BAYESIAN PREDICTIVE INFERENCE WITH SURVEY WEIGHTS

Yang, Lingli

Etd

BAYESIAN PREDICTIVE INFERENCE WITH SURVEY WEIGHTS

Public

Sample surveys play a significant role in obtaining reliable estimators of a finite population. In a world of "big data", a large amount of available non-probability samples are easier and faster to obtain than probability samples. In this research, we focus on binary data which occur in many different situations. The main idea of this research is to compare the performance of nine methods with different constructed survey weights, and we can use these methods for non-probability sampling after weights are estimated (e.g. quasi-randomization). In particular, we employ original weights, adjusted weights, adjusted standardized weights, and trimmed weights to build posterior distributions. We apply our models to the simulation study and compare their performance by posterior mean, posterior standard deviation, relative bias, posterior root mean squared error, and the coverage rate of 95% credible intervals. Also, we discuss an application on body mass index and compare these nine models.

Creator