ABSTRACT
This chapter discusses the situation where the values of one variable, called the response or dependent variable, depend on the selected values of one variable which can be determined by the experimenter. Such variables are variously called controlled, regressor, predictor, explanatory or independent variables, though the latter adjective is best avoided. The analysis of the linear regression model can be extended in a straightforward way to cover situations in which the dependent variable is affected by several controlled variables, or in which it is affected non-linearly by one controlled variable. To derive confidence intervals for the regression parameters and for the predicted values, it is necessary to make the assumptions of normality and constant variance as in linear regression. Some regression models involve a combination of multiple and curvilinear regression. Theoretical considerations may lead to a regression model which depends on a transformation of the controlled variables.
