Abstract: A computer method improves the robustness of population parameters estimated using sample responses. Two different factors that may impact the accuracy of estimates are considered. The first factor is referred to as statistical outliers. By statistical outliers, what is meant is observations that fall “statistically outside” of the other remaining observations in the sample. The second factor considered is the impact of the weight assigned to each observation on the overall parameter estimate. More specifically, the fact that weights assigned to each respondent are typically estimated and thus not exact is addressed. Consequently, the weights do not unduly influence the value of the parameter estimates.

Abstract: A computer method improves the robustness of population parameters estimated using sample responses. Two different factors that may impact the accuracy of estimates are considered. The first factor is referred to as statistical outliers. By statistical outliers, what is meant is observations that fall “statistically outside” of the other remaining observations in the sample. The second factor considered is the impact of the weight assigned to each observation on the overall parameter estimate. More specifically, the fact that weights assigned to each respondent are typically estimated and thus not exact is addressed. Consequently, the weights do not unduly influence the value of the parameter estimates.