ABSTRACT
In multiple regression the same goal of predicting a single variable remains, but several regression predictors are used. If one intends only to use a multiple regression predictor to estimate values of the variable for some new sets of predictors, then one can find the prediction equation without worrying about what it means. Such uses are rare; more often, one also wants to assign meaning to the predictor to say what it represents and to describe how it is related. Indeed, despite the prediction-oriented terminology, in many applications of multiple regression the interpretation is the only point of the analysis and actual predictions are never made. However, interpreting a multiple regression variable is less obvious than it seems, particularly when the component variables are substantially correlated. The geometric representation is very helpful in working out the relationships among the variables. Both this section and the next chapter concerned the task of interpretation.
