Mann-Whitney U Test

Dash

New Member
#1
Hello everyone,

Thank you all for providing and maintaining this forum.

I'm currently working with two data sets collected ten months apart, of about 20,000 observations of the time it takes individuals to complete a specific journey.

A normal probability plot shows that both data sets are heavily skewed to the left, which has led me to consider a Mann-Whitney U Test. I'm interested in knowing whether there is a statistically significant difference between data sets (alpha set at 0.05) based on a treatment on the latter.

I have not been able to find whether there is a limit to how large a sample can be when applying Mann-Whitney (Kruskal-Wallis seems to generally cap at 30 unless one is using special software?).

Any feedback on whether this test is fit for this study, as well as any other suggestions on what to do with such large and skewed data sets would be great appreciated.

Thank you in advance for your help.
 

Karabiner

TS Contributor
#2
With such a large sample size, you do not need to use U-test instead of t-test.
U-Test could be interesting, though, since means might not be representative in presence of huge skewness.

With Kind regards

K.
 

rogojel

TS Contributor
#3
And BTW even the slightest difference will be statistically significant - you will need to watch out for practical significance.

Regards