|
Abstract:
|
Two methods of interrupted time-series analysis known as ITSA and ITSACORR have become increasingly visible in published applied social science research during the past five years. Unfortunately, these methods contain fatal flaws that lead to greatly distorted intervention effect parameter estimates. The problems with the ITSE method (described in Huitema, 2004)are multiplied in the more recent ITSACORR approach. We demonstrate that neither the descriptive nor inferential properties of ITSACORR are satisfactory. Indeed, each component of the ITSACORR framework, including the structural model, the design matrix, the autocorrelation estimator, the ultimate parameter estimation scheme, and the inferential method contains fatal flaws. We recommend that the method not be used. Certain ARIMA and regression based methods provide far more satisfactory analysis options.
|
