ABSTRACT
Linear regression is about predicting a continuous response variable y from one or more predictor variables x. The linear regression model specifies the mean of the response variable as a linear combination of the predictor variables: E[y |x] = x′β . Although x is often random, we always condition on it in this chapter, hence we treat it as fixed. Since covariate combinations can consist of continuous variables or categorical variables, standard ANOVA and ACOVA models are special cases of these linear models. Theoretical results in this chapter rely heavily on multivariate normality, cf. Example B.1.
