If you registered for the October 3 SAT today, the last day College Board's regular deadline allows, you have about fifteen days. The question almost everyone in that position asks is whether those days can be compressed into two long weekends of review. They can be. It just will not buy you much.

The clearest evidence on this is a 2013 review in Psychological Science in the Public Interest by John Dunlosky, Katherine Rawson, Elizabeth Marsh, Mitchell Nathan and Daniel Willingham. They graded ten common study techniques on how well each holds up across ages, subjects and test formats. Two earned a "high utility" rating: distributed practice, meaning the same material revisited across separated sessions, and practice testing, meaning trying to produce an answer from memory. Rereading, highlighting and summarizing were rated low utility, despite being what most students reach for first.

Cramming is massed practice, the opposite of the first winner, and it usually involves rereading, one of the losers. That is the whole case in two sentences. What follows is why it holds, and what fifteen days can still do for you.

Why the crammed version fades

A long single session does something real: it stabilizes material in the short term. What it does not do is test whether the material survives without the session's own scaffolding around it.

When you review a concept at 11 p.m. and it comes back instantly, part of that ease is the concept still being on screen in your head. The context is fresh. Strip the context away, come back Friday, and the recall is harder. That difficulty is not proof the first session failed. It is the first honest measurement of whether anything durable was built.

This is why fluency is a bad instrument. Dunlosky's team noted that the low-utility techniques stay popular precisely because they produce a subjective feeling of smoothness that does not track long-term retention. You feel prepared at midnight and you are measuring the wrong thing.

Shana Carpenter, a psychology professor at Oregon State who has published more than 70 articles on spacing and retrieval, puts the tradeoff in four words.

Easy learning is easy forgetting.

Shana Carpenter
Professor of Psychology, Oregon State University

Carpenter's rule of thumb is that there is no magic interval. What matters is leaving enough time between sessions that the material is not still fresh when you return to it. In her framing, three hours in one sitting is worse than 30 minutes across six days, even though the clock reads the same. UC San Diego's psychology department, which calls the spacing effect one of the most replicated findings in learning science with more than 200 supporting studies, describes the pattern it argues against with uncomfortable precision: little preparation for weeks, then intense cramming two or three days out.

What happens when someone counts

Three studies put numbers on the gap, and none of them is about the SAT. Read them for the size of the effect, not for a promise about your score.

A 2025 experiment in an introductory programming course randomized 143 students so their practice dashboard counted either days practiced or total questions answered. The days group averaged 38.5 practice days against 17.3 for the questions group, while total questions came out nearly identical, 415.0 against 392.5. Final exam scores were 85.3 percent against 81.7 percent. Same volume of work, different distribution, meaningful difference in outcome.

A systematic review in The Clinical Teacher pooled 21,415 medical learners and found a standardized mean difference of 0.78 favoring spaced repetition over standard study on objective tests, with a 95 percent confidence interval of 0.56 to 0.99. That is a medium-to-large effect, on multiple-choice exams, in a population that crams for a living.

A 2026 study in the Journal of Experimental Psychology: Applied surveyed 185 and then 194 college students after their exams. Asking whether students had "spread out" their studying predicted nothing, which is the usual null result. Asking whether they had gone back and restudied the same concepts across sessions predicted exam scores in every exam measured, even after controlling for total study time.

Spacing is not the same as studying a different topic each day

This is the confusion that quietly ruins two-week plans. Cognitive scientist Cindy Nebel of The Learning Scientists points out that the effect measured in the laboratory is a spaced review effect: the same material, revisited, on separated days. Doing algebra Monday, reading Tuesday and grammar Wednesday is a timetable, not spacing. Nothing gets revisited.

Cramming A rotating topic schedule Spaced review
Same material seen twice? Twice in one night No Yes, on separate days
What it's good for Short-term stabilization Coverage Durable recall
What it feels like Fluent, then gone Productive Harder than it should be

Where this evidence gets thinner than the headlines

None of the studies above tested the SAT. They tested medical exams, a programming course, college finals and word pairs in a laboratory. The spacing effect is robust enough that assuming it transfers is reasonable, but the size of the gain on a digital SAT reading-inference question is not something anyone on this list has measured.

There is also a boundary worth knowing. A six-experiment 2025 study in the Journal of Experimental Psychology: Learning, Memory, and Cognition found that retrieval practice gave no advantage over restudy for novel, meaningless visual objects, and sometimes did worse, across feedback, delay and format variations. A 2026 health-professions study found retrieval benefits varied by content set, with at least one showing no advantage. Retrieval and spacing are strongest on meaningful material with an existing framework behind it: vocabulary, formulas, grammar rules, question types you can name. They are least proven on material that has no structure to hook into.

What would settle it for the SAT specifically is a randomized trial with distributed versus massed prep and official scores as the outcome. As far as these sources go, that does not exist.

The fifteen days you have left

Fifteen days is not nothing. It is enough for roughly ten separate review days, which is the variable the programming experiment found was doing the work.

  • Take one full practice section early, this week

    Not to feel bad. To generate a list of specific misses you can return to.

  • Study 30 to 60 minutes on most days rather than three hours on two days

    UCSD suggests one to two hours at regular intervals; Carpenter's comparison is 30 minutes across six days beating three consecutive hours.

  • Return to the same missed questions three or four days later

    Long enough that the answer is not fresh. This, not variety, is what the research measures.

  • Answer before you read the explanation

    Rereading an explanation produces recognition. Producing the answer produces retrieval, the other high-utility technique in Dunlosky's ranking.

  • Then read the explanation, especially when you were wrong

    In a 2026 study in Autism covering 240 adults, items missed at first attempt were later recalled at 0.47 accuracy with post-retrieval feedback and 0.06 without it. Checking the answer is not optional; it is most of the effect.

  • Spend the final days on pacing and review, not new content

    Split roughly 75 percent new work and 25 percent review earlier in the window, and flip toward review as October 3 approaches.

Two honest exceptions. If you are starting from zero on October 1, a single concentrated session is the least-bad option available, because massed practice does stabilize material briefly and briefly is the horizon you have. Just know what you bought. And if you are registering today mainly out of panic, the November 7 date has a regular deadline of October 23, which buys five more weeks of distinct study days for the cost of a later score release, typically two to three weeks after the test.

The pitfall that swallows most two-week plans is the one in step four. A student finishes a practice test, opens the answer key, reads twelve explanations, nods at each one, and logs two hours of study. Every explanation felt clear. None of them was retrieved. That is rereading with extra steps, and Dunlosky's team put rereading in the bottom tier for a reason.