SAT–ACT Percentile Comparison

My leading explanation is the changing ACT-taking population—especially the shrinking share of voluntary weekend testers—with genuine declines in academic performance also contributing. The evidence for both is stronger than the evidence that earning a particular ACT score has become intrinsically harder. ACT

A rising percentile at the same numerical score simply means a larger percentage of the reference group now scores at or below that level. It does not, by itself, distinguish a different group of students from worse performance among comparable students. ACT

The strongest population clue

Between the graduating classes of 2019 and 2025, ACT reports:

Group, classified by most recent test	Share in 2019	Share in 2025	Average score, 2019 → 2025
School-day testers	40%	69%	18.1 → 17.8
Weekend testers	60%	31%	22.4 → 23.0

The higher-scoring group became much smaller, even while its own average increased. ACT identifies changing school-day testing coverage and test-optional admissions as contributors to the participation shift. ACT

Applying the newer proportions to the older subgroup averages gives approximately 19.43, almost exactly the actual 2025 overall average of 19.4. That makes population composition a substantial explanation, not a speculative technicality. But this calculation concerns the mean, not the exact percentile movement at every score. ACT

Conceptually, this is a shift away from a more self-selected admissions-testing population toward a broader school-assessment population. Giving more weight to a lower-scoring distribution can raise the percentile of an unchanged score without making that score harder to earn.

But actual performance declines are also documented

ACT conducted a more controlled study using school-day testing at schools with high participation, adjusting for demographics, participation rates, test dates, and school differences. Relative to pre-pandemic cohorts, it found adjusted Composite-score declines of 0.66 points in 2021 and 0.59 points in 2022. That is evidence of deterioration beyond the simple school-day/weekend mixture change. ACT

An important qualification to the reweighting calculation: the school-day and weekend groups are not fixed populations either. Changes within each group can conceal achievement declines. Their relatively stable averages therefore do not establish that learning loss was negligible.

What I would conclude

Changing participation is my first explanation; genuine achievement decline is an additional, independently supported explanation. I would not assign percentages to their respective contributions from these aggregate tables. ACT

The important distinction is that a score can become rarer without representing more skill than it previously did. That can happen because stronger scorers comprise a smaller share of the testing pool, because comparable students perform worse, or both. Neither mechanism, on its own, means the ACT-to-SAT conversion should change.