Overview (GENLOG command)
GENLOG
is
a general procedure for model fitting, hypothesis testing, and parameter
estimation for any model that has categorical variables as its major
components. As such, GENLOG
subsumes
a variety of related techniques, including general models of multiway
contingency tables, logit models, logistic regression on categorical
variables, and quasi-independence models.
GENLOG
, following
the regression approach, uses dummy coding to construct a design matrix
for estimation and produces maximum likelihood estimates of parameters
by means of the Newton-Raphson algorithm. Since the regression approach
uses the original parameter spaces, the parameter estimates correspond
to the original levels of the categories and are therefore easier
to interpret.
HILOGLINEAR
,
which uses an iterative proportional-fitting algorithm, is more efficient
for hierarchical models and useful in model building, but it cannot
produce parameter estimates for unsaturated models, does not permit
specification of contrasts for parameters, and does not display a
correlation matrix of the parameter estimates.
The General Loglinear Analysis and Logit Loglinear
Analysis dialog boxes are both associated with the GENLOG
command. In previous releases, these
dialog boxes were associated with the LOGLINEAR
command. The LOGLINEAR
command
is now available only as a syntax command. The differences are described
in the discussion of the LOGLINEAR
command.
Options
Cell Weights. You can specify cell weights
(such as structural zero indicators) for the model with the CSTRUCTURE
subcommand.
Linear Combinations. You can compute linear combinations of observed cell frequencies,
expected cell frequencies, and adjusted residuals using the GRESID
subcommand.
Generalized Log-Odds Ratios. You can specify contrast variables on the GLOR
subcommand and test whether the generalized log-odds
ratio equals 0.
Model Assumption. You can specify POISSON
or MULTINOMIAL
on the MODEL
subcommand to
request the Poisson loglinear model or the product multinomial loglinear
model.
Tuning
the Algorithm. You can control the values of algorithm-tuning
parameters with the CRITERIA
subcommand.
Output Display. You can control the output display with the PRINT
subcommand.
Optional Plots. You can request plots
of adjusted or deviance residuals against observed and expected counts,
or normal plots and detrended normal plots of adjusted or deviance
residuals using the PLOT
subcommand.
Basic Specification
The basic
specification is one or more factor variables that define the tabulation.
By default, GENLOG
assumes a
Poisson distribution and estimates the saturated model. Default output
includes the factors or effects, their levels, and any labels; observed
and expected frequencies and percentages for each factor and code;
and residuals, adjusted residuals, and deviance residuals.
Limitations
- Maximum 10 factor variables (dependent and independent).
- Maximum 200 covariates.
Subcommand Order
- The variable specification must come first.
- Subcommands can be specified in any order.
- When multiple subcommands are specified, only the last specification takes effect.