Bayesian Pseudo-Empirical-Likelihood and Scale-Load Inferences from Complex Survey Data

• Activity Number: 308
• Type: Invited
• Date/Time: Tuesday, August 5, 2014 : 10:30 AM to 12:20 PM
• Sponsor: International Society for Bayesian Analysis (ISBA)
• Author(s): J. N. K. Rao and Changbao Wu*+
• Companies: Carleton University and University of Waterloo
• Keywords: scale load ; finite populatio parameters ; empirical likelihood ; pseudo posterior inferences

Abstract

Bayesian methods to inference on finite population parameters by using sample survey data face hurdles in all three phases of the inferential procedure: the formulation of a likelihood function, the choice of a prior and the validity of postrior inferences under the design-based frequentist framework. We first review the past work of Hartley and Rao (1968)based on a scale-load approach, similar to Owen's(1988) empirical likelihood approach. For complex survey data, we propose a pseudo-profile empirical likelihood approach to construct psuedo-posterior intervals, using non-informative prior, that are asymptotically valid under the design-based approach. We also extend the scale-load approach to complex survey data. Rao and Wu (2010) studied the above approaches for estiamting a finite population mean. We present extensions covering more complex finite population parameters such as regression coefficients and quantiles.