Earnings data is the gravitational center of choosing what to study. It is also the single most misread number in the entire decision, because it is almost always presented as one figure, a median, and one figure invites the reader to treat it as a personal prediction. It is not. Read honestly, with its range and its caveats in view, earnings data is a powerful guide to where a path leads. Read as a single point, it distorts the decision. This guide explains what the number hides and how to read it well, as a foundation for How to Choose a Major.
The Median Is a Starting Point, Not a Prediction
The figure you see most often is the median, and the first honest move is to understand exactly what it claims.
Definition
Median earnings
The middle value in the distribution: half of graduates earn more, half earn less. It is more honest than an average, which a few very high earners can inflate, but it is still a single point. It tells you the center of the range, not where any individual graduate will land within it.
A student who reads "median earnings: $70,000" and plans on $70,000 has made an unwarranted leap. The median says half earn more and half earn less; it says nothing about which half this student lands in, which depends on the specific program, the career chosen, the industry, the region, and the student's own performance. The median is the right place to start a comparison and the wrong place to end a prediction.
The Three Things a Single Number Hides
Behind every median earnings figure sit three distortions that a single number cannot show.
The range
A median averages over a wide spread. Two majors with the same median can have completely different ranges, one tight and predictable, one spanning from low to very high. The range is often more decisive than the median, and the median hides it entirely.
Geography
National figures average across regions with very different pay and living costs. A high median may reflect graduates clustered in expensive coastal cities, where the higher salary is partly eaten by higher rent. The number is not adjusted for where the money is actually earned.
The time lag
Earnings are measured years after entry, often a decade out, so the data describes graduates who started school long before today's applicant. It reflects an earlier cohort in an earlier economy, which is the best available signal but not a current one.
None of these makes the data useless. They make it a signal to be interpreted rather than a fact to be copied. A reader who keeps all three in view reads earnings figures the way they should be read: as a calibrated estimate of where a path tends to lead.
Use the Range, Not Just the Median
The single most important upgrade in reading earnings data is to look at the percentile band instead of the median alone.
Every major and program on the site reports a 25th-to-75th-percentile range alongside the median. That band is the honest picture: a quarter of graduates earn below the bottom, a quarter earn above the top, and half land inside. A field with a high median and a narrow band is a different bet from a field with the same median and a band three times as wide. The first is predictable; the second is high-variance, with strong upside and real downside. The median treats them as identical. The range tells them apart.
This is why Major vs Program vs Career insists on dropping from the major level to the program level: the program-level range is far more informative than the major-level median, because the major averages over programs that diverge. Read the band for the specific program the degree will actually be in.
The width of the band is itself a piece of information, and most readers ignore it. A narrow band says the path is predictable: graduates land close together no matter which corner of the field they end up in, so the median is a fair estimate of what one person can expect. A wide band says the opposite. It says the outcome depends heavily on choices made after enrollment, on the specific role, the employer, and the region, and that the median is a poor stand-in for any individual. When two fields show the same median, the one with the narrower band is the safer bet and the one with the wider band carries both more upside and more downside. The median alone cannot tell you which is which, so reading the median without the band is reading half the number.
There is a quieter reason to favor the range. A median is a single point, and a single point invites a single mental story: "I will earn this." A range refuses that story. It forces you to hold two numbers at once, a floor and a ceiling, and to ask honestly which end of the band your own plan resembles. That small act of holding two numbers instead of one is most of what reading earnings data honestly means in practice.
Self-Selection: Why the Number Is Partly About the Students, Not the School
The hardest caveat to see is that an earnings figure measures the graduates, not just the education. The two are tangled, and pulling them apart is where most casual readings go wrong.
Definition
Self-selection
The effect that arises when the people who choose a given path are systematically different from those who do not, so the outcome reflects who entered as much as what the path did to them. A program that admits high earners will report high earnings partly because of whom it enrolled, not only because of what it taught.
A selective school reports strong graduate earnings. Some of that is the education, the network, and the credential. But some of it is simply that the school admitted students who were already on a high-earning trajectory: stronger preparation, more family resources, more access to internships and connections. Those students would have earned more than average almost anywhere. The earnings figure cannot separate the part the school added from the part the student brought, and it never claims to. It reports the outcome, not the cause.
This is why a high earnings number is not, by itself, proof that a school or program will lift any individual student. The honest question is not "what do graduates of this program earn" but "what would a student like me earn here versus somewhere else," and no single median answers that. The figure is still useful, because where graduates land is real information about where a path tends to go. But treat it as a description of who comes out the other end, not as a guarantee of what the path will do for you specifically. The same caution applies in reverse: a program with a modest median may be doing excellent work with students who started from further back, and the raw number hides that entirely.
A Worked Example: Reading One Field Without Fooling Yourself
Abstract caveats are easy to nod at and easy to forget the moment a real number appears. Walk a single field through the honest read and the discipline becomes concrete. Take Computer Science, a field whose major-level median looks unambiguously strong.
Start at the major and resist the obvious move. The headline figure on the major profile is a midpoint, built from every program and every graduate underneath it, and the distance from the bottom of that distribution to the top is wide. A reader who stops here has learned only that the field, on average, pays well. That is true and nearly useless for a decision, because no one earns the average of a field; they earn the outcome of a specific program and a specific career.
Drop to the program level and the single number splits apart. A theory-heavy computer science track, an information-technology track, and a software-engineering track all sit inside the same major, and the labor market pays their skills differently. The program-level range on each college profile shows the split that the major median hid. Two students who both "studied computer science" can finish far apart, and the program-level band is the first place that gap becomes visible.
Now apply the caveats in order. The range tells you how much the specific program and role will decide the outcome. Geography tells you that a high figure may reflect graduates clustered in expensive tech hubs, where the larger paycheck is partly absorbed by rent, so the number is not as far above a lower-cost region as it first looks. The time lag tells you the figure describes people who entered college roughly a decade ago, in a different hiring market, which matters in a field that moves as fast as this one. And self-selection tells you that part of the strong number reflects who the field attracts, not only what it does for them.
None of that argues against the field. It produces a far better read than the headline: a calibrated sense of where the path tends to lead, what could move you up or down inside it, and what the number is and is not promising. A student who followed the field down to the software developer career, read the range, and weighed it against cost and growth made a decision. A student who saw "computer science pays well" made a wish. The figures were the same. The reading was not.
What the Federal Data Actually Measures
Reading honestly also means knowing the sample, because the source has boundaries that shape what the numbers can claim.
The College Scorecard earnings data, which feeds the figures on this site, covers students who received federal financial aid. That is a large and representative sample, but it is not every graduate; it skews toward middle-income students and excludes those who never took federal aid. It also measures earnings at fixed points after entry, counting everyone who enrolled, which makes it more honest than alumni-survey figures that quietly drop the people who did not do well. These boundaries are documented on the data sources page, and they argue for treating the numbers as a strong representative signal rather than a complete census.
Two features of this measurement deserve a closer look, because they cut in opposite directions and most readers feel only one of them. The first is who is counted. Because the sample is built from students who took federal aid, it leans toward middle-income graduates and leaves out both those who never borrowed and those whose families paid the full cost. That does not make the number wrong, but it does make it a number about a particular slice of graduates, and the slice is worth remembering when you compare a figure across two very different schools.
The second feature is a quiet strength. The figure counts everyone who enrolled and shows up in the wage records, not only the ones who finished or the ones who answered a survey years later. Alumni-reported salary numbers, the kind colleges sometimes publish in their own marketing, suffer from survivorship bias: the graduates who did well are more likely to respond, and the ones who struggled or left the field quietly drop out of the sample, dragging the reported average upward. The federal figure does not let people opt out of the count. That is exactly why it can look lower than a school's own published number while being the more honest of the two. When a college's self-reported earnings sit far above the federal figure for the same program, the gap is usually survivorship, not deception, and the federal number is the one to trust.
For a fuller walkthrough of where each figure on a college page comes from and how to read the page as a whole, see Reading a College Scorecard Page.
The Mistakes Honest Readers Avoid
Misreading earnings data is not random. The same handful of errors recur, and each has a clean fix.
The first is planning on the median as a personal salary. A student reads a field's median, mentally pencils that number into a future budget, and treats the rest of the decision as settled. The median says half earn more and half earn less; it says nothing about which half this student lands in. The fix is to plan against the bottom of the percentile range, not the middle. If the path still makes sense when you assume the lower end, it is a sound choice. If it only works at the median or above, you are betting on a result you cannot control.
The second is comparing a major median to a program median as if they were the same kind of number. A broad major average and a specific program average answer different questions, and lining them up side by side produces a false comparison. The fix is to compare like with like: program to program, or career to career, never a broad field against a narrow specialty. The Major vs Program vs Career guide explains why the levels are not interchangeable.
The third is ignoring the cost of living behind a high number. A figure that looks impressive may belong to graduates packed into the most expensive metros in the country, where the larger paycheck buys a smaller life. The fix is to read every earnings figure against where graduates actually work, and to convert a national number into a real one by asking what it buys in the places that path tends to lead.
The fourth is treating an old figure as a current one. The data describes a cohort that entered college years ago, and a field that has changed since then will mislead anyone who reads the number as today's reality. The fix is to pair the backward-looking earnings figure with a forward-looking one. Job Growth Projections covers how to read the Bureau of Labor Statistics outlook, which estimates where a career is heading rather than where it has been.
The fifth is using earnings as the only number in the decision. A high median attached to a field that is shrinking, or that costs far more to enter than it returns, is not a good outcome. The fix is to hold earnings alongside cost and growth, never alone. The ROI Calculator exists precisely to keep the earnings number tied to the price of getting it.
Every one of these mistakes shares a root: treating a summary statistic about a group as a forecast about an individual. The discipline that prevents all five is the same. Read the range, match the level, adjust for place and time, and never let the earnings figure stand alone.
How to Actually Read an Earnings Figure
The honest read is a short, repeatable sequence. Run it every time a single earnings number appears, and the common mistakes mostly disappear on their own.
- Identify the level. Is this number for a broad major, a specific program, or a career? A major median is a starting hypothesis; a program or career median is closer to a real answer. If it is a major figure, drop a level before trusting it.
- Find the range. Locate the 25th-to-75th-percentile band that sits alongside the median on every major and program profile. The width of that band tells you how much the specific path will decide the outcome.
- Anchor your plan to the lower end. Ask whether the decision still holds if you land near the bottom of the range rather than the middle. A choice that only works at or above the median is a bet, not a plan.
- Adjust for geography. Ask where graduates of this path actually work and what the number buys there. A high national figure tied to expensive metros is worth less than it first appears.
- Account for the time lag. Remember the figure describes a cohort from years ago, and pair it with a forward-looking job-growth projection for the career it leads to.
- Tie it to cost and growth. Put the earnings figure next to the price of the degree and the outlook for the career. Earnings alone is one factor of several, not the verdict.
Six steps, a couple of minutes, and the number stops being a slogan and becomes a forecast you can actually stand behind. This is the same sequence the ROI Calculator and Career Path Explorer automate; running it by hand once is what teaches you to trust the tools' output and to spot when a headline figure elsewhere has skipped a step.
Putting It Together
Reading earnings data honestly comes down to a short discipline:
| Do | Instead of |
|---|---|
| Read the 25th-to-75th-percentile range | Reading the median alone |
| Look at the specific program's data | Using the broad major average |
| Weigh earnings against cost and job growth | Treating earnings as the only factor |
| Treat the figure as a calibrated estimate | Treating it as a personal guarantee |
Used this way, earnings data does exactly what it should: it compares the realistic ranges of different paths and feeds the decision in How to Choose a Major alongside interest, aptitude, and job growth. The ROI Calculator pairs the earnings figure with cost, and the Career Path Explorer connects programs to the careers whose earnings you are reading.
Where This Fits
This guide is the data-literacy foundation of the choosing-what-to-study cluster. It supports the earnings-validation step in How to Choose a Major and pairs with Job Growth Projections for the forward-looking side of the picture. For the deeper meaning of the most common figure of all, see What Median Earnings 10 Years Out Actually Means, and for the question of how much weight earnings should carry against everything else, see Passion vs Paycheck. The honest reader's rule is simple: the median starts the conversation, the range informs the decision, and no single number predicts an individual outcome.
