SAVE Subcommand (COXREG command)

SAVE saves the temporary variables created by COXREG. The temporary variables include:

SURVIVAL. Survival function evaluated at the current case.

SE. Standard error of the survival function.

HAZARD. Cumulative hazard function evaluated at the current case. Alias RESID.

LML. Log-minus-log-of-survival function.

DFBETA. Change in the coefficient if the current case is removed. There is one DFBETA for each covariate in the final model. If there are time-dependent covariates, only DFBETA can be requested. Requests for any other temporary variable are ignored.

PRESID. Partial residuals. There is one residual variable for each covariate in the final model. If a covariate is not in the final model, the corresponding new variable has the system-missing value.

XBETA. Linear combination of mean corrected covariates times regression coefficients from the final model.

  • To specify variable names for the new variables, assign the new names in parentheses following each temporary variable name.
  • Assigned variable names must be unique in the active dataset. Scratch or system variable names cannot be used (that is, the variable names cannot begin with # or $).
  • If new variable names are not specified, COXREG generates default names. The default name is composed of the first three characters of the name of the temporary variable (two for SE), followed by an underscore and a number to make it unique.
  • A temporary variable can be saved only once on the same SAVE subcommand.

Example

COXREG VARIABLES = SURVIVAL WITH GROUP
 /STATUS = SURVSTA (1)
 /STRATA = LOCATION
 /CATEGORICAL = GROUP 
 /METHOD = ENTER
 /SAVE SURVIVAL HAZARD.

COXREG saves cumulative survival and hazard in two new variables, SUR_1 and HAZ_1, provided that neither of the two names exists in the active dataset. If one does, the numeric suffixes will be incremented to make a distinction.