You look up a college, find a middle 50% of 1400 to 1510, and compare it to your 1380. The comparison feels precise. It is not, and the reason is who is in that range.

The middle 50% a college reports in its Common Data Set describes the enrolled students who submitted that score. It is not a description of the entering class, because at a lot of colleges most of that class is not in it. Some colleges publish an admitted-student range instead, which is a different group again, so the first thing to establish is which one you are looking at.

Fewer than half the class had a score reported

GoodGoblin's College Admissions Data Report 2026 covers 1,034 colleges, with per-college score-submission shares drawn from Common Data Set and federal survey reporting. It publishes its underlying data, so we recomputed its headline figures ourselves rather than taking them on trust. They hold:

• 763 colleges report both an SAT submission share and an ACT submission share. Separately, 14 colleges in the dataset carry zero for both tests and are not among those 763, so a zero there is not being counted as a college that reported no submissions.
• At 487 of those 763, the two shares added together come to under 50%, so for a majority of the entering class neither test was part of what the college reported.
• Across the 684 of the 1,034 colleges with a comparable fall 2018 federal figure, the median share submitting an SAT was 65% for the fall 2018 class and 21% for each college's latest reported class.

One university in the set reports an SAT middle 50% of 1400 to 1510. For its fall 2025 entering class the dataset records about 16% of enrolled students submitting an SAT and about 6% submitting an ACT. Those are the compilation's figures as it recorded them from the university's reporting, not numbers we read off the university's own publication. We have not named it, though with the dataset linked above it is findable; the reason for leaving it out is that we have not checked its current requirements against its own admissions page, and a post about misreading published figures is the wrong place to be careless with one. So that SAT range summarizes the roughly one in six enrolled students who sent an SAT, not the class.

What the range is actually a sample of

Start with what is true by definition rather than by inference. The Common Data Set percentiles a college publishes are computed from the enrolled students who submitted that test. An admitted-student range, where a college publishes one, is computed from a different group, but the same point applies to it: it describes submitters within that group. That is how the figures are defined, so a range built from 16% of a class is a description of those students and of nobody else.

What the range would look like if the other 84% had submitted is not knowable from this data. The dataset records how many students sent a score, not what the scores of the students who did not send one would have been. Anyone telling you that non-submitters would have scored lower is reasoning from an expectation, not from these figures, and we are not going to do it either.

The point that survives without that assumption is narrower and still useful: a published middle 50% at these colleges is a summary of a minority of the class, so treating it as a portrait of the people who got in is a category error regardless of which way any selection runs.

Three things the range cannot do

A percentile is not a cutoff. A score below the 25th percentile is below three quarters of the submitted scores in that class. That is not the same thing as a minimum. A percentile is a description of a distribution, so if you need to know whether a minimum exists, look for it in the college's own stated requirements rather than inferring it from a range.

It cannot tell you whether submitting helps you. This is the one people most want. Submission shares and score ranges describe who submitted and what those scores were. They do not identify what would have happened to the same application without the score, and no aggregate of this kind can.

What the range does tell you is real, just smaller than people assume: where your score sits among the scores that were submitted at that college. That is genuine information. It is a statement about a distribution you can see, rather than about a decision you cannot.

It does not tell you the current policy. Submission data describes a class that already enrolled, so it lags. A college can change what it requires between the class the data describes and the one you are applying to.

Four checks before you send a score

The range is still useful. It is just answering a narrower question than people ask of it.

1. Check the requirement on the college's own admissions page, for your application year, not from a summary. Required, recommended, considered and not considered are four different things, and only the college can tell you which one applies to you.
2. Find the submission share if the college publishes one. It tells you how much of the class the range actually describes. A range covering a fifth of the class is a thinner description than one covering nearly all of it, which is worth knowing before you compare yourself to it.
3. Ask what the range is a range of. An SAT middle 50% describes the students who submitted an SAT. A college can publish an ACT range alongside it, covering a different and possibly overlapping group of students, and the two shares do not have to add up to the share of the class that submitted anything.
4. Do not try to settle the submit-or-not question from these numbers. They cannot answer it. If a college requires a score the decision is made for you. If it does not, the useful inputs are its own guidance and your own score, and the range is worth no more than the share of the class it covers, which varies enormously between colleges.

These figures are recomputed, not re-collected

The figures above come from one organization's compilation of federal and institutional reporting. We recomputed from its published dataset every number quoted in this post: the 763, the 487, the 684, both medians, the 14 zero rows, the 791 carrying a submission share, and the cohort counts behind the fall 2020 to fall 2025 span. Each matched or was counted directly. That is worth something: the arithmetic is right and the selection rules are inspectable.

It is not independent validation. We did not re-collect a single college's figure from its own Common Data Set, so if the compilation mis-recorded a value we would have reproduced the error faithfully. We also name no college here, because our rule is that a named college gets its current policy checked against its own admissions page first, and that check has not been done.

The data is also older than it looks and mixed in vintage. Each college contributes its own latest available figures, and counting those cohorts in the dataset gives a span from fall 2020 to fall 2025, with the large majority, 598 of the 791 colleges carrying a submission share, reporting a fall 2023 class. So the 487 and the 21% median describe a period rather than one admissions cycle, the 65% is fall 2018, and no part of it establishes any college's policy for the year you are applying.

And none of these figures can tell you whether submitting a score helps or hurts your own application. Submission shares and score ranges do not identify that, so anyone answering it confidently from this kind of evidence is going beyond it.

If you are weighing whether a score is worth sending at all, the companion question is what the test is actually measuring about you, which we wrote about in why practice questions expire even when their answers are still correct.