Counterbalancing
Distribute order effects across experimental conditions
Counterbalancing varies condition or stimulus order across participants to separate treatment effects from order effects. Complete counterbalancing uses all orders; Latin-square and Williams-type designs balance selected position or carryover features with fewer sequences. Randomising order can complement balancing, but balance against one type of order effect does not remove every possible carryover.
Use when participants complete several tasks or conditions and practice, fatigue or preceding stimuli could alter responses. Choose which order effects matter and whether treatment effects are sufficiently reversible.
Strengths
- Reduces systematic confounding of condition with order
- Efficient partial designs can limit the number of sequences
Limitations
- Complex carryover may survive the chosen balance
- Sequence imbalance returns through attrition or unequal recruitment
Know the boundary
Counterbalancing cannot erase irreversible learning or a lasting treatment effect.