Linear regression is a form of statistical analysis whereby values on one variable (the outcome variable, denoted by Y) are predicted from values on another variable (the predictor variable, denoted by X) with which they are correlated. Here, “predict” does not necessarily have a temporal meaning but merely indicates that values on the outcome variable are estimated using values on the predictor variable. The analysis normally has one or both of two objectives: first, to obtain specific predicted values on Y that correspond to specific observed values on X; and second, to [Page 1517]estimate the strength of this predictive relationship—that is, how well does X perform as a predictor of Y? The simplest case of linear regression, to be considered here, is where, in addition ...