Sage Statisticians in Social Sciences: Impact of Rubin’s Work

Contemporary social scientists have cast serious doubts over traditional statistical testing procedures and questioned the reproducibility of the findings. The Bayesian approach provides sound statistical tools to draw inferences about unknown parameters and potential outcomes in a methodical way. This paper reviews D. B. Rubin’s work from the orthodox Bayesian viewpoint and discusses how his brilliant ideas and suggestions should be applied when social scientists deal with real data. The discussion focuses on making inferences about causal relationships and handling missing data. It is argued that social scientists who are confident about both the Bayesian coherent system and the necessitated effective software for numerical solutions can build relevant statistical models and derive relevant information from the Bayesian analysis of real data. This paper specifically explains how to deal with the data, using examples from situations that social scientists should often encounter.