The HPQUANTSELECT Procedure

WEIGHT Statement

  • WEIGHT variable;

The variable in the WEIGHT statement is used as a weight to perform a weighted analysis of the data. Observations that have nonpositive or missing weights are not included in the analysis. If you do not use a WEIGHT statement, all observations that are used in the analysis are assigned a weight of 1.

The HPQUANTSELECT procedure mainly uses each valid weight as the scale factor of its relevant observation. Let bold upper X denote the design matrix before using weights, and let bold upper W denote the diagonal matrix whose diagonal elements are the weights. Then the HPQUANTSELECT procedure mainly uses the weighted design matrix bold upper W bold upper X for its computation. For example, the weighted version of bold upper X prime bold upper X is bold upper X prime bold upper W bold upper W bold upper X but not bold upper X prime bold upper W bold upper X. The only exception is in computing the standardized parameter estimates, which computation uses the weighted standard deviation of the dependent variable and all regressors. These standard deviations are based effectively on the diagonal of bold upper X prime bold upper W bold upper X.

Last updated: December 09, 2022