I've been presented with report that compares two items, each having over 400 data points. Because it's multivariate, the Hotelling T2 test was used and the result was the Hostelling ellipsoid with an explanation that because most of it overlapped, the two materials were not different. Would I be correct in asking for the p-value? I'm not that familiar with the test but it looks a lot like PCA, so does the same sort of matrix reduction occur?
$\begingroup$
$\endgroup$
5
-
$\begingroup$ what is your data ? multivariate - how ? what is your objective ? testing your assertion ? $\endgroup$– user10619Commented Mar 16, 2018 at 10:36
-
$\begingroup$ Yes, multivariate data and establishing (or not) that two materials are not different. $\endgroup$– BeerhunterCommented Mar 17, 2018 at 7:42
-
$\begingroup$ You are not clear about p-value and PCA. These techniques have quite different purposes. Are you interested in the interpretation of the Hotelling T2 test ? Indicate the results you have obtained. $\endgroup$– user10619Commented Mar 18, 2018 at 8:38
-
$\begingroup$ They are not my results. As presented, the report literally says in text only, that the results overlap, with the diagram also presented. My opinion is that the data, visually doesn't overlap enough, but to the reader, I would highlight a p-value to categorically prove the materials are not different. I can't present the actual data, it's not mine and there too much. I believe some numbers should be presented to prove the assertion. $\endgroup$– BeerhunterCommented Mar 18, 2018 at 12:18
-
$\begingroup$ Could you please edit your question to say exactly what you are saying in your comments ? And may be consider my comments? $\endgroup$– user10619Commented Mar 18, 2018 at 17:02
Add a comment
|