TRUMP: Tuned Ratio Unbiased Mean Predictor

• Activity Number: 301 - Weighting Adjustments
• Type: Contributed
• Date/Time: Tuesday, August 1, 2017 : 8:30 AM to 10:20 AM
• Sponsor: Survey Research Methods Section
• Author(s): Sarjinder Singh* and Stephen A. Sedory
• Companies: Texas A&M University-Kingsville and Texas A&M University-Kingsville
• Keywords: Calibration ; unbiasedness ; model assistance ; performance ; simulation study

Abstract

In this paper, we show that the new concept for tuning design weights in survey sampling developed by Singh, Sedory, Rueda, Arcos and Arnab (2015: Elsevier) leads to an innovative Tuned Ratio Unbiased Mean Predictor (TRUMP) with the assistance of a model. It is shown theoretically that the proposed TRUMP is more efficient than the ratio estimator due to Cochran (1940). Although there is no need of empirical investigations, but a small scale simulation study will be discussed. The proposed TRUMP has potential to be extended to the regression predictor and other complex survey sampling designs.