Why you can't just compare each p to .05
Three pairwise Mann-Whitney comparisons from the pencil-length data. Step through how Holm turns three raw p-values into three honest ones.