Differences between Mann-Whitney U and chi-square tests for if two-samples come from same distribution

by ABC Analytics   Last Updated August 01, 2020 09:19 AM - source

In a textbook, it introduces the chi-square goodness of fit test to test independence between two variables (organized in a contingency table), which can also be used to test if m samples are likely to come from the same distribution by using one dimension as samples and the other as values/bins.

Then in another chapter, it introduces the Mann-Whitney test, which seems to do the same thing. Both are nonparametric and chisquare can be applied to continuous distribution by discretizing the bins.

I am confused that they seem to do the same thing, so for the purpose of testing if two populations are the same (distribution and parameters), what're their differences?



Related Questions


Test for Equivalence version of MANOVA

Updated March 09, 2018 14:19 PM



testing of composite null hypothesis

Updated March 01, 2017 21:19 PM