Re: [R] CI for the median difference

2012-02-18 Thread Patrick Breheny
It seems to me that your two approaches are calculating CIs for different quantities. The bootstrap methods are calculating a CI for the difference in medians, while the Wilcoxon approach is calculating a CI for the median of the differences. If this were the mean, those would be the same, bu

[R] CI for the median difference

2012-02-18 Thread Vittorio Colagrande
Dear R-group, I have run into a problem in estimating confidence intervals for the median difference. I want to establish a confidence interval at (1- alpha) level for the difference between the medians of two indipendent samples (size n and m), by using the Wilcoxon distribution or with