Areas of Expertise

Causal Inference

Many social science questions concern the causes that underlie social behavior and social structure. In the social sciences causal inference is usually complex. Our researchers develop novel identification schemes in order to study complex social phenomena, such as civil conflict, polling, college attendance, and job loss. Researchers at CSS study machine learning approaches such as kernel methods and genetic algorithms to achieve covariate balance in pre-treatment covariates, facilitating the analysis of causal effects. They also study transportability, that is, combining observational and experimental studies in order to estimate population treatment effects, and causal effect heterogeneity, that is, how the effects of causes systematically vary across populations of interest. In social epidemiology, we study, g-methods for causal modeling, causal mediation with interaction analyses, and bias analysis.

Demographic Methods

Since their conception, surveys have a fundamental role in providing detailed information on social processes. However, survey sampling methodology currently faces many challenges. One is the study of hard-to-reach or otherwise “hidden” populations. These populations are characterized by the difficulty in sampling from them using standard probability methods. Typically, a sampling frame for the target population is not available, and its members are rare or stigmatized in the larger population so that it is prohibitively expensive to contact them through the available frames. Affiliates are developing statistical methodology to help improve understanding of such populations. We also study methods that incorporate survey information with external data registries, methods to adjust survey estimates for non-response, and methods for sensitive questions. Our researchers apply survey methodology to study a population’s political attitudes, opinions on civil conflict, and projecting the population of individuals with Alzheimer’s disease.

Social Data Science

The digital trails that institutions and individuals produce as they conduct their daily lives (i.e., “big data”) shed new light on demographic and social behaviors. Data Science is the science of examining a large amount of data with the purpose of finding patterns and drawing inferences by algorithmic or mechanical processes to derive insights. By linking these new data with traditional demographic methods and data (e.g., censuses, surveys, vital records), CSS affiliates study key questions of social and policy relevance. Affiliates also have expertise in network methods, including stochastic models of networks that allow for complex investigation of detailed social processes. CSS is bringing together data science and social science researchers around campus to develop expertise in relevant statistical, programming, and machine learning skills to advance social scientific knowledge.