Examining Data Absenteeism in Synthetic Health Data Across Six ASEAN Countries

This study examines data absenteeism in large language model (LLM)–generated synthetic health data across six ASEAN countries through a three-layer lens: (a) presence or absence, (b) associations, and (c) directionality and magnitude. By comparing synthetic datasets with cross-national survey data, the study evaluates whether LLMs can replicate demographic distributions and the relationships among social determinants, health technology use, and health outcomes. Results show that while synthetic data closely mirror basic demographic features, they often misrepresent behavioral relationships, producing weaker, exaggerated, or inconsistent associations. These findings suggest that data absenteeism extends beyond missing groups to include partial representations of behavioral patterns. Overall, the findings highlight both the promise and the limits of synthetic data as complementary tools for exploratory research and underscore the importance of relational validation to ensure responsible applications in health communication and equity studies.