This document provides instructions for conducting a 2-way repeated measures ANOVA (rmANOVA) with a crossover design in 3 steps: 1) Define the number of groups, time points, and within-subject factors, 2) Specify plots and adjustment for multiple comparisons, 3) Pay close attention to interaction effects in the results table to check for crossover bias and differences between interventions over time. Key results to interpret are interactions between group, intervention, and time, and separate rmANOVAs can further analyze time trends within each intervention.
2. The number of groups and the number of time
points for 2 way rmANOVA design have to be
defined on the first box. In this 3 time points and 2
interventions. So total combinations â 2*3=6
Put the within subject factors in the
box. Pay close attention to the
combination numbers â 1,1; 1,2 etc in
the parentheses. These signify the
intervention and the time point â
T1IV1; T1IV2; T2IV1 etc.
Put the crossover variable in the
between subjects factor box.
3. Specify the plots you want. You need them such
that you can make sense of the interaction of
Group with the other within subject variables.
And separate plots of intervention*time in the
two groups.
Put all the variables and interaction for
displaying means box, and put the alpha error
adjustment as Bonferroni.
Click the descriptives and effect size
estimates and homogeneity tests in âDisplayâ.
4. Results
Pay close attention to the within subject factors. They
should match your within subject variable groupings.
I am assuming you know how to go about this sphericity
table. If in doubt go through my rmANOVA notes.
5. Most important
table for you.
Look at p-values
for:
â˘Group*Interventi
on interaction â
for crossover
bias.
â˘Intervention
main effect â for
difference of
overall values of
dep var between
interventions.
â˘Time*Interventi
on interaction â
for trends of dep
var between the
interventions.
6. Use the estimated marginal means of
group*time*intervention for making line charts.
DO NOT try to interpret the between subject
effect of cross-over variable.
7. ⢠Interpret the profile plots
appropriately for interactions.
⢠Do separate rmANOVAs for the two
interventions separately to look at time
trends of dep var within each
intervention group.