Suggested answer to exercise: Estimation of sample size, 4

Question 4: What is meant by ‘power of 80%’?

Suggested answer

Some possible samples would produce differences larger than that in the whole population of patients, some would produce differences which were smaller. The larger the observed difference is, the smaller will be the P value for the significance test comparing the proportions with veno-occlusive disease between the treatment groups. The proportion of studies that will produce a difference big enough to be significant, if in the population the difference is that postulated, is called the power. This is also the probability that a study will produce a significance difference. Here it has been set to 80%, so that 80% of possible studies would produce a significant difference if defibrotide reduces the risk of hepatic veno-occlusive disease from 30% to 15%. This means, of course, that 20% of possible studies would not be significant.


Back to Exercise: estimation of sample size.

This page maintained by Martin Bland.
Last updated: 3 February, 2020.

Back to top.