Two-way ANOVA with Repeated Measures Explained: How to Make, Read and Report It
Two-way ANOVA with repeated measures analyses an outcome measured on the same participants under every combination of two within-subject factors, for example time (pre, post) crossed with condition (control, training). Because every participant provides data in every cell, between-person differences are removed from the error terms, which usually increases power.
When to use it
Use it when both factors are within-subject, the outcome is continuous, and each participant has complete data. If one factor varies between participants and the other within them, you need a mixed ANOVA, which is a different design.
Logic and notation
With n participants, factors A (a levels) and B (b levels), total variability is partitioned as:
SStotal = SSS + SSA + SSA×S + SSB + SSB×S + SSA×B + SSA×B×S
Each effect is tested against its own error term: FA = MSA/MSA×S, FB = MSB/MSB×S, FA×B = MSA×B/MSA×B×S. Assumptions: independence across participants, approximately normal residuals, and sphericity for any factor with three or more levels (assessed by Mauchly's test and corrected with Greenhouse-Geisser or Huynh-Feldt epsilon; Greenhouse & Geisser, 1959). Factors with two levels need no sphericity correction.
Original example (illustrative)
Four participants complete a memory test at two times (Pre, Post) under two conditions (Control, Training). Scores:
| Participant | Pre/Control | Pre/Training | Post/Control | Post/Training |
|---|---|---|---|---|
| 1 | 10 | 11 | 12 | 16 |
| 2 | 12 | 12 | 15 | 18 |
| 3 | 8 | 10 | 10 | 14 |
| 4 | 10 | 12 | 11 | 17 |
Cell means: 10.00, 11.25, 12.00, 16.25; grand mean = 198/16 = 12.375. Marginal means: Pre 10.625, Post 14.125; Control 11.00, Training 13.75.
| Source | SS | df | MS | F | p | Partial η² |
|---|---|---|---|---|---|---|
| Time (A) | 49.00 | 1 | 49.00 | 98.00 | .002 | .97 |
| Error (A×S) | 1.50 | 3 | 0.50 | |||
| Condition (B) | 30.25 | 1 | 30.25 | 27.92 | .013 | .90 |
| Error (B×S) | 3.25 | 3 | 1.08 | |||
| Time × Condition | 9.00 | 1 | 9.00 | 54.00 | .005 | .95 |
| Error (A×B×S) | 0.50 | 3 | 0.17 | |||
| Participants (S) | 28.25 | 3 |
Check: 49 + 1.5 + 30.25 + 3.25 + 9 + 0.5 + 28.25 = 121.75 = SStotal (Σx² − 198²/16 = 2572 − 2450.25). The interaction is the key result: the pre-to-post gain is 12.00 − 10.00 = 2.00 under Control but 16.25 − 11.25 = 5.00 under Training. Because the interaction is significant, interpret the simple effects before the main effects. Partial η² = SSeffect/(SSeffect + SSerror); report which η² you used, since values from different designs are not directly comparable (Lakens, 2013).
With n = 4, these results are for teaching only. Real studies need larger samples for stable estimates and adequate power.
How to make it in StatFAS
Open the Two-way ANOVA with Repeated Measures Calculator. Typically data are arranged in wide format, one row per participant and one column per factor combination.
APA 7 reporting
A 2 (Time) × 2 (Condition) repeated-measures ANOVA showed a significant Time × Condition interaction, F(1, 3) = 54.00, p = .005, ηp² = .95. Gains from pre to post were larger under Training (M = 5.00) than under Control (M = 2.00).
Common mistakes and alternatives
Treating repeated measures as independent groups, ignoring sphericity for factors with three or more levels, interpreting main effects despite a strong interaction, dropping participants with missing cells without saying so, and using partial η² without naming it. Alternatives include linear mixed models (better with missing data), the Friedman test for a single non-parametric repeated factor, and the general ANOVA calculator. For the plot, see the interaction plot.
Conclusion
This design gives each effect its own error term and removes person-to-person variability, but it demands complete data and careful treatment of interactions.
1.
statfas-two-way-anova-with-repeated-measures-interface.png, alt: "StatFAS Two-way ANOVA with Repeated Measures Calculator interface"2.
statfas-two-way-anova-with-repeated-measures-example-data.png, alt: "Example data entered in the StatFAS Two-way ANOVA with Repeated Measures Calculator"3.
statfas-two-way-anova-with-repeated-measures-settings.png, alt: "Two-way ANOVA with Repeated Measures Calculator settings in StatFAS"4.
statfas-two-way-anova-with-repeated-measures-result.png, alt: "Two-way ANOVA with Repeated Measures Calculator result generated in StatFAS"References
- Greenhouse, S. W., & Geisser, S. (1959). On methods in the analysis of profile data. Psychometrika, 24(2), 95–112.
- Lakens, D. (2013). Calculating and reporting effect sizes to facilitate cumulative science. Frontiers in Psychology, 4, 863.
- Maxwell, S. E., & Delaney, H. D. Designing experiments and analyzing data. Add edition, year and publisher.