In one sentence
In six lab experiments, psychology undergraduates who guessed whether a scenario was a true experiment learned no more — and transferred no better — than peers who saw the label with the example; what did help was feedback that explained why.
What the researchers did
Corral and Carpenter taught introductory psychology students (Iowa State and Colorado Boulder) a compact research-methods idea: which hypothetical studies count as true experiments. After a short slide set, everyone saw 12 scenarios (half true, half not).
Classification training: judge true vs not, then see feedback. Observation training: see the scenario with the category (Experiment 2 onward, at the same time, so people were less likely to self-test first). Feedback was either the label only or the label plus an explanation.
Post-tests used new scenarios: classify them, and application items (what is wrong / how to fix it). Timing varied: immediate (Experiments 1–2; n = 269 and 300), 10 minutes (3–4; n = 310 and 246), two days (5; 174 of 196 returned), one week (6; 180 of 194 returned).
What they found
- Classification vs observation: no reliable advantage for guessing first on classification or application items, at any delay they tested.
- Explanation vs label-only feedback: explanation generally produced better learning and transfer (clearest on classification questions; Experiments 1–5 and a combined analysis).
- That explanation edge was strongest immediately and weaker after a week.
- Observation learners were not simply failing to learn: when explanations were present, both training types improved.
- Experiment 2 presented labels simultaneously with examples so observation would not secretly become covert testing; the null training difference still held.
What this means for learners and educators
- “Make them generate the category” is not automatic transfer magic for this kind of complex, example-based concept. Worked examples with why can match self-classification.
- Spend the scarce minute on a reason, not only a tick. “Not a true experiment — no random assignment” beats “incorrect.”
- Retrieval practice still matters for facts you must produce later. This paper is a caution about over-generalising retrieval and generation to every transfer task.
Limitations and what we don't know yet
Materials were one domain (true experiments) and lab undergraduates. Application-question benefits of explanations were less consistent than classification-question benefits. Delays stopped at one week. The authors note that many popular theories predicted a classification win; this set of results does not license dropping retrieval elsewhere — it narrows where the advantage should be expected.
FAQ
Should I stop practising with closed-book questions?
No. This study compared two ways to learn research-design categories from scenarios. It is not a verdict on flashcards or exams.
Is reading explanations enough?
Explanations beat a bare label here. They still need later checks — the explanation benefit shrank by one week.
Why didn’t guessing help?
For these items, seeing a labeled example plus a reason may already supply the structure that a guess was supposed to build.