ABSTRACT
In this chapter different approaches to make factor analysis (FA) resistant against outliers discussed. There are two strategies to robustify FA: The first way is to find outlier identification and to exclude them from further analysis. This approach discussed in the next section, which also includes some basic tools and regression techniques from robust statistics essential for understanding the later sections. The second possibility, which will be the main focus of this chapter, is to build a FA method with the property that outliers bias the characteristic estimates. The resulting method called robust factor analysis, and accounts for violations from the strict parametric model. The robust method tries to fit majority data by reducing the impact of outlying observations. Results from robust factor analysis for changing only slightly if the outliers deleted beforehand from the data set. This approach results in a highly robust FA method with the more property that influential observations identified by inspecting the empirical influence function.
