THE IDENTIFICATION OF MULTIPLE OUTLIERS

One approach to identifying outliers is to assume that the outliers ha ve a different distribution from the remaining observations. In this a rticle we define outliers in terms of their position relative to the m odel for the good observations. The outlier identification problem is then the problem of identifying those observations that lie in a so-ca lled outlier region. Methods based on robust statistics and outward te sting are shown to have the highest possible breakdown points in a sen se derived from Donoho and Huber. But a more detailed analysis shows t hat methods based on robust statistics perform better with respect to worst-case behavior. A concrete outlier identifier based on a suggesti on of Hampel is given.