|
The objective of Regression Analysis: Statistical Modeling of a Response
Variable, Second Edition, is to provide tools necessary for using the modeling
approach for the intelligent statistical analysis of a response variable.
Although there is strong emphasis on regression analysis, there is coverage
of other linear models such as the analysis of variance, the analysis of covariance,
and the analysis of a binary response variable, as well as an introduction
to nonlinear regression.
The common theme is that we have observed sample or experimental
data on a response variable and want to perform a statistical analysis to
explain the behavior of that variable. The analysis is based on the proposition
that the behavior of the variable can be explained by: a model that (usually)
takes the form of an algebraic equation that involves other variables
that describe experimental conditions; parameters that describe how these
conditions affect the response variable; and the error, a catchall expression,
which simply says that the model does not completely explain the behavior
of the response variable. The statistical analysis includes the estimation of
the parameters, inferences (hypothesis tests and confidence intervals), and
assessing the nature (magnitude) of the error. In addition, there must be
investigations of things that may have gone wrong: errors in the data, poor
choice of model, and other violations of assumptions underlying the inference
procedures. |