Randomized Paired Comparison
A randomized paired comparison design forms pairs of similar units and randomizes treatment within each pair. This removes much of the between-unit variability and focuses inference on within-pair differences.
The main analysis target becomes the distribution of paired differences rather than raw outcomes.