I don't think that is correct. For starters, you would have to adjust the sig levels for the use of multiple tests (the STATS PADJUST extension command can do this for a variety of corrections). But, also, if the error terms are correlated, that would affect the tests.
Can you post the link to the YouTube link and to the papers you read?
------------------------------
Jon Peck
------------------------------
Original Message:
Sent: Thu May 28, 2020 08:26 AM
From: Cagdas Erbas
Subject: Multivariate Kruskal-Wallis Test
Hi,
Are there anyway to perform Multivariate Kruskal-Wallis Test on SPSS?
In my data, I have 6 dependent and 3 independent variables. Unfortunately one of my groups has less than 20 participants which are the first violation of MANOVA assumptions. Papers shows that I can perform the Multivariate Kruskal-Wallis Test for this analysis this particular data. I found a video on YouTuve which says if I perform all pairwise Kruskal-Wallis Test, it will count as Multivariate Kruskal-Wallis Test. Is this info true or is there any way to perform this analysis?
------------------------------
Cagdas Erbas
------------------------------
#SPSSStatistics