Seminario de Estadística
Miércoles 27 de abril de 2022 a las 11:30 h.
Departamento de Matemáticas, módulo 17, aula 520
Sonja Radas (Institute of Economics, Zagreb)
Título: What do they really think: Uncovering latent preferences through meta-predictions
Self-reporting is often used as a source of information in all walks of life, but respondents may not provide accurate self-reports. This is especially true in context involving private and sensitive topics such as health issues, unethical behavior, political affiliation, etc. Using game theory language, to self-report means to disclose whether a private signal has been received. In this paper we develop a simple theoretical model which describes how individuals use their personal signals, together with their initial beliefs and the information from their social circles, to form Bayesian predictions on the proportion of their peers who would declare their signal (i.e. meta-predictions). We show that, regardless of whether an individual chooses to declare the signal or not, there is an “imprint of the signal” which can be deduced from their meta-prediction. We conduct online experiments to demonstrate model validity.
[Talk based on joint work with Ivan Spajić Buturac]