You are here
Missing Unit Problem in Population Health (and Social Science) Research
Speakers: Rockli Kim, S.V. Subramanian
Date and Time: Thursday Mar 15th, 12:00pm - 1:30pm
Location: CGIS South 354
Abstract: An exclusive reliance on a single level – whether individual or geographic/ecological – as the unit of analysis is still pervasive in many fields of social sciences, including population health research. It is well recognized that patterns and relationships observed at any analytical unit is highly sensitive to the definition of the areal unit (i.e., “Modifiable Areal Unit Problem”). Yet, the fundamental premise in a single level framework, that there intrinsically exists one ideal unit of analysis and inference, is unrealistic given the complexity of the world we live in. While the advent of multilevel modeling provided opportunities to simultaneously model variation at multiple geographic scales, most studies in population health are restricted to two level of analysis at most. We present the thesis that treating observed differences in the results by choice of unit of analysis simply as a nuisance is conceptually problematic and can be empirically misleading (at worst) and in many instances provide an impoverished interpretation of the undertaken inquiry. The relative importance of one unit can be truly examined only when multiple scales that are thought to influence the outcome are simultaneously considered (i.e., "Missing Unit Problem"). In order to exemplify our thesis, we present our work concerning variously defined levels and a range of health outcomes (life expectancy, BMI, child anthropometric status and failures) in the context of the US and across low- and middle-income countries.
Rockli Kim is a postdoctoral research fellow at Harvard Center for Population and Development Studies. She received her ScD from the Department of Social and Behavioral Sciences and SM from the Department of Global Health and Population at the Harvard T.H. Chan School of Public Health. Her current research work focuses on investigating variations in individual and population health and well-being using multilevel statistical modeling. Additional research interests include assessing social determinants of child undernutrition and early childhood development in low- and middle-income countries.
S.V. Subramanian is a Professor of Population Health and Geography at Harvard University. He was the Founding Director of Graduate Studies for the interdisciplinary PhD program in Population Health Sciences at Harvard. He has published over 500 articles, book chapters, and books in the field of social and contextual determinants of health and nutritional inequalities in India and developing countries, and applied multilevel statistical models. He was named among Thomson Reuters Highly Cited Researchers (top 1% of cited publications in Web of Sciences) in 2015, 2016, and 2017. Subramanian is the Co-Editor-in-Chief for the international journal Social Science & Medicine (SSM), a Co-Senior Editor for the social epidemiology office of SSM, and the founding Co-Editor-in-Chief of a new journal SSM – Population Health.