In the third EMOS webinar on 18 Januray 2023 on the topic of "Non-probability sampling and big data, Ralf Münnich gave insights into:
Statistical inference and quality
Compensation methods (for biases)
Web surveys and big data
Conclusion and outlook
Ralf Münnich is full professor in economic and social statistics at Trier University. He has been participating in many European and national research projects, such as DACSEIS, AMELI, InGRID, as projects on the statistical methodologies of the German census. His main research interesests are survey statistics, computational survey statistics, us of modern data, and microsimulation methods. Since 2020, he is chairman of the German Statistical Society.
The aim of the webinar is to give an overview of problems using non-probability data such as web-surveys and big data and methods to overcome possible biases by using these data. Further, a basic understanding of ignoring the problems of using these data without considering how they were gathered will be built.
Webinar learning outcomes:
The basic strategies of compensating for biases induced by using non-probability are taught. These include the combination of probability and non-probability data via model-based methods such as imputation or matching or weighting.
Prerequisites: Elementary courses in statistics
Lenau, S., Marchetti, S., Münnich, R., Pratesi, M., Salvati, N., Shlomo, N., Schirripa Spagnolo, F. and Zhang, L.-C. (2021). Methods for sampling and inference with non-probability samples (Deliverable D11.8), Leuven, InGRID-2 project 730998 – H2020
Lenau, S.: Statistical and Machine Learning Methods for Handling Selectivity in Non-Probability Samples (2023). PhD Dissertation, Trier University.