ABSTRACT
The general denomination of regression m odels is used to identify sta tistical models for the relationship between one or more exp lan atory (in dependent) variables and one or more respon se (dependent) variables. Typical examples include the investigation of the influence of:
i) the amount of fertilizer on the yield of a certain type of crop; ii) the type of treatment and age on the serum cholesterol levels of pa
tients; iii) the driving habits and fuel type on the gas mileage of a certain make
of automobile; iv) the type of polymer, extrusion rate and extrusion tem perature on the
tensile strength and number of defects/unit length of synthetic fibers. W ithin this class, the so-called linear m od els play an im portant role for statistical applications; such models are easy to interpret, m athematically tractable and may be successfully employed for a variety of practical sit uations as in (i)-(iv) above. They include models usually considered in Linear R egression A n alysis, A nalysis o f V ariance (ANOVA), A nal ysis o f Covariance (ANCOVA) and may easily be extended to include Logistic R egression A n alysis, G eneralized Linear M odels, M u lti variate R egression , M u ltivaria te A nalysis o f V ariance (MANOVA) or M u ltivariate A nalysis o f C ovariance (MANCOVA).