Study

The way you're studying for the OITE probably isn't working

Omar Samara · 2026-09-07 · 4 min read

Nobody needs telling to do questions. The question is what happens in the four weeks before the exam, when the reading gets long and the tab with the question bank in it stays open but stops being used.

Rereading is what people actually do

Surveyed on their own study habits, 177 students were asked to list what they do. Rereading notes or the textbook came out on top at 83.6%. Self-testing came out well behind it.

The authors' explanation is the useful part. Rereading produces what they call an illusion of competence. The second pass through a page is smoother than the first, and that smoothness gets misread as knowing the material.

The experiment that separates the two

In 2006, students read prose passages and then either took recall tests on them or restudied them the same number of times. Everyone was tested again later — after five minutes, two days, or a week.

At five minutes, restudying won. At two days and a week, testing won, and won clearly.

Then the part worth sitting with: restudying made students more confident in their ability to remember the material, while it was leaving them with less of it. Confidence moved one way and retention moved the other.

It holds up on residents, at six months

This is not only a psychology-lab result. A randomised controlled trial ran it on 40 paediatric and emergency medicine residents in a real didactic conference. Half were tested on a topic and given a review sheet on the other; the other half got the reverse. Testing and studying happened right after teaching and twice more, about two weeks apart.

Six months later, the topics residents had been tested on scored 39%. The topics they had studied scored 26%. Effect size 0.91, p < 0.001.

Read those absolute numbers honestly. Six months on, everyone had forgotten most of it either way. The finding is the gap, not the score.

The broader review of the field lands in the same place: across ten study techniques, practice testing and spaced practice were the only two rated high utility. Rereading and highlighting were rated low.

The OITE-specific study, and its limits

One group tracked this prospectively in orthopaedics. Twenty-three residents at a single programme, all on the same question bank, tracked from July to the OITE.

The spread in effort was enormous — between 780 and 7,535 practice questions per resident.

The number of questions did not predict the raw OITE score. Total questions came out at p = 0.096, not significant even before controlling for anything. What did predict the score, once everything was in one model, was year in training, USMLE Step 1, and the percentage of practice questions answered correctly.

Now the caveats, because they are real ones. Twenty-three residents, one programme, one year. The authors say plainly that the study is probably underpowered to detect a question-count effect and ask for a multi-year study. The first author also discloses paid consulting for the question bank in question. This is a hint, not a law.

Taken with everything above, the plausible reading is that percent correct is a measure of whether retrieval is actually happening, and question count is a measure of how long the tab was open.

What a working dose looked like

A separate study put 60 residents across three programmes on a fixed protocol: 50 questions a day, five days running, one topic at a time. Scores went from 68.9% to 73.3%.

Note what that is and is not. It is a real, repeatable gain from a modest, structured week. It is also 4.4 points for 250 questions, which is a useful reminder of the size of the effect you are buying. The three platforms tested did not produce the same gain as each other, so the tool is not neutral either.

Four changes worth making

• Answer before you look. Cover the options, say the answer out loud, then read. Getting it wrong and then corrected beats reading the right answer smoothly.
• One topic for a few days, not everything at once. That is the pattern that worked in the trial above.
• Track percent correct, not questions done. It is the number with evidence behind it, and it is the one that hurts to look at.
• Write the miss down the same day, in your own words. Not the correct answer — the reason yours was wrong.

That last one is the one people skip, and it is the one that turns a question you got wrong into a thing you know. OrthoVaultOS is where you keep those, sitting against the topic rather than in a notebook you will not open again, so the next pass starts from what you actually missed.