Chasing change: repeated-measures analysis of variance is so yesterday!

Change and growth are the bread and butter of rehabilitation research, but to date, most researchers have used less than optimal statistical methods to quantify change, its nature, speed, and form. Hierarchical linear modeling (HLM) (random/mixed effects or latent growth or multilevel modeling, individual/latent growth curve analysis) generally is superior to analysis of (co)variance and other methods, but has been underused in rehabilitation research. Apropos of the publication of 2 didactic articles setting forth the basics of HLM, this commentary sketches some of the advantages of this technique.