Celia Sianipar, Adji Achmad Rinaldo Fernandes, Solimun Solimun, Septi Nafisa Ulluya Zahra

Bayesian Path Modeling Simulation for Stunting Analysis

Introduction

Bayesian path modeling simulation for stunting analysis. Assess Bayesian path modeling for stunting analysis using Monte Carlo simulation. Explore sample size, model complexity, and error variance effects, providing insights for public health causal analysis.

Abstract

This research assesses the interaction between sample size, model complexity, error variance, and their collective impact on the execution of Bayesian path modeling within the context of stunting, using Monte Carlo simulation over 100 datasets over 18 different conditions. The results indicate that the sample size positively impacts the stability of the model, but the more intricate the model and the more noise there is in the data, the greater the effect. A 0.3 error variance unpredictedly lowered the RMSE on various complex scenarios (eg. for nonlinear structures, it decreased from 0.222 to 0.079), and low error variance, combined with nonlinear pathways, led to lower CI coverage (<0.85) and lower ESS, indicating there was difficulty recovering the true parameters A unique contribution of the research is that, in Bayesian modeling, sample size is not the only driver of model complexity, noise, and stunting research. This information is beneficial in providing evidence on how to choose priors, structure models, and conclusively derive results for Bayesian causal analysis in the field of public health.

Full Text

Comments

You need to be logged in to post a comment.