In the analysis of time series data, it is frequently of interest to compare a set of forecasts made at some time point when a change in level is suspected to have taken place with actual observations. This paper discusses various statistical techniques for making such comparisons, and illustrates these techniques in terms of an actual example concerning the monthly averages of atmospheric ozone concentration of Azusa, California. This particular comparison is of interest because new automobile emissions standards that were introduced at the end of 1970. These measures might have reduced ozone below levels expected if no new standards had been introduced.
展开▼
