Avi Singh is a Senior Fellow in the Center for Excellence in Survey Research at NORC and has more than 35 years of postdoctoral research experience. His research interests include survey design, estimating functions, small-area estimation, nonlinear mixed and latent variable modeling, data analysis corresponding to complex surveys, categorical data and time series, and protection of confidentiality and analytic utility in data dissemination.
Singh is an innovator who conducts research on the development of advanced statistical products and new applications of existing methods to meet client and project needs. At NORC, he is currently leading efforts to create Public Use Files for Centers for Medicare & Medicaid (CMS) that will allow researchers to access data from all CMS claims datasets. For this project, Singh made innovative modifications of GenMASSC, an advanced disclosure technique developed earlier by him. He is also developing a new system within the NORC Data Enclave to provide automated disclosure treatment of analysis output generated by remote users of the confidential microdata.
Singh previously worked at Cornell University, Memorial University, Statistics Canada, and RTI International. His experience in academia, government, and the private sector has honed his ability to combine fundamental ideas from different fields of statistics, in particular, survey sampling and mainstream statistics. At RTI he was the lead developer of the methodology of Generalized Exponential Modeling for sampling weight calibration, which is now in common practice in all of their surveys and is part of the SUDAAN software. He also developed MASSC (Micro-agglomeration with Substitution, Subsampling, and Calibration), a product for confidentiality protection of microdata, which resulted in a proprietary product as well as a patent award. At Statistics Canada, his new method of regression composite estimation for combining or blending information over time has been implemented since 2000 to produce more efficient monthly estimates of unemployment and employment.