Use of pooled time series in the study of naturally occurring clinical events and problem behavior in a foster care setting.

Abstract

Pooled time series is an underused analytic technique with the potential to increase researchers' ability to exploit clinical data. This article demonstrates the value of pooled time series by analyzing the behavior of youths in a specialized foster care treatment setting in response to a naturally occurring clinical event: changes in the number of youths living together in a treatment foster care setting. Pooled time series moves beyond typical clinical analyses with an increased capability of controlling statistically for complex within-subject effects and with a clinically useful measure of effect size. The complexity of the intrasubject data made it virtually impossible to determine the relevant significance (i.e., clinical meaning) of the clinical event by the use of standard n = 1 visual analysis procedures or standard statistical methods (e.g., chi square). After things such as autocorrelation and individual time trends were statistically controlled, each additional youth increased the number of problematic behaviors by one behavior per youth per day on the Parent Daily Report.