You've measured a variable in three or more groups, and the means (and medians) are distinct. Is that due to chance? Or does it tell you the groups are really different? Which groups are different from which other groups?