Last summer, I wrote a series of posts arguing that the redesigned ACT had created a problem colleges would eventually need to address. The new ACT is shorter, with fewer questions determining each section score, which means individual questions have a greater effect on the final result. That creates more opportunity for section scores to move around from one administration to another even when a student’s underlying ability has not changed very much. Superscoring inflates that variation because students can take the ACT repeatedly, discard unfavorable fluctuations, and preserve favorable ones. Take the test enough times, and eventually the right combination of section scores can fall into place. That was the idea behind what I called the ACT slot machine.
In an earlier part of this series, I argued that colleges had a relatively simple way to reduce this problem: stop superscoring the ACT. At the time, that was mostly a prediction about how colleges might eventually respond to the redesigned test. I suspected that admissions offices would eventually have to decide whether a composite assembled from the best section scores across several administrations still represented what they wanted an ACT score to represent. We now have at least one confirmed example of a highly selective college moving in exactly that direction.
Swarthmore Ends ACT Superscoring
As recently as July 2025, Swarthmore College explicitly superscored the ACT. Its standardized-testing policy stated that the admissions committee would consider a student’s highest submitted Composite and subsection scores and would recalculate a new Composite using subsection scores earned on different test dates.
By August of 2025, that policy had changed. Swarthmore no longer superscored the ACT and instead considered the highest Composite earned on a single test date. [Given that I had argued just a month earlier that colleges should stop superscoring the ACT, I would of course like to take full credit for the change; unfortunately, Swarthmore has not yet acknowledged my influence.]
Why Did Swarthmore Change Its Policy?
Swarthmore has not publicly said, at least as far as I can find, that it stopped superscoring because of the redesigned ACT. There is, however, some evidence about what the admissions office was thinking during this period. Three months after the superscoring reversal, The Phoenix, Swarthmore’s student newspaper, published an article about the college’s decision to continue its test-optional policy. In that article, Dean of Admissions Jim Bock discussed the uncertainty created by changes to standardized testing, and the ACT in particular. The article noted that the redesigned ACT had not yet been sufficiently studied for reliability or incorporated into updated concordance tables, and Bock said that “it is unclear when new validity testing is occurring.” The article described the admissions office as hesitant to rely heavily on assessments that were changing rapidly.
If we read between the lines, that connection makes sense. Superscoring becomes more problematic precisely when an admissions office has questions about how much meaning to attach to variation in individual section scores. A superscore collects across multiple administrations the highest observed score in each section. If some portion of those differences reflects measurement noise rather than a meaningful change in student ability, superscoring systematically preserves that noise in one direction. The strongest observed result survives, while the weaker result disappears.
Suppose, for example, that a student’s underlying Reading performance is roughly equivalent to a 32. On one administration she scores a 30, and on another she scores a 34. That four-point difference does not necessarily mean her reading ability changed dramatically between the two Saturdays; some portion of the movement can simply reflect normal testing variation. If a college looks at a single-sitting composite, both good and bad variation remain embedded in that score. A strong Reading result helps, but it has to coexist with whatever happened in English and Math on the same test date.
Superscoring changes the statistical dynamic. The 30 effectively disappears while the 34 can survive permanently, and the student can repeat the same process in English and Math. After enough administrations, the resulting superscore can become meaningfully higher than the score the student would typically produce on any one test date, even if the student’s underlying level has remained relatively stable. The more volatile individual section scores become, the more valuable repeated testing and superscoring become together.
The SAT-ACT Distinction Is the Most Interesting Part
Perhaps the most notable aspect of Swarthmore’s reversal is what the college did not change. Swarthmore continues to superscore the SAT. This is therefore not a broad rejection of superscoring or simply another manifestation of the college’s test-optional philosophy. Swarthmore has chosen to treat the ACT and SAT differently.
That distinction is central to the argument I have been making throughout this series. Colleges frequently treat the SAT and ACT as interchangeable admissions measures, but there is no reason that their score-use policies must be identical if the statistical behavior of the tests differs. In earlier parts of this series, I argued that repeated testing creates a particularly substantial superscoring advantage on the ACT. If the redesigned ACT also creates unresolved questions about reliability and validity—the exact concerns Swarthmore’s dean of admissions subsequently raised—then superscoring the SAT while using the highest single sitting on the ACT is a coherent policy choice.
Again, I cannot say that this was Swarthmore’s stated rationale. The college has not publicly connected those dots. But we now know the chronology. Swarthmore was superscoring the ACT in July 2025. By August, it had stopped. In November, its dean of admissions was publicly talking about uncertainty surrounding the redesigned ACT’s reliability, validity testing, and comparability with the SAT. That does not prove causation, but it is more suggestive than a policy change occurring in a vacuum.
Is This the Beginning of a Trend?
One college is not enough to establish a trend. I have spent time looking for other selective institutions that recently stopped superscoring the ACT, and the evidence becomes messy very quickly. Third-party databases can be stale, college websites are sometimes vague, and the rollout of the redesigned ACT has coincided with unrelated changes in test-optional and test-required policies. I am therefore not prepared to say that colleges broadly are abandoning ACT superscoring.
But Swarthmore moves this question beyond the theoretical. When I started writing this series, ending ACT superscoring was a policy response that I thought colleges should consider. We now know that at least one highly selective institution actually made that change, and we know almost exactly when it happened. Swarthmore was still superscoring in July 2025 and had stopped by August.
Whether other colleges follow is now the interesting question. If other selective institutions begin retaining SAT superscoring while moving the ACT to a highest-single-sitting policy, Swarthmore may turn out to have been an early example of a broader reassessment. If no one follows, this may simply be an idiosyncratic decision by one admissions office. Either way, the possibility that colleges might respond to the redesigned ACT by abandoning superscoring is no longer hypothetical. The first confirmed reversal has happened.