West Virginia's Common Data Set Reports Two Different Acceptance Rates for the Same Class

West Virginia University's 2024-25 Common Data Set reports how many students it admitted for Fall 2024 in two places, and the two places do not agree. One says 15,570 admits out of 20,150 applications, which is a 77.27% acceptance rate. The other says 17,155 admits out of 20,148 applications, which is 85.14%. Same school, same entering class, same PDF. The headline acceptance rate moves 7.87 points depending on which page you read.

That is the largest example of a defect we found at eight schools while building our Real Acceptance Rates series, across nine separate disagreements. Before going further, the honest framing: most of the other eight are tiny. Several are a single student. The reason to publish them together is the pattern, not the size of the gaps, and anyone who reads this as eight schools with a West Virginia-sized problem has read it wrong.

How One Section Ends Up With Two Answers

Section C1 of the Common Data Set is the part everyone quotes. It asks for the number of applicants, the number admitted, and the number who enrolled for the fall entering class. The standard form asks for those counts broken out by sex — men, women, and in some filings a third row for another or unknown category. Many schools also fill in an optional supplement that asks for the same three counts broken out by residency.

Those are two cuts of one group of people. Every applicant has a sex category and a residency category, so both sets of rows describe the identical cohort from two angles, and adding up either should land on the same total. When a school fills in both and they disagree, the filing is telling you two different things about one class.

This is not the Early Decision problem we wrote about earlier in this series. That defect is a wrong label: the cohort year printed at the top of Section C21 is part of the form's own question text, inherited from whichever version of the template a school filled in, and often a year behind the filing it sits inside. The school did not type it. You can read the whole story in our breakdown of the Common Data Set's Early Decision section.

What is on this page is different in kind. No form template explains two conflicting sets of numbers in one section of one filing. The counts are entered by the institution, and the disagreement is between things the institution entered.

The Nine Disagreements

Every figure below comes from the school's own Common Data Set filing.

Eight schools, nine disagreements. NC State appears twice because its admitted and enrolled counts each disagree independently.

Read the gap column before anything else. Haverford's enrolled count differs by one student out of 386. Drexel's differs by one out of 1,948. Buffalo's applied, admitted, and enrolled counts each differ by exactly one. Those are rounding-scale discrepancies in a survey that runs dozens of pages, and they change no rate anyone would ever cite. A one-student gap in a form this long is unremarkable, and we would not have written anything if that were the whole finding. Oberlin's six-applicant gap and Brandeis's four-admit gap sit in the same category — worth recording because we record everything we cannot reconcile, not because they change what an applicant should do.

The three at the top are different. UIC's Fall 2023 filing prints an enrolled total of 4,697 that its own components sum to 4,647, a 50-student gap internal to a single line of arithmetic. NC State's Fall 2024 filing disagrees with itself by 446 on enrolled students and 180 on admits. And West Virginia's disagrees by 1,585 admits, the only one here large enough to move a published acceptance rate by an amount a family planning a college list would notice.

Why West Virginia Is the One That Matters

The other eight rows are a data-quality note. West Virginia's is a number problem, because the two answers produce two different acceptance rates and both of them are printed by WVU for the same cohort:

A school with a 77% acceptance rate and one with an 85% acceptance rate describe meaningfully different admissions environments, and WVU's own filing supports either description.

We publish the residency figure in our West Virginia acceptance rate breakdown, and we want to be precise about why. Two reasons, both modest on their own: the residency table's four components sum exactly to its own totals, and 15,570 is continuous with the 15,923 WVU reported the prior year, while 17,155 would be a sharp break. That is a judgment about which figure looks more likely to be right, not a resolution. West Virginia publishes nothing that settles which number is correct, and we do not claim to know. If WVU clarifies, we will change what we publish.

The same caution applies to every other row in the table. At none of these schools do we know which figure is the true one. We know only that they cannot both be.

What This Does Not Mean

No motive is available here and none should be offered. Filling out the Common Data Set is a long clerical exercise, the residency breakdown is an optional supplement, and an optional field that is blank, partially filled, or pulled from a slightly different query than the required field is an entirely ordinary outcome. Nothing in this data indicates that any of these schools set out to publish a favorable number, and we are not suggesting it.

It also does not mean these eight institutions are unreliable in general, or that the Common Data Set is a bad source. It remains the most detailed public admissions data that exists, this entire series is built on it, and most of the filings we have read hold together fine. The takeaway is a narrow one: a Common Data Set figure is not self-verifying. Being printed in an official filing does not mean it survived a check, because in these nine cases the filing does not agree with itself.

The scope is eight schools among the filings in our Real Acceptance Rates corpus where both the sex breakdown and the residency supplement were filled in completely enough to compare. That is not a random sample of American colleges, and we are not extending any ratio beyond the named schools.

The Check You Can Run Yourself

This takes about three minutes per school and needs nothing but the school's own PDF.

1. Find the school's Common Data Set — usually on the institutional research office's website — and open Section C1.
2. Locate the sex breakdown. Add the men, women, and other or unknown rows for applicants. Do the same for admits and for enrolled students.
3. Locate the residency supplement below it, if the school filled it in. Add its rows for the same three counts.
4. Compare the two sets of totals, line by line. They are describing one cohort, so they should be identical.
5. While you are there, check each printed total against its own components. UIC's 50-student gap shows up on that step alone, without needing the residency table at all.

If the two sets disagree, the school's own filing does not agree with itself, and any acceptance rate computed from either side is uncertain by that margin. A one- or two-student gap is noise. A gap in the hundreds or thousands means the rate you were about to quote has a range, not a value, and you should say so if you use it.

Where to Go From Here

If you are building a list around published acceptance rates, the practical version is short: a rate is a claim, and claims can be checked. Before a number shapes where you apply, spend three minutes confirming the filing it came from is internally consistent. Our free college list builder works from rates we have already run this check against, and our application deadline tracker keeps the dates straight once the list is set. If you would rather talk through how much weight a school's published rate deserves in your case, a free strategy call is a good place to start.