Skip to content

Latest commit

 

History

History
10 lines (5 loc) · 1.24 KB

unit_06_part_3.md

File metadata and controls

10 lines (5 loc) · 1.24 KB

Part 3: Confidence Intervals, Practical Significance and p-Hacking

In this section, you are going to learn about several concepts of hypothesis testing in practice.

First you will learn about confidence intervals, which are regions produced from a sample and test that overlap the (unknown) true population value with at some known rate. You'll read and watch a lecture about the concept, but there is a real risk of mis-interpreting these confidence intervals. This is where the careful language of statistics is at odds with the vernacular use of the term "confidence interval".

Second you will learn about providing context for the results of parameter estimates. We will provide a few strategies for providing this context, but none of these will always be correct. Understanding whether a result that is statistically significant is practically significant is very much shaped by context.

Third, and finally, we will talk about issues concerning running many tests, but reporting only a subset of those tests. This is a problem that is present in academic research (sometimes deemed the "replication crisis"), but it is also very much a problem that is present in statistical work that is directed toward business decision-making.