2. Appraising the relevance of results

2.3 The eight steps

7. What are the results ?

Statistical power

A statistically significant difference is of interest only if it is clinically pertinent.When the difference is not significant, the reviewer will say that the results are negative. But is it a "true negative"? It could possible be, in fact, a "false negative".

A difference or a desired effect may actually be present, but the sample size may have been too small to demonstrate it. Or else the difference or outcome may be less significant than expected hypothetically, in which case the sample size was still too small to demonstrate it (we say that the study has low statistical power). A type II error is one that affirms that there is no difference or outcome, when in actual fact, there is one. This assumption is comparable with false negative diagnostic tests.