Research Statement

How population processes differ across space and society

I examine how the unequal distribution of fertility, mortality, and migration shapes population patterns across spatial and social strata — and their intersections. The throughline across my dissertation and my forecasting work at IHME is methodological: the small-area estimation techniques I build to reveal hidden subnational inequities are the same techniques that make national and global population forecasts more accurate, and the forecasting work in turn shows me where finer-grained subnational estimates are most needed.

Dissertation: subnational estimation

3 chapters · survey harmonization + Bayesian small-area models

Three projects using data harmonization across survey sources and Bayesian hierarchical small-area models to capture unequal patterns of fertility, mortality, and migration within large, heterogeneous countries.

Chapter 1
Fertility and social transformation in India

Estimates fertility by state, education, and urban–rural residence in India, 1990–2020. Statistical methods addressing data sparsity uncover two previously masked dynamics: unequal fertility declines across the regional and education gradient, and a concentration of births among rural, low-educated women even as school enrollment has expanded. Kitagawa decomposition separates behavioral from compositional change to clarify how fertility will continue to unfold in a large heterogeneous country.
Chapter 2
Mortality inequality across spatial and societal intersections in the U.S.

Estimates life expectancy by state, educational attainment, and race-ethnicity — dimensions rarely considered jointly in prior work. The approach jointly accounts for geography, education, and race-ethnicity while estimating misclassification in death-certificate data. Findings show persistent young-adult mortality gaps by education, concentrated among non-Hispanic White and Black populations with less than a high school education — a key driver of recent U.S. life expectancy stagnation.
Chapter 3
Racialized and gendered patterns of internal mobility in the U.S.

Estimates sex-, race-ethnicity-, and age-specific migration flows across U.S. states by harmonizing multiple data sources within Bayesian hierarchical regression models. Findings reveal persistent racial and gender disparities in migration opportunity, reflecting labor markets, social networks, and institutional barriers on who gets to move.

Forecasting at IHME

Global, national & subnational

Beyond the dissertation, I've led fertility and migration forecasting for the Future Health Scenarios team — work that draws directly on subnational estimation methods, and that in turn reveals where those methods matter most.

Fertility forecasting

Rebuilding the covariate structure

Reconstructed forecast models to add under-5 mortality and population density alongside existing covariates (education, contraceptive met need), reflecting the theoretical and empirical links between child mortality, density, and fertility decline — and improving how much of the fertility trend the model explains across time and regions.

Migration forecasting

From indirect to direct estimates

Led the reconstruction of the migration forecasting pipeline, shifting from indirect to direct estimates of past trends and integrating climate shocks as drivers of future flows — the same harmonization logic developed for the dissertation's migration chapter, applied at global scale.

Policy scenarios

Simulating pro-natalist policy impacts

Developed policy-relevant population scenarios, incorporating the rising importance of pro-natalist measures in low-fertility countries to simulate how policy interventions reshape age structure, dependency ratios, and spatial distribution.

Methodological throughline

Subnational rigor, global scale

Small-area estimation surfaces where national averages mask real heterogeneity; forecasting then asks how that heterogeneity evolves. Each project sharpens the other — better subnational models improve forecast covariates, and forecast residuals point back to where subnational estimation is weakest.

Research directions

Next
Quantifying the future burden of population aging
Next
Applying the amenability framework to cancer and Alzheimer's disease
Next
Closing the feedback loop: climate and population processes
Next
Disparities in migration as a climate adaptation strategy