Abstract
We describe a test statistic for a changepoint in a sequence of repeated measurements. This statistic is non-parametric, and may be motivated from either frequentist or Bayesian considerations. We apply this procedure to data arising from a clinical sleep study of low-energy emission therapy.