FINITE POPULATION CORRECTIONS OF THE HORVITZ-THOMPSON ESTIMATOR AND THEIR APPLICATION IN ESTIMATING THE VARIANCE OF REGRESSION-ESTIMATORS

Kott (1988) first derived finite population corrections (fpcs) for the Horvitz-Thompson estimator under unequal probability sampling without replacement. No numerical comparison of variances with the fpcs was m ade. We derive some of the fpcs again in a new and more intuitive way without resorting to large sampling theory. We demonstrate how to inco rporate these fpcs with the jackknife variance estimator of the genera lized regression estimator (Sarndal 1980). Simulation results showed t hat adjustments by these fpcs resulted in variances with considerable less bias. One of the fpcs is also robust.