I struggled with understanding how to use Bayesian statistics in my professional work. More specifically, I am struggling understanding the building blocks of Bayesian statistics. What I do see is that Bayesian statistics involves probabilistic thinking, and very clever sampling of data distribution. On the higher level conceptualization, the ideas of prior belief, and updating your belief to get a posterior estimation are intuitively appealing. However, when it comes to mechanics of applying Bayesian logics, I am struggling a lot.
This struggle does not deter me from enjoying reading about Bayesian logics being applied in real life situations. Recently I listened to a Data Skeptic podcast episode about data representations, data visualization, and how we citizens or audience are unconsciously being educated about statistics through reading and interacting with very well-thought out work of data journalists and data scientists at news organizations. The specific example that professor Jessica Hullman mentioned is an example from the New York Times’ interactive graphic representations of inequality in America. The interactive exercise asks the reader to provide their prior belief about inequality in America, and then presenting them with data, thus nudging them toward updating their belief. This is such a brilliant statistics exercise from a very well-respected news organization in the US. I wish that more news organizations in the world exercise what the New York Times does: using data journalism to educate the public about social issues in a scientific way.
For those who would want to listen to the podcast episode, it could be found here. Let me know what you think about Bayesian statistics, and how one can implement Bayesian statistics in social sciences.