Abstract
This chapter presents brief descriptions of several topics in regression analysis—some that might occur because of problems with the data set and some that are extensions of the basic analysis. It discusses methods used to alleviate the effects of missing values, defines dummy variables and shows how to code them as well as a brief introduction to regression trees. Missing values are often classified into three types: missing completely at random; missing at random; and nonrandom missing data. For nonrandom missing data, the methods for treating nonresponse depend on the particular response bias that has occurred and it may be difficult to determine what is happening. One method that has been used to improve on single imputation is multiple imputation. Imputed values are generated from the incomplete data set using one of the stochastic imputation methods discussed and used to create multiple complete data sets.