METHOD Subcommand (LOGISTIC REGRESSION command)

METHOD indicates how the independent variables enter the model. The specification is the METHOD subcommand followed by a single method keyword. The keyword METHOD can be omitted. Optionally, specify the independent variables and interactions for which the method is to be used. Use the keyword BY between variable names of an interaction term.

• If no variable list is specified, or if the keyword ALL is used, all of the independent variables following the keyword WITH on the VARIABLES subcommand are eligible for inclusion in the model.
• If no METHOD subcommand is specified, the default method is ENTER.
• Variables that are specified on CATEGORICAL are replaced by sets of contrast variables. The set of contrast variables associated with a categorical variable is entered or removed from the model as a single step.
• Any number of METHOD subcommands can appear in a Logistic Regression procedure. METHOD subcommands are processed in the order in which they are specified. Each method starts with the results from the previous method. If BSTEP is used, all remaining eligible variables are entered at the first step. All variables are then eligible for entry and removal unless they have been excluded from the METHOD variable list.
• The beginning model for the first METHOD subcommand is either the constant variable (by default or if NOORIGIN is specified) or an empty model (if ORIGIN is specified).

The available METHOD keywords are as follows:

ENTER. Forced entry. All variables are entered in a single step. This setting is the default if the METHOD subcommand is omitted.

FSTEP. Forward stepwise. The variables (or interaction terms) that are specified on FSTEP are tested for entry into the model one by one, based on the significance level of the score statistic. The variable with the smallest significance less than PIN is entered into the model. After each entry, variables that are already in the model are tested for possible removal, based on the significance of the conditional statistic, the Wald statistic, or the likelihood-ratio criterion. The variable with the largest probability greater than the specified POUT value is removed, and the model is reestimated. Variables in the model are then evaluated again for removal. When no more variables satisfy the removal criterion, covariates that are not in the model are evaluated for entry. Model building stops when no more variables meet entry or removal criteria or when the current model is the same as a previous model.

BSTEP. Backward stepwise. As a first step, the variables (or interaction terms) that are specified on BSTEP are entered into the model together and are tested for removal one by one. Stepwise removal and entry then follow the same process as described for FSTEP until no more variables meet entry or removal criteria or when the current model is the same as a previous model.

The statistic that is used in the test for removal can be specified by an additional keyword in parentheses following FSTEP or BSTEP. If FSTEP or BSTEP is specified by itself, the default is COND.

COND. Conditional statistic. This setting is the default if FSTEP or BSTEP is specified by itself.

WALD. Wald statistic. The removal of a variable from the model is based on the significance of the Wald statistic.

LR. Likelihood ratio. The removal of a variable from the model is based on the significance of the change in the log-likelihood. If LR is specified, the model must be reestimated without each of the variables in the model. This process can substantially increase computational time. However, the likelihood-ratio statistic is the best criterion for deciding which variables are to be removed.

Example

LOGISTIC REGRESSION 
  VARIABLES = PROMOTED WITH AGE JOBTIME JOBRATE RACE SEX AGENCY
 /CATEGORICAL RACE SEX AGENCY
 /METHOD ENTER AGE  JOBTIME
 /METHOD BSTEP (LR) RACE SEX JOBRATE AGENCY.

• AGE, JOBTIME, JOBRATE, RACE, SEX, and AGENCY are specified as independent variables. RACE, SEX, and AGENCY are specified as categorical independent variables.
• The first METHOD subcommand enters AGE and JOBTIME into the model.
• Variables in the model at the termination of the first METHOD subcommand are included in the model at the beginning of the second METHOD subcommand.
• The second METHOD subcommand adds the variables RACE, SEX, JOBRATE, and AGENCY to the previous model.
• Backward stepwise logistic regression analysis is then done with only the variables on the BSTEP variable list tested for removal by using the LR statistic.
• The procedure continues until all variables from the BSTEP variable list have been removed or the removal of a variable will not result in a decrease in the log-likelihood with a probability larger than POUT.