CRITERIA Subcommand (COXREG command)

CRITERIA controls the statistical criteria used in building the Cox Regression models. The way in which these criteria are used depends on the method specified on the METHOD subcommand. The default criteria are noted in the description of each keyword below. Iterations will stop if any of the criteria for BCON, LCON, or ITERATE are satisfied.

BCON(value). Change in parameter estimates for terminating iteration. Alias PCON. Iteration terminates when the parameters change by less than the specified value. BCON defaults to 1E−4. To eliminate this criterion, specify a value of 0.

ITERATE(value). Maximum number of iterations. If a solution fails to converge after the maximum number of iterations has been reached, COXREG displays an iteration history showing the last 10 iterations and terminates the procedure. The default for ITERATE is 20.

LCON(value). Percentage change in the log-likelihood ratio for terminating iteration. If the log-likelihood decreases by less than the specified value, iteration terminates. LCON defaults to 1E−5. To eliminate this criterion, specify a value of 0.

PIN(value). Probability of score statistic for variable entry. A variable whose significance level is greater than PIN cannot enter the model. The default for PIN is 0.05.

POUT(value). Probability of Wald, LR, or conditional LR statistic to remove a variable. A variable whose significance is less than POUT cannot be removed. The default for POUT is 0.1.

Example

COXREG VARIABLES = SURVIVAL WITH GROUP AGE BP TMRSZ
 /STATUS = SURVSTA (1)
 /STRATA = LOCATION
 /CATEGORICAL = GROUP 
 /METHOD BSTEP
 /CRITERIA BCON(0) ITERATE(10) PIN(0.01) POUT(0.05).
  • A backward stepwise Cox Regression analysis is performed.
  • CRITERIA alters four of the default statistical criteria that control the building of a model.
  • Zero specified on BCON indicates that change in parameter estimates is not a criterion for termination. BCON can be set to 0 if only LCON and ITER are to be used.
  • ITERATE specifies that the maximum number of iterations is 10. LCON is not changed and the default remains in effect. If either ITERATE or LCON is met, iterations will terminate.
  • POUT requires that the probability of the statistic used to test whether a variable should remain in the model be smaller than 0.05. This is more stringent than the default value of 0.1.
  • PIN requires that the probability of the score statistic used to test whether a variable should be included be smaller than 0.01. This makes it more difficult for variables to be included in the model than does the default PIN, which has a value of 0.05.