the t-test conteols for type I error for one test only. Dunnett's test controls for the trype I error to comapre each group to a control. Thus, if there are 4 groups, Dunnetts' test will control type I error at .05 to compare each of three to a fourth. Unless you manually control it, doing three individual t-tests will have an increased type 1 error rate between .05 and (1 - .95**3), assuming an alpha of .05.
Paul R. Swank, Ph.D. Professor, Developmental Pediatrics Medical School UT Health Science Center at Houston -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of Troopers Sent: Wednesday, June 18, 2003 10:49 AM To: [EMAIL PROTECTED] Subject: t-Test vs Dunnett's Test Could you explain me what is the difference between a t-Test and a Dunnett's Test to compare several groups? . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . ================================================================= . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
