1

I'm trying to analyze data from a crossover study, where two treatments ($A$ and $B$) are applied to $n$ patients (each patient receive both treatments).

For each treatment I have an outcome with multiple measurement. For example patient 1 for treatment $A$ has $m_{A,1}$ measurements and for treatment $B$ has $m_{B,1}$ measurements. The measurement are relative to the same variable.

I'm interested in proving that treatment A and B do not differ.

If I had one measurement for each candidate and each treatment I would have done a paired test (paired t-test or Wilcoxon Signed-Rank test)

At the moment I'm averaging the measurements and conducting a paried test on the average.

Is there a bettere way to do it?

Teoz
  • 11
  • 2
  • I think crossover studies are not the best to investigate the difference between two different treatments but to investigate especially the best sequence of administration – GGA Jan 27 '16 at 22:16
  • I think crossover study can be used to investigate differences between treatment if washout period and carryover effects are taken into account. What worries me most, is that for each sample I have multiple measurements. I think this answer [here](http://stats.stackexchange.com/questions/181563/analyzing-repeated-measures-experiment-with-multiple-treatment-groups-and-multip?rq=1) can be a starting point. – Teoz Jan 29 '16 at 11:05

0 Answers0