The HPQUANTSELECT Procedure

SELECTION Statement

  • SELECTION <options>;

The SELECTION statement performs variable selection. The statement is fully documented in the section SELECTION Statement in Chapter 4: Shared Statistical Concepts.

The HPQUANTSELECT procedure supports the following suboptions in the METHOD= option in the SELECTION statement to specify the corresponding effect selection methods:

NONE

specifies no model selection.

FORWARD

specifies the forward selection method, which starts with no effects in the model and adds effects.

BACKWARD

specifies the backward elimination method, which starts with all effects in the model and deletes effects.

STEPWISE

specifies the stepwise regression method, which is similar to the forward selection method except that effects already in the model do not necessarily stay there.

By default, the METHOD=STEPWISE option is used in the SELECTION statement. If you do not use the SELECTION statement, the HPQUANTSELECT procedure fits the full model that is specified by the MODEL statement; this is equivalent to specifying the METHOD=NONE option in the SELECTION statement. For information about all the selection criteria that are used in PROC HPQUANTSELECT, see the section Criteria Used in Model Selection.

The DETAILS=ALL and DETAILS=STEPS options produce "Fit Statistics" and "Parameter Estimates" tables, which provide information about the model that is selected at each step of the selection process.

TEST=value

specifies a method to compute significance levels for the selection process. Table 13.4 summarizes these methods.

Table 13.4: Options for Significance Levels

Value of TEST=

  

Method

LR1

 

Likelihood ratio test Type I

LR2

 

Likelihood ratio test Type II

WALD

 

Wald score test


By default, PROC HPQUANTSELECT uses the Wald score to compute significance levels. If you specify the IID suboption in the SPARSITY option of the MODEL statement, the Wald score test uses the iid form of the covariance matrix to compute the Wald score and the associated significance levels. Otherwise, the non-iid form of the covariance matrix is used. The sparsity functions for both Type I and Type II likelihood ratio tests are estimated under the iid assumption no matter whether you specify the IID suboption.