Several techniques, of increasing complexity, have been covered in this book. From fitting single means, to fitting multiple means, to fitting situations where the regressor is a continuous variable, specific techniques have been demonstrated to address a wide variety of statistical settings. This chapter introduces an approach involving general linear models, which encompasses all the models covered so far and extends to many more situations. They are all unified under the technique of least squares, which fits parameters to minimize the sum of squared residuals.
The techniques can be generalized even further to cover categorical response models, and other more specialized applications.