Hello,
I have several ordered groups (20), each with several observations. Each
group has fewer observations roughly linearly from 30 to 0. Each
observation is a proportion. As I know the max and min values for a
proporion are 1 and 0, I am adding these values to each group to allow
bootstrap for groups with <= 1 observations. Sometimes the bootstrap
confidence interval is > 1, which makes me think that I am doing something
deeply wrong.
How can I do bootstrap properly with observations which are proportions?
What special precautions should I be taking for very low sample sizes with
bootstrap?
Is this a bad question for R-help?
Thanks,
Dan.
