Avi Singh is an Institute Fellow at AIR. His primary responsibilities include providing leadership, consultation, and collaboration with researchers across AIR and mentoring of junior researchers. He has a wide area of interest with expertise in design and analysis of spatial and temporal surveys, combining probability and nonprobability surveys, small area and latent variable modeling, and protecting confidentiality and quality of data.
Prior to joining AIR, Dr. Singh was associated with NORC, RTI International, Statistics Canada, and Cornell and Memorial Universities. At Statistics Canada, his novel method of modified regression for combining data from rotating panel surveys has been in use since 2000 for monthly production of unemployment rates from the Labor Force Survey. 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 RTI surveys and is part of the SUDAAN software. He also developed a product termed MASSC for confidentiality which resulted in a proprietary product and the award of a patent. At NORC, he developed a new method of model-over-design integration for efficient estimation from nonprobability supplements to probability samples.
Dr. Singh is an innovator who takes keen interest in conducting research on the development of advanced statistical tools and products for difficult but practical problems. At AIR, he is testing a new method of randomly split zones for samples of size one for dealing with the commonly prevalent problem of high nonresponse in surveys.