|
Abstract:
|
Hartley and Rao (1968) proposed a nonparametric likelihood approach to inference from sample survey data called the "scale-load" approach. This method was discovered 20 years later (Owen, 1988) in mainstream statistics under the name "empirical likelihood" (EL), and its theory is discussed in detail in Owen's 2001 book. The EL approach is useful particularly for confidence interval estimation. Only recently, the EL has been revived in sample survey literature, and methods that take account of survey design features have been proposed in the context of using known population information and constructing confidence intervals on finite population parameters. This talk will present an overview of new developments in EL inference from sample survey data, focusing on known population information, imputation for missing data, integration of surveys, and confidence intervals.
|
