Crossover Design Sample Size

#1
Hi all

I'm having trouble figuring out the sample size needed for a 2x2 crossover study. The population hasn't been studied yet so I don't have exact exploratory data to call on regarding anticipated effect sizes but the best fit would suggest a medium-large effect size.

I'm not particularly strong on stats though and have been informed not to use effect sizes to figure out the sample size. So I guess with no exploratory studies to replicate and if I'm not to use effect sizes, how do I figure out the sample size?

I don't think I can simply work it out based on the analysis I want to run because there may be ethical implications with using too many of my population at this stage of research.

Any ideas? Or signposts to papers?

Thank you :)
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
Nope you need effect sizes to guide the planning. You can also base sample size on precision estimates, but those still need to be anchor on an effect!