Estimation of Education Indicators in East Java Using Multivariate Fay-Herriot Model
Novia Permatasari, Azka Ubaidillah
Abstract
Open-access reader
Novia Permatasari, Azka Ubaidillah
Abstract
Open-access reader
Education is an important aspect in improving human resources. Data availability of education indicators in a low administrative level is needed as a basis for education planning in that region. The problem of sample size when provide a low administrative level data can be overcome by indirect estimation, namely Small Area Estimation (SAE). SAE is able to increase the effectiveness of the survey sample size by using the strength of neighbouring areas and information from auxiliary variables related to the variables of interest. We obtain simulation study to compare multivariate model to univariate model and implement multivariate model to estimate three education indicators which are obtained from the National Socio-Economic Surveys by Statistics Indonesia. Simulation results are in line with previous studies, where the multivariate Fay-Herriot model with p variable has smaller of mean squares error (MSE) than the univariate model. The model implementation to estimate CrudeParticipation Rate (APK), School Participation Rate (APS), and Pure Participation Rate (APM) also shows that the multivariate model produces smaller RRMSE than the direct estimates. It can be concluded that multivariate model is able to produce more efficient estimates than direct estimation and univariate model.
OpenAlex reports 1 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Education is an important aspect in improving human resources. Data availability of education indicators in a low administrative level is needed as a basis for education planning in that region. The problem of sample size when provide a low administrative level data can be overcome by indirect estimation, namely Small Area Estimation (SAE). SAE is able to increase the effectiveness of the survey sample size by using the strength of neighbouring areas and information from auxiliary variables related to the variables of interest. We obtain simulation study to compare multivariate model to univariate model and implement multivariate model to estimate three education indicators which are obtained from the National Socio-Economic Surveys by Statistics Indonesia. Simulation results are in line with previous studies, where the multivariate Fay-Herriot model with p variable has smaller of mean squares error (MSE) than the univariate model. The model implementation to estimate CrudeParticipation Rate (APK), School Participation Rate (APS), and Pure Participation Rate (APM) also shows that the multivariate model produces smaller RRMSE than the direct estimates. It can be concluded that multivariate model is able to produce more efficient estimates than direct estimation and univariate model.
Key concepts: Univariate, Multivariate statistics, Statistics, Multivariate analysis, Estimation, Sample (material), Econometrics, Mathematics