An adaptation of Page's CUSUM test for change detection (Q2568547)

The authors consider the problem of change detection in a parametric setup. An extension of the common CUSUM test is proposed which is based on the score function and accounts for nuisance parameters. Under the null hypothesis an almost sure approximation is obtained for the sequence of test statistics by partial sums of iid random variables. The test is illustrated by an example of monitoring a nested random effects model, where changes in different parameters may occur.