lisijie515

Register on Conventional: 2007/04/02 Posts: 234
5th Floor
lisijie515
Actually rank sum test is a permutation test, and permutation foursquare test for null basically rarely very far away from equality foursquare of distributions. Therefore, the rank sum test is actually testing whether two distributions in the same. If we assume the same shape as the two distributions, it is then that the two positions inspection distribution (median); If the distribution is normal, which is the inspection means; if the shape is different, it is in the inspection stochastic ordering. Rough but inaccurate to say that is to look at a random sample from each depicting a distribution inside foursquare the first than the second largest probability is not 0.5. So, even though the median and IQR same, it does not represent the same two distributions, rank sum test entirely possible significant.
COS Forum | Statistics Capital is proudly powered by bbPress.