Reviewing a Missed Question the Long Way

Why the fast review does nothing

You miss an item. You open the rationale. You read that option C was correct because of a thing you already knew, and you think, right, of course, I should have seen that.

Then you move on. Nothing changed.

That feeling has a name worth borrowing from cognitive psychology: recognition. Recognising an explanation is much easier than producing one, and it feels almost identical from the inside. It is the reason a review session can feel productive and leave no trace.

The slow review is built to block recognition. It makes you produce your reasoning before you are allowed to see the answer's.

The four questions, in order, before the rationale

Cover the rationale. Do not peek. Then work through these on paper or out loud.

• What was the stem actually asking me to do? Say the task in your own words, without looking at the options.
• Why did I pick what I picked? Not what the answer is. What my reason was at the time.
• Why is the correct option correct? Build the case yourself, from the stem, as if you were explaining it to someone.
• What would have to change in the stem for my answer to become correct? This one is the most useful and the most skipped.

That last question is doing something specific. It forces you to name the exact feature of the scenario that discriminates between your pick and the key. Until you can name it, you have not learned the item, you have only learned its answer.

Only now open the rationale. Compare their reasoning to yours, not their letter to yours.

Classify the reason before you close the item

You missed it for a reason, and the reason belongs in one of a small number of buckets. Content gap. Misread the task. Eliminated on instinct. Right answer, wrong sequence. Scope or role.

Assign one. Write the letter or word next to the item number and move on.

The categories themselves, how many to use, and what each one tells you to do next are worked out in tracking what you miss by reason not by topic. Use whatever set you settle on consistently, because the value is in the tally, not in any one label.

Here is why this beats topic tracking. A tally that says you are weak in endocrine tells you to review endocrine, which is a lot of hours. A tally that says two thirds of your misses were misread tasks tells you to change how you read stems, which is an afternoon.

If your eliminations are not currently recorded with reasons, that is the upstream fix. Eliminating distractors on rationale covers the two-second notation that makes this classification possible at all.

Logging it in one line

The log is not a notebook of rewritten rationales. Nobody rereads those.

One line per missed item. Item source and number, the task the stem set, your reason, the correct reason, and the category. That fits on a single line in a spreadsheet or a page in a notebook.

Something like: qbank item 412, task was next action, I chose the first action, key was the step after the completed one, misread task.

Written that way, a month of misses is scannable in five minutes and the pattern is visible without any analysis. That scannability is the entire design goal.

Add one more field if you can stand it. A yes or no for whether you were confident when you answered. Confident misses and uncertain misses need different responses, and you cannot reconstruct the difference later.

Confident misses are the ones to look at first. A confident miss means something in your model is wrong rather than missing, and wrong beliefs do not get corrected by exposure to more questions. They get corrected by being contradicted directly, which is what the fourth question is for.

Uncertain misses are usually gaps, and gaps close with ordinary study.

Two different piles, two different responses. Sorting them takes one extra column.

How much review a session deserves

More than feels natural, and the ratio matters more than the total.

The slow review takes roughly two to three minutes per missed item once the habit is built. Do the arithmetic on your own session before you plan it, because a set with fifteen misses carries a review tail you have to actually make room for.

If the tail does not fit, do fewer questions. A shorter set fully reviewed teaches more than a longer set skimmed, and the second option mostly produces the feeling of having worked.

The temptation runs the other way, because question counts are visible and review is not. A day with 100 items and no review looks better in a log than a day with 30 items and a full tail. Only one of them changes how you read the next stem.

Count reviewed items, not answered items. That single change to what you record changes what you do.

How to structure the session around that tail, including whether to review as you go or at the end, belongs to what a practice question session should look like.

Two variations worth knowing

Review items you got right but were unsure about. Same four questions. A lucky guess and a reasoned answer look identical in a score report and are completely different facts about your readiness.

And if you are on the PN track, add a fifth question to role-based items: was the option eliminable on scope before any clinical reasoning happened? Who does this questions and how to read them covers that shape, and question analysis with a PN scope in mind covers the broader habit.

The prerequisite under all of it stays the same. If you cannot name the task a stem set, none of the four questions have anything to bite on, and what a question stem is actually asking you to do is where that starts.