Wednesday, September 21, 2011

The problem with low powered studies.

This figure shows that amount of overlap between the null hypothesis (Ho) and the alternative hypothesis (Ha) when the power of a study is only 15%. When the p-value is < 0.05, it is reasonable to reject Ho since you are 3 times more likely to be right than wrong. But when the p-value is > 0.05, it is not possible to reject or accept Ho or Ha since you are just about as likely to be wrong as to be right, i.e. you would be almost as accurate by just flipping a coin.