Soubhik is a research methodologist in the Methodology and Quantitative Social Sciences department at NORC at the University of Chicago. He specializes in the application of statistical and computational methods to the social sciences. Soubhik has over seven years of experience applying a diversity of data science tools including generalized linear models, machine learning, causal inference, Bayesian inference, and data visualization to substantive applications ranging from social media text analysis to government program evaluation to pre-election polling.

At NORC, Soubhik leads a wide range of data science tasks across different projects. On complex survey weighting, he designed, built, tested, and applied a suite of statistical methods in R to combine probability and non-probability surveys for the newly developed Rapid Survey System (RSS) from the National Center for Health Statistics (NCHS). Soubhik also created a reproducible pipeline to generate and test survey weights for Pew Research Center’s Religious Landscape Study III (RLS III), a mixed-mode study involving both dual-frame random digital dialing and address-based sampling. On data visualization, he built a dynamic data visualization dashboard to enable data collection and quality control for the 2023 round of the Survey of Doctorate Recipients (SDR). Additionally, Soubhik has built a dashboard to help diagnose weights for AP-NORC political polls. On causal inference and statistical modeling, Barari led an evaluation of mode effects in the SDR and is leading an evaluation of generative AI tools for survey interviews.

Before joining NORC, Soubhik was a research scientist at SurveyMonkey leading data science efforts on the polling team, from weighting weekly pre-election surveys to developing open-source R packages. He has also served in data science roles at Microsoft Research, Harvard’s Institute for Quantitative Social Science, and MIT’s Political Methodology Lab. Soubhik’s research has appeared in academic journals such as Nature Scientific Data and media outlets such as The Atlantic and Scientific American. Soubhik regularly gives talks on data science, social science, and public opinion at organizations such as the NY Open Statistical Programming Meetup, the R Gov Conference, Columbia University, Meta, AAPOR, and TEDx.