Mathematica worked with FinMark Trust’s (FMT) research facility Insight2Impact (i2i) to develop a procedure to eliminate bias in short message service (SMS) survey data using a statistical technique called multilevel regression with post stratification (MRP).
- Causal inference
- Bayesian statistics
- Methods for modeling correlated and longitudinal data
- Machine learning
- Analysis of survey data
- Semiparametric regression
- State Health Policy
- Human Services
- Family Support
- Child Welfare
- International Research
Jonathan Gellar is an expert in developing and applying Bayesian and frequentist statistical methods to address policy questions. He has led research projects and statistical tasks across a wide range of policy applications in domestic and international settings, including Medicare and Medicaid policy evaluation, technical assistance for Medicare and Medicaid programs, education, child welfare, military mental health, and financial inclusion.
At Mathematica, Gellar primarily focuses on health care policy evaluation. He led the quantitative analysis for Mathematica’s evaluation of managed long-term services and supports, part of the Medicaid 1115 demonstration. In this role, he led all statistical aspects of the evaluation, including study design, propensity score matching, and data analysis using Bayesian and frequentist methods. He has led statistical tasks on several other Medicare or Medicaid evaluations, including the Comprehensive Primary Care and Comprehensive Primary Care Plus initiatives, the Medicaid Money Follows the Person evaluation, and the Independence at Home demonstration. He is the co-principal investigator and lead statistician for a project developing an algorithm to predict near-term academic outcomes among students in Allegheny County, Pennsylvania. He is also the deputy project director for a project investigating the use of SMS messages as a platform for surveying hard-to-reach populations in Africa and Asia.
In 2015, Gellar joined Mathematica after completing his Ph.D. in biostatistics at the Johns Hopkins Bloomberg School of Public Health, where he developed methods at the intersection of functional data analysis, survival analysis, and the analysis of longitudinal data. In addition, he conducted several analyses with data collected in clinical settings, including clinical trials and cohort studies.