A Box-Jenkins Approach to Modeling Outliers in Time Series Analysis

The sociologist or historian who wants to analyze time series data is often confronted with the fact that the data do not meet the requirements of the statistical models that he or she would like to apply. One of the problems commonly encountered is “outliers†that, if not treated properly, may distort model identification and parameter estimation. Statisticians working within the Box-Jenkins approach to time series analysis have recently developed a detection and estimation procedure that promises to be quite effective in modeling outliers. This procedure is introduced here in the version given by Chung Chen and Lon-Mu Liu. Several examples with simulated and real world data are presented.