Three classic designs for comparing conditions, each with characteristic strengths and weaknesses.
Independent groups (between-subjects)
Different participants in each condition. Person A is in the music condition; Person B is in the silence condition.
Strength: no order effects (each person does the task only once). Weakness: large individual differences between groups can mask the IV effect, unless N is large or random allocation is used carefully.
Repeated measures (within-subjects)
The same participants in every condition. Person A does the memory task with music and with silence.
Strength: each participant is their own control — eliminates individual differences. Weakness: order effects — practice (gets better) or fatigue/boredom (gets worse) by the second go. Counter with counterbalancing (half do music first, half silence first) so order effects cancel out.
Matched pairs
Two separate groups, but participants are matched on a relevant variable (IQ, age, gender, prior knowledge). The closest pair is split, one to each condition. Combines some advantages of both other designs.
Strength: reduces individual differences without order effects. Weakness: hard work — need a large pool to find good matches; never a perfect match; participants drop out can leave singletons.
How to choose
- Tasks that can't be repeated (e.g. learning a new skill once) → independent groups or matched pairs.
- Tasks where participant variables matter a lot but order effects can be controlled → repeated measures.
- Limited participant pool → repeated measures.
- Strict scientific rigour with large N → independent groups with random allocation.
Identifying the design in a question
Key clue: count the participants per condition.
- "30 participants total, 15 per condition" → independent groups.
- "30 participants, each doing both conditions" → repeated measures.
- "30 pairs, each pair split between conditions" → matched pairs.
⚠Common mistakes— Common errors
- Confusing matched pairs with repeated measures.
- Forgetting to mention counterbalancing as the fix for order effects.
- Listing strengths/weaknesses without linking them to why (e.g. "no order effects" without explaining what an order effect is).
AI-generated · claude-opus-4-7 · v3-deep-psychology