Identifying treatment effects in univarate time series using a joint estimation procedure

Researchers use interrupted time-series analysis to identify the influence of treatments on social or organizational processes. Methods for analyzing interrupted time-series, such as intervention analysis and the Chang et al. procedure have limitations that could lead to erroneous conclusions. This paper presents an application of a new joint estimation procedure in gauging the impact of treatments. This procedure locates outliers (treatments) and estimates the series parameters jointly and, consequently, is relatively robust. As examples demonstrate, this robustness helps in identifying the impact more accurately. Finally, the authors believe this procedure should gain general acceptance with social scientists and organizational researchers given that it allows for greater flexibility and is easier to use.