Send the score: what the 2026 test-optional data actually says

2.9 percent. 7.2 percent. Those are the same applicant, at the same elite institution, in the same admissions cycle. With a test score on the file, the chance of admission is 7.2 out of 100. Without one, it is 2.9 out of 100. The applicant is high-achieving and from a disadvantaged background, which is exactly the applicant the test-optional policy was written to help. The gap is not a rounding error; it is the policy's own finding, and it points the other way from the reason the policy exists (Sacerdote, Staiger and Tine).

I count the totals before the argument, so here are the totals first. Three studies came out in 2026 from three different directions, and they do not agree on everything. What they agree on is enough to change a decision an applicant can actually make.

What the three studies are

• One elite institution, applicant-level files. Sacerdote, Staiger and Tine, *American Economic Review: Insights* (September 2026). They read the actual applications at a single elite school and find that high-achieving applicants from disadvantaged backgrounds submit scores at too low a rate once the score becomes optional. Reporting a score more than doubles their admission probability, from 2.9 to 7.2 percent. Their read: an elite office does not treat a score in isolation. It reads the score against the applicant's background, so a number can lift a strong file from a weaker background instead of capping it (AER:I).
• National, institution-level, pre-2020 adopters. RTI International, *Educational Researcher* (March 31, 2026). Where a school went test-optional before the pandemic, the policy is not linked to any overall change in first-year retention or six-year graduation. But split by selectivity the sign flips: at more selective institutions, six-year graduation for men, women and White students rises by about two percentage points; at less selective institutions it falls by about two points for White students (RTI).
• National, IPEDS. Walker and Scott at Tulane (published online August 20, 2026). Total applications are unchanged. Admission rates rise 3.4 percentage points. Yield drops 3.0 points. Racial minority enrollment is unchanged overall; the Black share rises 0.7 points. Their word for it: more admissions, limited movement on the diversity goal the policy was built for (Walker and Scott).

Three studies, one table

Read the first row twice. It is the whole story, and it is the opposite of the slogan.

Three words that are not the same policy

The argument gets flattened because the vocabulary is not precise. These are different policies, and a school's label changes the advice.

• Test-required. A score is mandatory for everyone. The baseline every other policy is measured against.
• Test-optional. A score is not required, and if it is submitted it is read. This is the policy the three studies examine, and the one the decision below is about.
• Test-flexible. A score is not required, but a strong one can be used to stand out in a crowded file.
• Test-blind. A score is submitted and then ignored. Rarer, and a different decision entirely.

If a school is test-optional, a submitted score is read and can work for you. If it is test-blind, a submitted score is thrown away. The same test score, two opposite outcomes, decided by one word on the admissions page.

Why "optional" can backfire

The mechanism is not mysterious. When the score stops being required, a high-achieving student from a disadvantaged background often decides not to send it. The test is expensive, the preparation is uneven, and "optional" reads as "you probably do not need it." The admissions office then reads a missing score as *there was nothing to report*. At a school that reads files in context, the missing score removes the one piece of evidence that could lift a strong applicant from a weaker background.

It shows up most where the preparation gap is widest: the first-generation student, the student from a public school without a test-prep budget, the one whose counselors have not sent scores to elite schools before. For the applicant who already holds a 1550 and a prep course, optional costs nothing. For the one the policy was written for, optional can cost the difference between a file that gets read and one that gets skipped.

That is the gap between the slogan and the data. "Optional" is neutral on paper. It is not neutral in a file. The policy that was supposed to remove the barrier ends up removing the applicant's own best document.

The yield number the offices will feel

The third study's yield figure is the one an admissions office feels first, even if applicants never see it. Yield is the share of admitted students who actually enroll. A 3-point drop, alongside a 3.4-point rise in admissions, means the admitted class is not converting to enrollment at the old rate. The study reports the number; the reason is inference, and I will mark it as such. The likely story is that the marginal admits, the ones admitted because the bar moved, are less certain to enroll, and the aid attached to them may not scale. If a program's yield is sliding, its funding model is about to be tested. That is a question for the office, not a headline.

The column this index does not carry

The honest limit of what this site can tell you: test-optional status is a policy, set school by school and year by year, and it has moved a great deal in the last few cycles. It is not a stable property of a campus the way tuition or degree levels are, so there is no column for it in the index. Do not read a campus page here and assume it is optional or required. Check the admissions page of each school, for the cycle you are applying to, before you decide.

If a school cannot tell you, in one line, whether a score helps or is required in the cycle you are applying in, that is a question worth asking and an answer worth keeping. I write that down too.

The rule, counted out

Strip the studies to the decision an applicant actually faces, and it comes out as a rule with one exception. The exception is the part that surprised me.

1. If your score is above the school's median, send it. A score above the median is evidence, not a risk. The data gives you no reason to hide it.
2. If you are high-achieving from a disadvantaged background, the rule is the same and the reason is the strongest. This is the applicant the policy failed, and this is the applicant for whom a reported score does the most work. 2.9 to 7.2 is not a small number.
3. If your score is far below the school's median, the question is whether sending it costs you a seat or a full ride. That trade is school-specific and is not in any of these studies. Ask the office and get the answer in writing.

The slogan was "optional." The data says the optional decision is the one that matters, and for the applicant the policy was meant to help, the answer is usually to send the number.

Atlas ranks and figures on this page are rounded planning values compiled in 2025/26 for orientation. The study findings are as reported by the named journals and institutions: Sacerdote, Staiger and Tine in American Economic Review: Insights, the RTI analysis in Educational Researcher, and Walker and Scott's IPEDS analysis. Verify each school's current test policy for your cycle before you apply.