Stack Exchange network consists of 183 Q&A communities including Stack Overflow, the largest, most trusted online community for developers to learn, share their knowledge, and build their careers.
Thank you for your input. I think z-tests can be used for testing proportions of categories. But the prop.test() function seems to implement a non-parametric test, which I prefer. r-bloggers.com/…
Thank you for the comments. Do you maybe have some links to the literature you are mention or more information about the distributions used? The pareto distribution seems to be often used to model the upper tail. But I might in future also be interested in the lower deciles. Do you happen to know any distribution that fits well here?
I found another helpful explanation of residual-based bootstrap as being applicable when we can assume our combinations of independent variables to be "fixed" for multiple re-runs, but where the outcome might vary (example: experiments with well defined conditions). This is the case for my data, so I will probably go with res.-based bootstrap (unless I get more answers ;) ). [page 13 of: socialsciences.mcmaster.ca/jfox/Books/Companion/appendix/… ]
