GLZ Syntax - Overview

The GLZ syntax consists of command lines, which are terminated by semicolons. The complete command syntax for GLZ is shown below. Note that terms in curly brackets ({}) denote alternate specifications (e.g., MBUILD = ALL or MBUILD = BESTSUBSET); optional keywords are enclosed in brackets ([]); all keywords are shown in upper case letters (even thought the syntax interpreter is not case sensitive).

GLZ and GLM syntax conventions
The GLZ syntax conventions are essentially identical to the GLM syntax conventions [see General Linear Model (GLM), with only a few additions and omissions relating to 1) the stepwise and best subset analyses, 2) the technical parameters for the iterative estimation procedure, and 3) the specification of models for categorical dependent (response) variables (i.e., when the distribution selected on the GLZ Startup Panel - Advanced tab is either Binomial, Multinomial, or Ordinal multinomial]. Also, in GLZ you cannot specify 1) within subject (repeated-measures) designs, 2) mixed-model ANOVA and ANCOVA designs for random effects, or 3) multivariate analyses (with multiple dependent (response) variables). The following is a list of keywords available in GLZ; note that the General Regression Models (GRM) (see the next paragraph) and General Partial Least Squares Models (PLS) modules also use similar syntax, with sets of additional keywords (options) that are specific to those particular applications.
GLZ and GRM syntax conventions
Many of the model building options (see MBUILD) are identical to those available in the General Regression Models (GRM) module (see GRM syntax conventions), with the exception of some of the technical parameters that guide the respective model building methods.

[GRM;]

RESPONSE [=] { Variable }
    { Variable (Value List) };
    [GROUPS [=] { NONE }
    { Variable [(Value List)
     .......
    Variable [(Value List)] };]
[COVARIATE [=] { NONE }
    { Variable List };]
[DESIGN [=] { NONE }
    { Effects };]
[INTERCEPT  [=] { INCLUDE }
    { EXCLUDE };]
[PARAM [=] { SIGMA }
    { OVERP };]
[COUNTV [=] { NONE }
    { Variable };]
[SDELTA [=] { Value };]
[SURFACE [=] { NONE }
    { Covariate Variable List };]
[MIXTURE [=] { NONE }
    { Covariate Variable List };]
[SAMPLE [=] { NONE }
     { Variable (Value) };]
[MBUILD [=] { all }
    { FORWARD };
    { BACKWARD };
    { FSTEPWISE };
    { BSTEPWISE };
    { BESTSUBSET };]
[MAXSTEP [=] Integer Value;]
[P1ENTER [=] Value;]
[P2REMOVE [=] Value;]
[BESTCRIT [=] { LIKELIHOOD }
    { SCORE }
    { AIC };]
[MAXITER [=] Value;]
[CONVERGE [=] Integer Value;]
[INITIALS [=] { NONE }
    { USER [(Value List)] };]
[OFFSET [=] { NONE }
    { Variable };]
[MAXSUB [=] Integer Value;]
[OUTPUT [=] { NONE }
    { List of options (integers) };]
Example
Shown below is an example specification for a stepwise regression analysis that includes two categorical predictors and their interaction, as well as continuous predictor variables. To run this example, use example data file Exp.sta, and on the GLZ Startup Panel - Advanced tab select Normal and Identity as the Link function and Distribution, respectively.

GLZ;

RESPONSE = STRESS_R;
GROUPS = GROUP GENDER;
COVARIATE = CORRECT1 CORRECT2 CORRECT3;
DESIGN = CORRECT1+CORRECT2+CORRECT3 + GROUP | GENDER;
MBUILD = FSTEPWISE;
MAXSTEP = 10;
P1ENTER = .10;
P2REMOVE = .10;

For more examples, see the GLZ Syntax Examples section.

Upper case and lower case letters
The syntax editor is not case sensitive; i.e., DEPENDENT, dependent, or Dependent are all valid specifications of the same keyword.
Order of keywords
The order of keywords should follow that implied in the complete listing of all keywords shown above. Note that, while most of the options can be specified in any order, the Design statement must follow the Response, Groups, and Covariate specifications. In general, when a keyword references a variable or effect previously defined, then that specification must occur before the respective keyword reference.
Comments
You can insert comments into the syntax by placing them in curly brackets (e.g., { This is a comment }), or after a pound sign (e.g., ... # This is a comment; all text starting at the pound sign to the end of the respective line will be treated as a comment). You can place comments in between any keywords of the syntax, and you can include several comment lines. Comments will be ignored when the syntax is parsed.

See also, GLZ - Index.