Connect with us


Slow down and THINK – Econlib



Slow down and THINK - Econlib

When you think of statistics, do you think of a useful tool for real-world analysis, or does the phrase “lies, damned lies and statistics” come to mind? Regardless of your answer to that question, Jeremy Weber wrote his new book: Public Policy Statistics, for you. In this episode, host Russ Roberts invites Weber to talk about it.

Weber argues that no statistics textbook includes the integration of context, purpose, and audience with statistical analysis. That is a problem. Roberts congratulates Weber on his use of illustrations instead of equations, and describes how he views statistics in college as more of a cooking class. Weber sees them more as vocational education; both are excellent analogies! What works best for you?

Of course, I have many more questions I could ask… As always, we’ll limit ourselves to a few questions, and we hope you’ll take a moment to share your thoughts. As Russ says, we’d love to hear from you!

1- How is learning statistics in school like a cooking class or learning to use a chainsaw? What’s wrong with these ways of thinking about statistics? What does Weber mean when he compares concepts that are concept-dependent with context-less? (Think of statistics versus physics, perhaps.)

2- Is statistical analysis being used more often as a weapon or for the search for truth in the political process? How do you think politicians would perceive Weber’s book, and why?

3- Roberts wonders to what extent you can look at data without taking theory into account. How does Weber describe the correct relationship between data and theory? How do today’s increased computing power and amount of data enable analytics? more difficult? How can analysis today be more superficial, while at the same time having more data behind it?

4- The (in)famous distinction between correlation and causation is brought up towards the end of the conversation. Roberts says one of the things he liked about Weber’s book is that he makes a point that goes much deeper than that. What’s that point? How should we do the size of causality, and how do analysts use statistical significance as a crutch?

5- Until now the focus has been on what is wrong with statistics and the way it is taught. What did you learn from this conversation about how it needs to be improved? Is an understanding of statistics a necessary part of citizenship education? Why or why not?