I am using a mixed model in lmer to analyse some phenotype data across 3 levels of a treatment I am interested in, blocked by a random factor I am not interested in (just want to control for). So my model is basically:

model <- lmer( thing ~ treatment + (1|block) )

I have pilot data that gives me a measure of the variance of the phenotype 'thing' in this population. So I was hoping to use a power analysis to ask the question -

'Given the variance X in thing, how many samples would be needed to find significant differences across treatments?'

I was planning on running this power analysis for different hypothetical effect sizes. The thing is, I haven't even thought about power analysis since a distant stats lesson some time in my undergrad bio degree, and at that time I certainly wasn't told anything about dealing with them in mixed models.

I had a look at potential R functions and could only come up with the pwr package, which doesn't seem to have functionality for mixed models. I also came across some tutorials which started with a simulated dataset, but it didn't seem to address what I thought would be a simple problem (either that or I didn't understand).

I simply want to give the variance and [hypothetical] effect size, and get the sample size needed to find a significant difference - within the mixed model. Is this more complicated than I thought? Has anyone got any pointers of where I could look for some clear instruction, or functions I've missed that might be helpful?

Thanks in advance for any help,



migrated from stackoverflow.com Jun 15 '16 at 12:49

This question came from our site for professional and enthusiast programmers.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.