Change in outcome trajectory over time in treatment vs. control group

Question

I have randomized controlled trial data (2 conditions - randomization on the individual) and I would like to look at 2 psychological scale continuous variables over time as outcomes, in 2 separate analyses. I have about 250 people with complete data at 5 discrete time points. My hypothesis is that anxiety decreased over time for the entire sample, with no difference by treatment condition, and that self esteem increased over time, but only for the treatment group.

I would ideally like to use all time points and not throw out data. However, I am concerned with gist change, not the specific shape of the trajectory. For example, a fairly linear decrease, a monotonic nonlinear decrease, and a nonmonotonic trajectory with an overall decrease across the 5 times would all support my anxiety hypothesis. I would ideally like to run both an unadjusted and a covariate-adjusted model. I have varied degrees of familiarity with methods like linear mixed effects regression, difference in difference regression at multiple times, and mixed ANOVA, but it's been challenging to figure out what is most appropriate.

Do you have any recommendations about where to start?

I can most easily use SAS or SPSS, in case that's of consequence for your response.

Answer 1

I just wanted to follow up that I researched this more and determined that growth curve modeling would be the most appropriate method since it will allow me to contrast slopes across groups and test for a statistically significant difference in slopes, as well as include all data points for all participants and avoid losing data due to missed follow up time points.