Stunting Modeling Using Robust Regression: Maximum Likelihood, Method of Moment, and Generalized Maximum Likelihood Estimations Yuliana Susanti, Hasih Pratiwi, Respatiwulan, Sri Sulistijowati Handajani, Muhammad Bayu Nirwana, Ilmia Hamidah
Universitas Sebelas Maret, Surakarta, Indonesia
Abstract
Stunting is a serious health problem facing the world. Stunting results from chronic or recurrent undernutrition, usually associated with poverty, poor maternal health and nutrition, frequent illness, and inappropriate early feeding and care in early life. According to WHO, in 2019, the global prevalence of stunting reached 29.9%, and Asia was in second place with a percentage of 31.9%. In 2021, the stunting rate in Indonesia was 24.4%, which is above the WHO standard of 20%. Factors that are thought to affect stunting are iron supplement tablets, exclusive breastfeeding, proper sanitation facilities, and toddlers suffering from diarrhea. One method to analyze the factors that influence stunting is robust regression. Robust regression is a method used when there are some outlier data or, the data is not normally distributed. This research aims to develop a robust regression model for the number of stunting in Indonesia based on the influencing factors. Several estimation methods that can use in robust regression are maximum likelihood (M) estimation, method of moment (MM), and generalized M (GM) estimation. The results showed that the model with GM estimation was the best because it had the largest adjusted R squared and the smallest AIC. Partial tests showed that all independent variables significantly affect the number of stunting in Indonesia.
Keywords: stunting, robust regression, M estimation, MM estimation, GM estimation
