Mathematics and Statistics 11(3): 579-585, 2023 http://www.hrpub.org
DOI: 10.13189/ms.2023.110315
Small Area Estimation of Illiteracy Rates based on
Beta-Binomial Model using Hierarchical
Likelihood Approach
Etis Sunandi
1,2
, Khairil Anwar Notodiputro
1,*
, Indahwati
1
, Agus M Soleh
1
1
Department of Statistics, Faculty of Mathematics and Natural Sciences, IPB University, Indonesia
2
Department of Mathematics, Faculty of Mathematics and Natural Sciences, University of Bengkulu, Indonesia
Received December 31, 2022; Revised April 20, 2023; Accepted May 9, 2023
Cite This Paper in the Following Citation Styles
(a): [1] Etis Sunandi, Khairil Anwar Notodiputro, Indahwati, Agus M Soleh , "Small Area Estimation of Illiteracy Rates
based on Beta-Binomial Model using Hierarchical Likelihood Approach," Mathematics and Statistics, Vol. 11, No. 3, pp.
579 - 585, 2023. DOI: 10.13189/ms.2023.110315.
(b): Etis Sunandi, Khairil Anwar Notodiputro, Indahwati, Agus M Soleh (2023). Small Area Estimation of Illiteracy Rates
based on Beta-Binomial Model using Hierarchical Likelihood Approach. Mathematics and Statistics, 11(3), 579 - 585.
DOI: 10.13189/ms.2023.110315.
Copyright©2023 by authors, all rights reserved. Authors agree that this article remains permanently open access under
the terms of the Creative Commons Attribution License 4.0 International License
Abstract Small Area Estimation (SAE) is a statistical
method used to estimate parameters in sub-populations
with small samples. This study aims to develop a Beta-
Binomial model on SAE with a Hierarchical Likelihood
(HL) approach. The model built is called the SAE-BB-HL
model. This research begins by deriving a formula for
estimating model parameters analytically. A good fit is
calculated with the Mean Square Error of Prediction
(MSEP) and bias. This study used simulation data and data
from the National Socio-Economic Survey (SUSENAS)
and Village Potential (PODES) of Bengkulu Province for
2021 collected by Statistics Indonesia (BPS). The
simulation study aims to evaluate the SAE-BB-HL model.
Simultaneously, the application study aims to predict the
illiteracy rate per sub-district in Bengkulu Province. The
simulation study results show that the parameter estimates
of random area distribution are very close to the actual
parameters. It also reveals that the bias and MSEP
estimates of the proportion of HL are lower than the direct
estimates. In addition, the results of this study show that the
SAE-BB-HL model can improve the accuracy and
precision of proportion estimation. Applying the SAE-
BB_HL model to real data shows that the predictive value
of the illiteracy rate tends to be higher when compared to
the direct estimator.
Keywords Binary Response, Mean Square Error,
Overdispersion, Small Sample
1. Introduction
Small Area Estimation (SAE) is a statistical method for
estimating parameters in a subpopulation based on a small
number of samples to provide estimates with adequate
precision [1]. Small area models are classified into two
types based on the type of data available. The first is the
area-level model. This model connects the direct estimator
and the auxiliary area variable. The second model is a unit-
level model that connects the unit values of the estimator
directly to the unit level of auxiliary variables with known
area mean values and certain area auxiliary variables. The
SAE model can be applied to both continuous and discrete
response variables such as binary. Research using binary
data is prone to overdispersion. The Beta-Binomial
distribution can be used to overcome the problem of
overdispersion in binary data [2]. Beta-Binomial model
development on small area estimation has been carried out.
The model parameters are estimated through the Bayesian
approach.
According to [3], parameter estimation in a mixed model
with a Beta-Binomial hierarchy can be done through
Hierarchical Likelihood (HL). This method is claimed to
be better than the Bayes approach analytically. The HL
method can reduce the bias of estimating binary data