s-news
[Top] [All Lists]

Multiple comparisons - general stat question

To: <s-news@lists.biostat.wustl.edu>
Subject: Multiple comparisons - general stat question
From: "John Sorkin" <jsorkin@grecc.umaryland.edu>
Date: Mon, 28 Apr 2003 17:03:06 -0400
I am currently trying to respond to a reviewer. Any help would be
appreciated. 

The reviewer insists upon a significant omnibus F test prior to
performing an adjustment for multiple comparisons using the method of
Dunnett-Hsu. I have three groups, on control and two treatments. I am
comparing each treatment to the control group. The comparisons were
planned ad hoc. I am not comparing one treatment group to the other
treatment group. I have told the reviewer that Dunnett's test can be
performed without a prior significant omnibus F test. The reviewer does
not believe me. Can anyone suggest a winning argument, or better a good
reference?

Thanks
John 

John Sorkin MD, PhD
Chief, Biostatistics and Informatics
Baltimore VA Medical Center GRECC and
University of Maryland School of Medicine Claude Pepper OAIC
410-605-7119 
john@grecc.umaryland.edu

<Prev in Thread] Current Thread [Next in Thread>