Covariate Selection for Generalizing Experimental Results
Researchers are often interested in generalizing the average treatment effect (ATE) estimated in a randomized experiment to non-experimental target populations. Researchers can estimate the population ATE without bias if they adjust for a set of variables affecting both selection into the experiment and treatment heterogeneity.Although this separating set has simple mathematical representation, it is often unclear how to select this set in applied contexts. In this paper, we propose a data-driven method to estimate a separating set. Our approach has two advantages. First, our algorithm relies only on the experimental data. As long as researchers can collect a rich set of covariates on experimental samples, the proposed method can inform which variables they should adjust for. Second, we can incorporate researcher-specific data constraints. When researchers know certain variables are unmeasurable in the target population, our method can select a separating set subject to such constraints, if one is feasible. We validate our proposed method using simulations, including naturalistic simulations based on real-world data.
Co-Sponsored with The Center for Social Statistics
FP3.0: Increasing the uptake of Long-Acting Reversible Contraceptives (LARCs) among adolescent females and young women in Cameroon
In sub-Saharan Africa, 25% of teenagers have started childbearing (ICF, 2015). While young women describe many of these births as planned and intentional, women under the age of 20 also have the greatest percentage of mistimed/unintended pregnancies compared to all other age groups. For example, in Cameroon, more than 30% of the births to this group were unwanted or wanted later (DHS 2011). Low age at first birth has a significant impact on the spacing of births and timing of future pregnancies. It may also reduce accumulation of human capital for both the mother and the child.
Co-sponsored with the UCLA Luskin Senior Fellows Speaker Series
Social Science in the Age of Genomics
The cost of genetic information has been dropping at a rate faster than that of Moore's law in microcomputing. As a result, the science of genetic prediction has improved by leaps and bounds in recent years, and with it has emerged a novel field: sociogenomics. Sociogenomics seeks to integrate genetic and environmental information to obtain a more robust, complete picture of the causes of human behavior. This talk will highlight some recent examples of sociogenomic research, touching upon issues such as adolescent peer effects, racial discrimination, assortative mating, and fertility patterns. The talk will conclude by discussing the social and policy implications of genetic prediction.
Co-sponsored with the Public Policy and Applied Social Science Seminar Series
Fabíola Bof de Andrade, Fundação Oswaldo Cruz, Instituto de Pesquisas René Rachou, Brazil & James Macinko, UCLA This seminar will provide an introduction to the newest study in the Health and Retirement Study (HRS) family, the Brazilian Longitudinal Study of Aging. The speakers will describe the overall study design and the main topics covered, highlight […]
