This is a test version of Biostars. For the public version, visit https://www.biostars.org.
How to estimate the power of association test for small sample size

I am planning to use association tests such as burden test, but I only have a few sample size. Are there any papers/references, beside this guide ( http://faculty.washington.edu/tathornt/SISG2015/lectures/Taipei2015/Taipei2015session07.pdf ), that would tell how many samples I need to achieve certain power (or the minimum number of samples). I know there are some papers that discuss about the power calculation using at least 500 samples, but I need something for sample size below 50.

association test
• 4,093 views • link
updated by Alastair Kerr Alastair Kerr • written by MAPK MAPK

2 answers

G*Power is what I recommend to our researchers for power calculations. Note that you still need a decent understanding of the method but a decent intro can be found on the wikipedia page.

• 0 views • link
written by Alastair Kerr Alastair Kerr

The power you need for a certain association is always dependent on the expected effect size or odds ratio. If you expect an odds ratio of 1.1 you need a far bigger cohort than for an odds ratio of 20.

• 0 views • link
written by WouterDeCoster WouterDeCoster

Thanks, but when you say effects size or odds ratio, do you mean the minor allele frequency? How could we correlate MAF with odds ratio?

• 0 views • link
written by MAPK MAPK

Log in to answer this question.

More posts like this