For crossover designs
Download as PPTX, PDF4 likes5,496 views
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.
For crossover designs
- 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 – 2*3=6 Put the within subject factors in the box. Pay close attention to the combination numbers – 1,1; 1,2 etc in the parentheses. These signify the intervention and the time point – 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 “Display”.
- 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: •Group*Interventi on interaction – for crossover bias. •Intervention main effect – for difference of overall values of dep var between interventions. •Time*Interventi on interaction – 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.