I'm used to working with very large data sets in my regression analyses, but
I've encountered a situation where my N is 24, and my k (number of
independent vars) is 9. I may be able to expand the sample to N=36, but in
any case, how do you decide when a sample is "too" small to be useful? Does
it just depend on your results (e.g., standard errors are really large)?
