I've got data that requires a split plot repeated measure ANOVA. Problem is I can't get SPSS to do post hoc on the repeated measures with all groups.
There are repeated measures taken for each subject. There are also several groups. SPSS 17 is giving the group-time interaction, but to do the post hoc, it appears one has to do it group by group? That seems cumbersome! Does graphpad also have similar problems?
Let Y1 to Y3 be the scores for the 3 levels of RMfactor, and let the between-Ss factor be Group. You can get all pairwise comparisons for the main effect of RMfactor and for the simple main effects by using COMPARE on the /EMMEANS sub-command. For example:
GLM Y1 Y2 Y3 BY Group
/WSFACTOR=RMfactor 3 Polynomial
/METHOD=SSTYPE(3)
/EMMEANS=TABLES(Group)
/EMMEANS=TABLES(RMfactor) COMPARE ADJ(BONFERRONI)
/EMMEANS=TABLES(Group*RMfactor) COMPARE(Group) ADJ(BONFERRONI)
/EMMEANS=TABLES(Group*RMfactor) COMPARE(RMfactor) ADJ(BONFERRONI)
/CRITERIA=ALPHA(.05)
/WSDESIGN=RMfactor
/DESIGN=Group.
Three methods of adjustment are available:
ADJ(LSD)
ADJ(BONFERRONI)
ADJ(SIDAK)
Look up EMMEANS in the Command Syntax Reference manual for more info.
For more powerful post hoc procedures I would run a Holm or Hochberg adjustment by hand (well actually I'd do them in R, but they are easy enough to work out by hand). The exception is for only three means (then the LSD method is very powerful without inflating the Type I error rate).
Please check this file. I hope it will help you
http://www.plantsciences.ucdavis.edu/agr205/Lectures/2011_Lectures/L12b_RepMeasures.pdf
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA