If the groups are independent, I'd use the Games-Howell procedure, which is the multiple comparison analog to the Welch t-test for unequal n's.
At 12:51 PM 03/30/2003 -0800, Excel wrote:I am comparing 4 class means. Each having different sizes. I have ran a F-test and there is a significant difference between the means. I want to run a Post Hoc test to find out which ones. Do I use Tukey? Do you need equal class sizes? Scheffe?
now, even in a fully randomized experiment ... group ns may not be identical but ... when you talk about "classes" ... and different ns ... it suggests to me that you are using INTACT classes ... is that the case?
if that IS the case ... you have a much larger problem than trying to figure out the correct post hoc test
. . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
~~~ Lise DeShea, Ph.D. Assistant Professor Educational and Counseling Psychology Department University of Kentucky 237 Dickey Hall Lexington KY 40506 Email: [EMAIL PROTECTED] Phone: (859) 257-9884
"... for only by varied iteration can alien conceptions be forced on reluctant minds."
-- Herbert Spencer, in the preface to The Data of Ethics, 1881.
_________________________________________________________
dennis roberts, educational psychology, penn state university
208 cedar, AC 8148632401, mailto:[EMAIL PROTECTED]
http://roberts.ed.psu.edu/users/droberts/drober~1.htm
. . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
