Rapid Responses to:

LETTERS: Karl-Heinz Jöckel and Andreas Stang Issues in reporting epidemiological studies: Study raises several concerns BMJ 2005; 330: 146-c [Full text]

Rapid Responses: Submit a response to this article

Rapid Responses published:

Type I errors

J Martin Bland   (19 January 2005)

Type I errors 19 January 2005
J Martin Bland, Prof. of Health Statistics University of York, YO10 5DD Send response to journal: Re: Type I errors	It is not true that 'Studies that are statistically underpowered . . . more often suffer from type I and type II errors compared with larger studies.' Type I errors are those when we conclude that there is evidence for an effect when in the larger population there is no such effect, e.g. we say the difference in the sample is significant when there is no difference in the larger population. Type I errors can only occur when the null hypothesis is true, i.e. when there is nothing to find. In this case, the probability of a type I error is the significance level which we decide in advance, usually 0.05. It has nothing to do with the sample size. Competing interests: None declared