The abuse of Power, P-values and the Misnomer of Statistical “Testing”

In estimating the sample size in the design of a randomised controlled trial (RCT) certain assumptions are necessary. Critical amongst these is an educated guess as to how much difference the intervention might be expected to make and whether this will be of clinical importance. Typically, investigators will estimate a likely value for the defined primary outcome in the control group, and then decide what a clinically meaningful difference would be in the intervention group as the “effect size”. Roughly speaking, the larger the difference or effect size that the intervention is anticipated to make, the smaller the sample required for the study.