We ran a mixed design repeated measures ANOVA for an RCT examining the utility of a specific treatment on knee range of motion (flexion). Our between subjects factor was study condition (treatment vs. control) and our repeated measure was Initial Flexion vs. Discharge Flexion. We showed a significant effect for TIME, but not TIME*Condition. However, the between subjects test showed a significant difference for Condition. How can I interpret this?
Using a standard framework: your TIME main effect is about time collapsing over CONDITION. Your CONDITION variable is about condition collapsing over TIME. Your interaction is simply stated as "the effect of time depends on condition" or "the effect of condition depends on time". You are making an accept/reject decision about the interaction, but a more nuanced (and appropriate) approach would be to look at the effect size. If the effect size of your interaction really is 0, then you only have evidence that A) there were differences (regardless of condition) as a function of time and B) control was different from treatment. A is probably acceptable for you. It is B without a Condition x Time interaction that is bothering you. B indicates that there are differences between treatment and control, but that the effect of time is the same for both. Therefore, there may have been baseline differences between those in treatment and control. You should check for this using a between subjects test using only the Initial Flexion data points. However, a better solution is just to plot the means for all four cells of your design. This plot will show both your main effects and your interaction (or lack thereof).