Confusing “Significant” and “Important”

Null-hypothesis significance testing (NHST) is a well-established and widespread statistical technique. It allows a researcher to calculate the probability that variables which appear to be correlated actually are correlated.

However, NHST is often mistaken for a judgement on the importance of a empirical results, as opposed to being only a statement about the chances of it being in error.

We hope to explain the problems with NHST in a way that encourages responsible use of NHST when it is an appropriate tool to use, and the use of other techniques when NHST is not appropriate.