Date of Degree


Document Type


Degree Name



Public Health


Deborah Balk

Committee Members

Diana Romero

Heidi Jones

Shiro Horiuchi

Michele Kiely

Subject Categories

Asian Studies | International Public Health | Maternal and Child Health | Other Social and Behavioral Sciences | Social Statistics | Spatial Science | Women's Health


Indonesia, stillbirths, neonatal mortality, population-survey, spatial analysis, random effects


Little research has focused on the determinants of stillbirth and neonatal mortality in Indonesia despite the fact that the country has one of the highest rates of stillbirths in the world and one of the highest rates of neonatal mortality in Asia. Several areas have not been explored due to data limitations, such as shared risk factors and measurement challenges, which can inform health policy and interventions and guide future research. This dissertation addresses this gap by comparing measurement challenges for stillbirth and neonatal mortality and identifying risk factors for both using Indonesian reproductive health survey data. Differences in estimates of stillbirth and neonatal mortality rates are compared from cross-sectional and longitudinal surveys and each outcome is analyzed in light of observed individual and geospatial risk factors. The potential contribution of unobserved heterogeneity or frailty is also evaluated to adjust for model misspecification regarding clustering of births within mothers. The three papers provide a more thorough overview of stillbirth and neonatal mortality risks in Indonesia than is available in current published research. This study is also among the first to compare two household surveys in Indonesia to assess implications of survey data quality and design for national and sub-national rate estimates. The three main findings are: 1) birth history data focused on livebirths could significantly undercount stillbirths leading to underestimated stillbirth mortality rates, 2) unobserved heterogeneity is an important component of risk due to clustering of births by mother, and 3) geographic variables in current reproductive health surveys are not sufficient to determine whether mothers have appropriate access to health services. The findings imply that stillbirth is insufficiently measured through current methods, risk factor estimations for neonatal mortality rates should adjust for maternal clustering effects, and survey sampling frames should accommodate geographic diversity beyond urban-rural designations.