Geisser Seymour
School of Statistics, 313 Ford Hall, University of Minnesota, Minneapolis, MN 55455 USA.
Stat Med. 2003 Nov 15;22(21):3337-46. doi: 10.1002/sim.1635.
We revisit a former problem that led to much commentary regarding a proposed use of testing whether profiles for several groups in a repeated measurement scenario were parallel. Here we review previous approaches and propose Bayesian model selection to indicate whether the profiles can be considered parallel and if so support an estimation procedure for the profile differences.