Sample size t test - median and SD
Description of test and variables
This sample size calculation using the median and standard deviation (SD) (or variance) instead of the mean for positively skewed data was developed by Aidan G. O’Keeffe, Gareth Ambler and Julie A. Barber at the University College London, UK and published at BMC Medical Research Methodology in 2017.
The authors argue that outcomes with skewed distributions are very common in healthcare research and sample size calculations are based on estimates on a transformed scale, which are usually difficult to obtain. In contrast, estimates of median and variance on the untransformed scale are generally easier to specify.
This sample size calculation compares log-transformed data using a log-normal distribution for a two-sample t-test .
Worked Example
Video recorded by Kevin Mitchell