 |
 |
Dear Martin, Thank you for your response. Unfortunately when I run the batch entering [1 1 -1 -1] as a contrast vector for an F test, I get an "invalid contrast" error. Following McLaren's recommendations, I am running a flexible factorial with the following specifications: Factor 1: subjects (indep, equal var) Factor 2: A (dep, equal var) Factor 3: B (dep, equal var) Then for Main Effects and Interactions I select ME for the 3 factors and the A-B interaction, which yields a design matrix of 8 conditions: A1 A2 B1 B2 A1B1 A1B2 A2B1 A2B2. For the t-contrasts I use: 1 -1 0 0 .5 .5 -.5 -.5 (tests A1>A2) -1 1 0 0 -.5 -.5 .5 .5 (tests A2>A1) 0 0 1 -1 .5 -.5 .5 -.5 (tests B1>B2) 0 0 -1 1 -.5 .5 -.5 .5 (tests B2>B1). I'm not sure why then when I enter 1 1 -1 -1 as an F contrast, I get an error, whereas something like 1 -1 0 0 .5 .5 -.5 -.5 (the first t-contrast vector) would run normally. Could anyone explain what is going on here? I'll go ahead and see if 1 1 -1 -1 works for Flex factorials that don't consider the factor 'subject', although, as I understand, this would not be the correct way of evaluating repeated measures designs. Best regards, Eduardo