ABSTRACT
The previous chapters have dealt almost exclusively with hypothesis
testing techniques for the analysis of contingency tables. In this
chapter an alternative approach w i l l be considered, namely that of
fitting models and estimating the parameters i n the models. T h e
term model refers to some 'theory' or conceptual framework about
the observations, and the parameters in the model represent the
'effects' that part icular variables or combinat ions of variables have
in determining the values taken by the observations. Such an
approach is common i n many branches of statistics such as regression
analysis and the analysis of variance. M o s t c o m m o n are linear models
which postulate that the expected values of the observations are
given by a linear combina t ion of a number of parameters. Techniques
such as maximum l ike l ihood and least squares may be used to
estimate the parameters, and estimated parameter values may then
be used in identifying which variables are of greatest importance i n
'predicting' the observed values.
