A comparison of uncertainty for unstratified, proportional-, and Neyman-allocated population estimates.

A comparison of the variance σ2 and the SEM (Standard Error of the Mean) of the Horvitz-Thompson (H-T) estimator for 1,000 simulated sampling trials, and a fixed sample size of 990. For the unstratified control case (A), all sections were assigned to a single stratum, in contrast to 4-level optimal stratification using either proportional (B) or Neyman allocation (C). The stratification variable is “persons per residential structure” and Table 2, subtable 2a, specifies the samples per stratum.
A comparison of uncertainty for unstratified, proportional-, and Neyman-allocated population estimates.