UNS Conference Portal, The 1st International Conference on Mathematics: Education, Theory & Application

Font Size: 
Semiparametric Model-Based approach for SmallArea Estimation under Informative Sampling
Angela Nina Rosana Chytrasari, Sri Haryatmi, Danardono Danardono

Last modified: 2016-10-31

Abstract


Model-based approach has been frequently used in small area estimation. Underinformative sampling design, model holding for the sample may differ from model holding inthe population. Ignoring an informative sample may lead to erroneous inference. On the otherhand, the function of parametric mean model may restrictive in practice. The aim of this paperis to study an estimation method to predict area means based on nested eror regresion modelunder informative sampling scheme and unspecified function of mean model. The sample modelis extended by adding function of unit selection probability as an auxiliary variable to accountthe informativeness. Unspecified mean function and the additional term are approached thenby penalized spline. Bootstrap estimator of mean squared prediction error is also proposed.