CitationSavitz, David A. & Olshan, Andrew F. (1998). Describing Data Requires No Adjustment for Multiple Comparisons: A Reply from Savitz and Olshan. American Journal of Epidemiology, 147(9), 813-4.
AbstractThe context for discussing multiple comparisons is statistical hypothesis testing and p values, which we dealt with cursorily in our earlier commentary (1) and which Dr. Thompson addresses in detail (2). He organizes the sequence of questions as follows: Should hypothesis tests be used? If they are used, should they be based onp values? If p values are used, should they be adjusted for multiple comparisons? We fully agree with his first two answers, namely "infrequently" and "no," but appear to differ on the third question, to which he answers "yes." Even on this point, however, the differences are more in strategy than in the underlying goals.
Reference TypeJournal Article
Journal TitleAmerican Journal of Epidemiology
Author(s)Savitz, David A.
Olshan, Andrew F.