Abstract

The empirical likelihood-based methods provide interesting ways to analyze complex survey data. Using various estimating equations, it easily incorporates many model and population-level constraints. It can also be used to justify a semi-parametric likelihood-based inference of model parameters when due to the complex sampling procedure the observed data has a distribution different from the model. We discuss the implications of a specific form of constraint on empirical likelihood-based parameter estimates. We focus on the properties of the optimal weights as well as on the estimation of standard errors. Finally, some implications of the methodology in the analysis of other well-known estimators will be discussed.