The next three sections provide basic information about the
modeling modules. As noted above, each of the models involves a quantity 
that is modeled as a function of
parameters and covariables using one of the general EPICURE model forms. The default model in
each of the modules is the loglinear model 
. Loglinear models can be specified and
fit using the command FIT varlist, where
varlist is a list of covariates to be included in the model.
A complete description of the EPICURE models is given in Fitting Models
with Epicure of this Guide and in Risk Models in
EPICURE Programs of the EPICURE Command Summary. In the following
sections we write the model as 
to emphasize the general models,
but it may be easiest to think in terms of a simple loglinear models as you read
the next few sections.
As described in Categorical Variables in EPICURE Models, categorical variables specify levels of factors for use in modeling or the computation of summary statistics. Variables are designated as categorical with the LEVELS command. When a categorical variable is included in a model, an indicator variable is created for each level or factor. Model specifications can include ordinary (continuous) variables, categorical variables, and products of two continuous or categorical variables. When a model includes the product of a categorical and a continuous variable, a separate parameter is estimated for each level of the categorical variable. When the product of two categorical variables is included in a model, parameters are estimated for each level of the cross-classification defined by the levels of the two factors.
The basic EPICURE model can be extended to include
multiplicative strata parameters. The extended models can be written as 
, where the 
are the stratum parameters. The
strata are defined by a set of categorical variables. Because the stratum
parameters are estimated using a computationally efficient algorithm that does
not require the inversion of large matrices, models can include large numbers of
such parameters. Stratification is discussed further in Stratified Models.
Model specification in EPICURE requires specification of key variables, such as the response variable, number of trials, or time at risk. These key variables are described below and in more detail in Key Variables for the Regression Programs. As noted below, each module has its own set of key variables and automatically recognizes certain default names for these variables. In addition, as discussed in Key Variables for the Regression Programs of this guide and in Key Variable Commands of the EPICURE Command Summary, you can easily specify alternative names for these variables.