NotesNudge the notes you never go back to

The Testing Effect

The testing effect is the finding that attempting to recall material produces better long-term retention than restudying it for the same amount of time.

It is the single most useful result in this field for anyone with a pile of notes, and it comes with a caveat that almost never travels alongside it.

A related workplace concept is mouse jiggler detection software, which offers a useful contrast with the personal knowledge-work problem discussed here.

Reviewed August 9, 2026.

The finding

Roediger and Karpicke's 2006 experiments are the canonical reference: participants who were tested on material retained substantially more at a delay than participants who restudied it, despite restudy feeling more effective at the time.

The effect has since been replicated widely, in laboratories and classrooms, for facts, vocabulary, concepts and procedures.

For an independent external reference related to this topic, see SAGE Journals.

Two features make it unusually practical.

Unsuccessful attempts still help. You do not need to remember for the attempt to be worth making. Trying and failing, then seeing the answer, beats simply seeing the answer.

And there is a forward effect. Reviews of research on this describe testing of earlier material improving learning of subsequently presented new material — a well-replicated laboratory phenomenon. Retrieval does not only consolidate what you tested; it appears to leave you better placed for what comes next.

The caveat that rarely travels with it

A meta-analysis of spacing and retrieval practice in mathematics found something different. Across seven studies and 32 effect sizes manipulating testing versus restudy, the weighted mean effect was g = 0.18, with a 95% confidence interval crossing zero — which the authors describe plainly as the testing effect not being robust in that domain.

Spacing held up in the same analysis, at around g = 0.43 for isolated material and g = 0.24 within a course. Testing did not.

That does not overturn the general finding. It means the effect depends on material more than the popular summaries admit, and that a result established largely on facts and vocabulary should not be assumed to transfer to conceptual or procedural domains at the same size.

Anyone quoting the testing effect as a universal law is quoting more confidently than the literature does.

Why it feels wrong to do

The most practically important thing about this result.

Restudy feels more effective while you are doing it. Rereading is fluent; the material is right there and comprehensible, and that fluency reads as mastery. Attempted recall is halting, uncomfortable, and produces visible gaps.

So the technique that works feels worse than the technique that does not, at the moment of choosing. This is why people who know about the testing effect still reread — knowing about a fluency illusion does not dispel it, any more than knowing about an optical illusion straightens the lines.

The countermeasure is structural: build the attempt into the routine so that it does not require choosing. Fifteen seconds of attempted recall before opening a note, every time, as a rule rather than a decision.

What it means for notes

Do not reread the note. Try to reconstruct it first.

Given a title, spend fifteen seconds attempting to say what it contained and why you kept it. Then open it and see the gap. That is the whole application, and it is the difference between resurfacing and retrieval.

Write titles that make attempts possible. A title stating a claim gives you something to reconstruct; a topic label gives you nothing to try. This is a capture-time decision that determines whether review can work at all.

And accept the failures. A blank is not a wasted attempt. It is the condition under which seeing the answer does the most.

What it does not license

It does not mean testing everything. Benefits asymptote and attention is finite.

It does not extend to insight or creativity. Retention is what was measured.

And it does not justify a product that shows you notes. Showing is restudy. The effect is about the attempt, and a tool that does not ask you to try has implemented the comparison condition.

The short version