrong thing to put in front of someone who came looking for a bachelor’s, so the two are ranked in separate tracks and you choose which one you are looking at.

The associate score answers one question: what is this degree worth if you stop here? For a registered nursing or radiologic technology degree that is the right question, because the associate degree is the credential that qualifies you to work. For psychology, teacher education, social work or general studies it is close to the wrong question, because the qualifying credential in those fields is a bachelor’s degree and the associate is a step toward it. Those programs show small earnings premiums here, and the small premium is real but it is not the point of the program.

What we do not yet compute is the transfer pathway. A student who finishes an associate degree at a community college and then completes a bachelor’s has paid community-college prices for the first two years instead of four-year prices. The saving can be substantial and this score never sees it, because each credential is priced on its own. Reading the associate figure as the cost of choosing that route therefore overstates the downside: the honest comparison is between a whole four-year path and a whole transfer path, and that is a calculation we have deliberately deferred rather than approximated.

What this cannot tell you

  • Four-year colleges only. Community colleges are not in this data at all, and that matters: an associate degree in nursing earns a median $82,434 on two years of tuition and would often outrank a bachelor’s degree here. The rankings are therefore tilted toward four-year degrees in a way you cannot see. We are working on it.
  • The comparison wage is statewide, not local. A search that crosses a state line compares against a different yardstick even though the job market is the same. Philadelphia and Camden, New York and northern New Jersey, Kansas City, and Cincinnati and northern Kentucky are all affected.
  • The price is the college’s, not the program’s. Nobody publishes cost by major, so two programs at one college differ only in what they pay. That is reliable for putting programs in order and not reliable as an exact dollar figure.
  • Earnings are measured four years after finishing, for students who took federal financial aid. A program whose students mostly pay cash is measured on a minority of its graduates. Anyone still enrolled in further study at that point is left out entirely, which understates majors whose graduates usually go straight on to another degree.
  • These are not this year’s salaries. They are what the classes of 2018 and 2019 were earning in 2022 and 2023, converted to 2024 dollars. Inflation is accounted for; the job market those students walked into is not today’s. The comparison figure for a high-school graduate is drawn on the same basis and from the same years, so the gap between them is measured like for like even though both levels are dated.
  • Small programs are absent, not bad. The federal government does not publish earnings for programs with fewer than 16 graduates. Every row shows its graduate count, because a median from 44 graduates is a weaker claim than one from 900.
  • Some figures cover several campuses. Where a college reports under an identifier shared with its branch campuses, the row says so and the earnings describe them jointly.
  • It cannot separate the college from its students. A program that pays well partly because of who it admits will look better than it is. Ranking within a selectivity band limits this; it does not remove it.

The federal earnings test

Some rows are flagged as failing a federal earnings test. That means graduates of the program do not out-earn a typical high-school graduate by enough to clear the threshold Congress set, which puts the program’s access to federal student loans at risk. Three hundred bachelor’s programs currently fail, concentrated in fine arts, theatre, music and religion. No nursing program fails, out of 864.

Sources

Earnings and the comparison wage come from federal program-level earnings data. Cost, completion, admissions and enrollment come from IPEDS, the Department of Education’s annual survey of every college. Locations come from IPEDS and the Census Bureau’s zip code centroids. Everything is recomputed when new federal data is released.

The technical record, including every test we ran and failed, is kept with the data rather than on this site.

Why this ranks programs and not colleges. Most of the variation in what graduates earn sits between programs at the same college, not between one college and the next. An average across a whole college therefore averages away the thing that decides the answer, which is why every ranking here is a program at a college and why no college on this site carries a single return figure of its own.