IASS - Reducing measurement and sampling biases in non-probability surveys
| Date | 27 Feb 2026 |
| Time | 14:00 CET - 15:30 CET |
| Level of instruction | |
| Instructor |
Aditi Sen
|
| Registration fee | |
In the age of big data, non-probability surveys are becoming increasingly abundant. Data integration techniques involving both probability and non-probability surveys are being extensively used for improving estimates of finite population parameters. While much of the existing research has focused on mitigating selection bias in non-probability surveys, the issue of measurement error within these surveys remains relatively unexplored. Statistical methods devised with the purpose of reducing selection bias are appropriate for reliable estimation, only under the assumption of accuracy of survey responses. Motivated by a recent case study of the Pew Research Center, our research addresses bias from both measurement and sampling errors in non-probability surveys. In this article, we propose a new data integration method that uses multiple probability and non-probability surveys and leverages machine learning models to construct a composite estimator. The proposed composite estimator integrates probability and non-probability surveys, when both contain response variables of interest. We analyse the performance of this estimator in comparison to an existing composite estimator in literature, analytically as well as empirically, using multiple survey data. Finally, we identify conditions under which the proposed estimator outperforms estimators based solely on probability surveys.
Instructors
About the instructor
Aditi Sen is a PhD Candidate in Applied Statistics at the Department of Mathematics, University of Maryland College Park. She holds a master’s degree in Statistics from the University of Calcutta. Her research focuses on statistical data integration, empirical Bayes and hierarchical Bayes methods for granular estimation using complex surveys and Bayesian machine learning using neural networks. As a research assistant at the Joint Program in Survey Methodology, she has contributed to the Sampling Methodology Capacity Building Initiative for World Bank employees in Tanzania. She has worked with advisor Prof. Partha Lahiri on important small area estimation problems, demonstrating effectiveness of empirical Bayesian methodology on polling data for election projection, and Covid-19 survey data for mask effectiveness in the U.S. As a data science intern at iSpot.tv, she has industry experience in implementing predictive modeling techniques and statistical inference from complex survey data, including non-probability samples. She is the winner of the ASA Edward C. Bryant Scholarship for Outstanding Graduate Student in Survey Statistics and WSS Outstanding Graduate Student 2025. Prior to graduate study, she has extensive experience in data analytics, building machine learning models for spend categorization of credit cards, likelihood prediction of personal loans and other financial products in the Hongkong and Shanghai Banking Corporation.