The Department of Statistics, Faculty of Science and Mathematics (FSM), Diponegoro University (UNDIP) held a virtual public lecture through the World Class University (WCU) Visiting Professor Online program. This activity was chaired by Alan Prahutama Ph.D, a lecturer in the Department of Statistics UNDIP.
The speaker at this activity was Dr. Zita C Villa Juan Albacea, professor at the Institute of Statistics (INSTAT), Collage of Arts and Sciences (CAS), University of the Philippines, Los Banos (UPLB), Laguna, Philippines. Dr. Zita is an expert in the field of Small Area Estimation (SAE). She has been involved in several SAE projects by the Philippines Statistics Authority (PSA), Asian Development Bank (ADB), and the United Nation of Women.
Dr. Zita gave a presentation topic entitled “Meeting the Sustainable Development Goals (SDGs) with SAE”. Before the presentation by Dr. Zita, the Vice Dean of FSM for Resources, Dr. Eng. Adi Wibowo, gave a speech at this activity. He said that the SDGs are a field that can be explored by statistical scientists to be used as a basis for decision-making. He also hopes for international collaboration, especially in the field of research that can be used to improve development, especially related to the SDGs.
In the initial presentation, Dr. Zita explained about the 17 SDGs and the components of each SDGs group. Statistics are used by the government to support decision-making or policy.
One of the data that can be used as a guideline in measuring the SDGs is survey data. The problem is that survey data is used for estimation at the district or provincial level (aggregate level). If the survey data is used for lower level estimation, it will result in greater variance. To overcome this, it can be done in various ways, therefore one of the techniques to overcome this is with SAE. SAE is a statistical technique used to estimate population parameters for small geographical areas or sub-populations (such as sub-districts and sub-districts) that have a very small sample size.
Furthermore, Dr. Zita explained the steps in SAE modeling including (1) identification of the objectives of SAE modeling; (2) Identify the indicators needed to form the model; (3) Identify the desired level of aggregation of the distribution; (4) Determine the dataset to be used; (5) Identify the computing software to be used; (6) Identify the approach to be used; (7) SAE modeling estimation; (8) Evaluation of the quality of the estimated results; (9) To make a detailed presentation of the modeling estimation results.
The SAE method can be approached with two approaches, namely direct estimation and indirect estimation. Direct estimate uses sample data from the small area. This results in low accuracy because the variance value is too large. Meanwhile, indirect estimation uses a model-based methods approach. This approach method is the most common in SAE modeling. These approaches include the Fay-Herriot model, the Hierarchical Bayes model, and Empirical Best Linear Unbiased Prediction (EBLUP). SAE provides better accuracy for small area modeling.