Online Program

Return to main conference page

All Times ET

Program is Subject to Change

Tuesday, June 15
Tue, Jun 15, 11:30 AM - 1:00 PM
Treatments for Nonresponse: Imputation, Weighting, and Other Models

Variance estimation of calibration estimators under nonresponse in survey sampling (308178)

*Per Gösta Andersson, Stockhom University 

Keywords: Calibration, confidence intervals, nonresponse

An essential part of the estimation process of totals or means of a finite population is the variance estimation. When we have to deal with nonresponse this problem becomes more delicate. We will show how some previously suggested techniques work when constructing confidence intervals for a population total based on normal approximation assumptions. The quality of the intervals are measured in terms of coverage and length. However, intervals of this type often suffer from inadequate coverage and noncoverage probabilities. In our investigation we will as point estimators consider different types of calibration estimators using auxiliary information at the sample level and/or the population level. To improve on the quality of the resulting confidence intervals a theoretically motivated adjustment technique will be suggested and empirically demonstrated.