Introduction
In biomedical and clinical research, repeated measurements on the same subjects over time are common. For example, researchers may measure blood pressure at baseline, after 1 month of treatment, and again after 3 months. Because these measurements are taken from the same individuals, they are statistically dependent. Standard ANOVA cannot be used in such cases because it assumes independence of observations.
To analyze longitudinal data properly, we use Repeated Measures ANOVA.
Repeated Measures ANOVA evaluates whether there are statistically significant differences in the mean of a continuous variable measured at multiple time points within the same subjects.
In this article, we demonstrate how to perform and interpret Repeated Measures ANOVA in MedCalc, using systolic blood pressure data measured at:
- BP_Baseline
- BP_1Month
- BP_3Month
The analysis includes:
- Sphericity testing
- Greenhouse-Geisser correction
- Within-subject effects
- Trend analysis
- Bonferroni-adjusted pairwise comparisons
- Clinical interpretation
This tutorial is ideal for medical researchers, PhD scholars, and biostatistics students.
Study Design
Number of Subjects
n = 10 subjects result
Each subject has three systolic blood pressure measurements:
| Time Point | Mean (mmHg) | Std. Error | 95% CI |
|---|---|---|---|
| BP_Baseline | 149.9 | 0.6046 | 148.53 – 151.27 |
| BP_1Month | 142.2 | 0.7272 | 140.55 – 143.85 |
| BP_3Month | 135.0 | 0.8165 | 133.15 – 136.85 |
There is a clear downward trend in blood pressure over time.
📥 Download the Repeated Measures ANOVA Dataset (Excel)
Sphericity Assumption
Repeated Measures ANOVA assumes sphericity, meaning that the variances of differences between time points are equal.
From your MedCalc output:
| Method | Epsilon |
|---|---|
| Greenhouse-Geisser | 0.661 |
| Huynh-Feldt | 0.730 |
Because epsilon is less than 1, sphericity may be violated. MedCalc automatically applies corrections.
Test of Within-Subjects Effects
Sphericity Assumed
| Source | SS | DF | F | P |
|---|---|---|---|---|
| Factor | 1110.467 | 2 | 3486.35 | < 0.001 |
Greenhouse-Geisser Corrected
| DF | F | P |
|---|---|---|
| 1.322 | 3486.35 | < 0.001 |
Interpretation
Even after correction, P < 0.001, indicating a highly significant difference in systolic blood pressure across time points.
Trend Analysis
Trend analysis helps determine the pattern of change.
| Trend | t | DF | P |
|---|---|---|---|
| Linear | -63.857 | 9 | < 0.0001 |
| Quadratic | 3.000 | 9 | 0.0150 |
Interpretation
- Significant linear trend indicates a consistent decrease over time.
- Significant quadratic trend suggests slight curvature in decline.
Clinically, this implies steady blood pressure reduction with treatment.
Pairwise Comparisons (Bonferroni Corrected)
| Comparison | Mean Difference | P-value |
|---|---|---|
| Baseline vs 1 Month | 7.7 | < 0.0001 |
| Baseline vs 3 Month | 14.9 | < 0.0001 |
| 1 Month vs 3 Month | 7.2 | < 0.0001 |
All comparisons are statistically significant.
This confirms:
- Significant reduction from Baseline → 1 Month
- Further reduction from 1 Month → 3 Month
- Largest reduction from Baseline → 3 Month
Clinical Interpretation
The treatment appears highly effective:
- Average reduction after 1 month ≈ 7.7 mmHg
- Average reduction after 3 months ≈ 14.9 mmHg
A 15 mmHg reduction in systolic blood pressure is clinically meaningful and may significantly reduce cardiovascular risk.
Why Repeated Measures ANOVA Is Important
Repeated measures designs:
- Increase statistical power
- Reduce variability
- Require fewer subjects
- Control for individual differences
MedCalc simplifies:
- Sphericity correction
- Trend testing
- Post hoc comparisons
- Confidence interval reporting
Conclusion
This Repeated Measures ANOVA analysis in MedCalc demonstrates a highly significant reduction in systolic blood pressure over time.
Key findings:
- Strong overall time effect (P < 0.001)
- Significant linear decreasing trend
- All pairwise comparisons significant
- Clinically meaningful reductions
Repeated Measures ANOVA is an essential statistical method in longitudinal biomedical research, and MedCalc provides a robust, user-friendly environment for its implementation.
