Penentuan Nilai Pembobotan dan Penduga Ragam Pada Penarikan Contoh Bertahap (Studi Kasus : Survei Pra Pemilihan Gubernur dan Wakil Gubernur Jawa Barat di Daerah Pemilihan Kota Bogor)

Abstract

Sampling can be performed by using one or more stages. These stages are useful in order to obtain representative samples. It is often observed that parameters are estimated by assuming the samples are taken using simple random sampling (one stage) although they are taken using multistages which lead to invalid results. This study focus on estimation of parameters considering the stage in multistage sampling. As an illustration, we use pre-election survey of Governor and Vice Governor of West Java in Bogor City which involving three stages of selection, namely : (1) village within district, (2) neighborhood within village and (3) household within neighborhood. These stages are taken into account during estimation of number of voters in form of weights which are obtained by considering the selection process. It is obtained that the estimation using the weights are closer to the true number of voters released by KPU Bogor compared to the one not using the weights.