The effectiveness of small area estimation with best linear unbiased prediction method on various sample sizes with resampling simulation on SUSENAS data in the case of mean years school of Kecamatan (sub-district) in Surabaya

SDGs aim to improve the quality of life from one generation to the next. One of the SDGs goals is quality education. One of the indicators that make up this goal is Mean Years School (MYS) In Indonesia, MYS is generated from Susenas. Susenas methodology is designed for the purposes of estimating macro indicators up to the district/city level, so the data from the Susenas cannot be used directly to estimate smaller areas/domains. In this study, MYS at the sub-district level will be estimated using SAE. SAE is a method that is often used to produce better precision in estimating a parameter in a small area, in this case a sub-district. SAE method used in this study is EBLUP. In SAE with the EBLUP based on the area level and using accompanying variables, it is able to provide a better estimation value than the direct estimation results. From the available samples, the adequacy of SAE precision can be determined by performing bootstrap resampling simulations at different sample sizes. Furthermore, an evaluation of the simulation has been carried out by looking at the goodness of the model and the comparison of the estimation results for each sample size. It found that even though the sample size was smaller than the Susenas sample, it could produce an estimated value that was almost the same as the entire Susenas sample size, besides that the R-Square value of the overall sample size also remained in the range of 40%.