Bayesian Experimental Assessment of Social Technologies: Methodological and Methodical Aspects

Abstract

In modern sociology the Bayesian methodology for analyzing sociological data is practically not represented. The basics of applying the Bayesian approach are not taught at Russian sociological faculties, and empirical sociological research is not conducted within the framework of this approach. At the same time, this methodology has significant prospects within the context of experimental sociological research, in particular, when solving the problem of assessing the effectiveness of new social technologies and innovations. The main purpose of this study is demonstrating of the methodological and methodological aspects of the application of the Bayesian approach using modern statistical analysis software, as well as its advantages in comparison with the traditional frequentist approach in the field of experimental assessment of social technologies. The principles of operation of J.K. Kruschke's universal comparative Bayesian model BEST (Bayesian estimation supersedes the t test), reproduced with Python library PyMC, were revealed. The obtained results of the study not only prove the methodological superiority of the Bayesian approach when applied to the assessment of social technologies, but can also be used by other researchers to tune existing or build new comparative Bayesian models, contributing to the popularization of the Bayesian approach in social research.