When the classical least squares assumptions are not met, particularly with regard to the normality and homoscedasticity (equal variance) assumption, several transformations were well known long before the work of Box and Cox [1].
Chatterjee correctly states that Box and Cox [1] in their original research (and, actually, many others, e.g., ]3]) were mainly seeking power transformations for the response (i.e., dependent) variable only, while Box and Tidwell [2] were seeking power transformations for independent variables only.
307] states that, "The power transformation introduced by Box and Cox [1] and given by
While Box and Cox [1] focused mainly on the case of (3), they [1, pp.