It is recommended to correct the sample size when leading a community randomized controlled trial that considers the community as a research unit and not individuals or households. I would like to know if this recommendation is to be applied.
PS. Check out Pepperdine, too.
http://cadres.pepperdine.edu/ccar/define.html
Yes - you must allow for clustering by community in both the power calculation and the statistical analysis. There are plenty of papers available showing how this is done. try this web page:
http://www.healthknowledge.org.uk/public-health-textbook/research-methods/1a-epidemiology/clustered-data
Thanks Adrian for the contribution. It still remains an issue. I would like to know how to correct sample size when the effective sample size is very low. for example, I planned n=160 in each cluster and the effective sample size after taking into account ICC and Design Effect is 98 ?
- Badges
- Science topic
- Similar topics
- Medicine
- Disease
- Infectious Diseases