Gang Fang's Blog

A follow-up question from yesterday is how Bayesian statistics and statistical testing relate? They both seem to seek to make inferences about populations based on sample data.

My thought: before answering this question, I want a concept map that show me where hypothesis testing fits in the subject of statistics. I wonder if it is frequentist?

My answer: “seek to make inferences about populations based on sample data” is the broadest statement I can make regarding inferential statistics, and therefore anything falling into this category provides a mean to this goal.

Bayesian statistics uses probability to represent uncertainty about the world. The uncertainty is viewed as subjective belief of events. “The world” can be thought of as the population and the parameters are treated as random variables.

Hypothesis testing aims at establishing certain claims about population parameters or properties, based on sample data by ruling out randomness in the sample.

So we can see they use different mechanics for very similar purposes. One thing to note is Hypothesis testing is rooted in the Frequentists school of probability while it is still used by Bayesian.