2

I have a study in which 60 subjects took both drug A on one day and drug B on another. The order of drug administration was counterbalanced, such that 30 subjects took drug A on the first day and 30 took drug B on the first day. (Each subject then received the other drug on the second day.) Our dependent variable was a continuous measure of performance on a behavioral task. I would like to evaluate whether drug condition affects behavioral performance, but I suspect that there may be an effect of the order of drug administration.

I believe the goal would be to treat subject as a random effect, drug condition as a within-subject repeated measure, and order of drug administration as a between-subjects measure. Would a split-plot ANOVA be appropriate? My data happen to be in Matlab; as a bonus, would anyone know of a way to do so in Matlab?

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
Andy
  • 21
  • 1

1 Answers1

1

This is a cross-over design. Typically to compare drugs in this case you would look at differences paired on subjects. It is not a split plot design.

Michael R. Chernick
  • 39,640
  • 28
  • 74
  • 143