ABSTRACT
The multiple linear regression is handled in almost the same manner as the simple linear regression. Some of the highly sophisticated computer packages allow tests and confidence intervals about any weighted sums of the regression coefficients. Unfortunately, they tend to be somewhat unwieldy to use. In multiple linear regression the deterministic part of the model is usually some algebraic formula which is a weighted sum of the regression coefficients, using known weights. By contrast, the terms of the deterministic part of an analysis of variance model seem to be indexed variables and maybe a constant. The main difference is that the known weights occurring in these formulas are either 0 or 1. Mathematically the analysis of variance model is treated the same way as any multiple linear regression model. In the analysis of variance model, each indexed variable represents a sequence of distinct regression coefficients, and each of these regression coefficients must be estimated separately.
