Effect of resampling schemes on significance analysis of clustering and ranking