The SURVEYPHREG Procedure

The replicate weights variance estimation method is a general-purpose variance estimation method that uses the replicate weights and replicate coefficients that you provide by using the REPWEIGHTS statement and the REPCOEFS= option, respectively.

If you provide your own replicate weights in a REPWEIGHTS statement but do not specify replicate coefficients in a REPCOEFS= option, then the default replicate coefficient depends on the VARMETHOD= option in the PROC SURVEYPHREG statement as shown in the following table:

Value of VARMETHOD=	Default Replicate Coefficient
None specified
BOOTSTRAP
BRR
JACKKNIFE

Let be the estimated proportional hazards regression coefficients from the full sample, and let and be the estimated regression coefficients and the replicate coefficient for the rth replicate, respectively. PROC SURVEYPHREG estimates the covariance matrix of by

with R degrees of freedom, where R is the number of replicates.

If you specify the CENTER=REPLICATES method-option, then PROC SURVEYPHREG computes the covariance matrix of by

where is the average of the replicate estimates as follows:

If one or more components of cannot be calculated for some replicates, then the variance estimator uses only the replicates for which the proportional hazards regression coefficients can be estimated. Estimability and nonconvergence are two common reasons why might not be available for a replicate sample even if is defined for the full sample. Let be the number of replicates where are available, and let be the number of replicates where are not available. Without loss of generality, assume that are available only for the first replicates; then the jackknife variance estimator is

with degrees of freedom. Alternatively, you can use the VADJUST=AVGREPSS option in the MODEL statement to use the average sum of squares for the invalid replicate samples. For more information, see Variance Adjustment Factors.