Skip to main content
v2026.11,986 entries · CC-BY 4.0

Direct comparison

Within vs Between Subjects: Differences & Examples

Within-subjects designs test every participant in every condition; between-subjects designs give each person one. Compare power, sample size and carryover.

Written and maintained by CASRAI Editorial Board

Last updated

How do Within-Subjects and Between-Subjects compare side by side?

The table below compares Within-Subjects and Between-Subjects across 14 dimensions, from also called to checking counterbalancing.

Side-by-side comparison

DimensionWithin-SubjectsBetween-Subjects
Also calledRepeated-measures designIndependent-groups design
Participants per conditionSame participants in every conditionDifferent participants in each condition
Individual-differences varianceRemoved from the error term (each subject is own control)Remains in the error term
Sample size for equal powerSmaller, for the same effect sizeLarger, for the same effect size
Main confound riskCarryover, practice, and fatigue/order effectsBetween-group differences unrelated to the manipulation
How the confound is controlledCounterbalancing (Latin square, ABBA) or a washout periodRandom assignment to condition
Typical test (2 conditions)Paired-samples t-testIndependent-samples t-test
Typical test (3+ conditions)One-way repeated-measures ANOVAOne-way between-subjects ANOVA
Key analysis assumptionSphericity (equal variance of pairwise differences)Homogeneity of variance across groups
Participant burdenHigher — multiple sessions/conditions per personLower — one condition per person
Cannot be used whenThe manipulation only works once, or its effect cannot fully wash outRecruiting enough participants for every group is not feasible
Fixed participant traits (e.g. hand preference, clinical vs non-clinical group)Not possible - a person cannot hold both levelsRequired - each person belongs to one level
Typical UX or A/B useSmall-sample usability tests of simple variations such as button placementLarge-scale A/B tests and comparisons of whole flows, with random assignment
Checking counterbalancingAdd order group as a between-subjects factor; a non-significant effect suggests it workedNot needed - there is no order to balance

Ask CASRAI · free to try

Ask about Within vs Between Subjects: Differences & Examples

2 free questions a day, no account.

An AI assistant for research: funder rules, compliance, methods and publishing. It links the sources behind its answers and labels answers that come from a web search.

Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.

Works on this site and inside Claude, Cursor and the AI tools you already use.

Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays open to read, with no account and no card.

Common questions

Common questions about Within-Subjects vs Between-Subjects

Which design needs fewer participants?

+

Within-subjects, for a given effect size and power target, because removing between-person variance from the error term increases statistical power per participant. A between-subjects design needs a larger total sample to reach the same power.

Can a study combine both?

+

Yes — a mixed (split-plot) design manipulates one variable between subjects (e.g., treatment vs control group) and another within subjects (e.g., repeated time points), which is the standard structure for a pre-post study with a control group.

Is a crossover trial a within-subjects design?

+

Yes. A crossover trial is the clinical-research application of a within-subjects design, where each participant receives multiple treatments in sequence, with a washout period between them specifically to manage carryover.

What replaces repeated-measures ANOVA when sphericity is violated?

+

A Greenhouse-Geisser or Huynh-Feldt correction to the degrees of freedom, or a linear mixed-effects model, which does not assume sphericity at all.

Within-subjects vs between-subjects: which is better?

+

Neither is better for every study. When there is a choice, within-subjects is usually preferred because it needs fewer participants and removes individual differences from the error term, but between-subjects is better when conditions interfere with each other, when a manipulation works only once, or when the factor is a fixed trait.

Is within-participants the same as within-subjects?

+

Yes. Within-participants, within-subjects, repeated-measures and related-samples all describe a design where each participant takes every condition; between-participants, between-subjects and independent-groups describe the design where each participant takes one.

How do you counterbalance a within-subjects design with three or more conditions?

+

Use a Latin square so each condition appears once in each position across order groups, with the number of participants a multiple of the number of conditions. A balanced Latin square also equalises which condition follows which, but only exists for an even number of conditions.

Which statistical test goes with each design?

+

For two conditions, a paired-samples t-test for within-subjects data and an independent-samples t-test for between-subjects data; for three or more, a repeated-measures ANOVA and a one-way between-subjects ANOVA respectively.

Referenced across the research world

University of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logoUniversity of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logo
  • University of Cambridge logo
  • Columbia University logo
  • Crossref logo
  • University of Edinburgh logo
  • Harvard University logo
  • University of Oxford logo
  • Princeton University logo
  • Stanford School of Medicine logo
  • University College London logo
  • ORCID logo

View CASRAI adoption →