chapter  4
Simple Models: Statistical Inferences about Parameter Values
Pages 28

Although the sums of both the absolute errors and the squared errors seem to be reasonable alternatives, we decided in Chapter 3, for reasons of efficiency, practicality, and tradition, to define total error as the sum of the squared errors. In the simplest models, where we are estimating the single parameter β0, the choice of the sum of squared errors as the definition of error implies that the best estimate is the sample mean (as we proved in Chapter 2).