Statistical techniques are used across scientific disciplines, so if we want to understand how science works and whether it is rationally justified, we need to understand statistics. My hope is that thinking explicitly about the foundational issues from a philosophical perspective will help us to do scientific and political work -- that our lives will be improved as a result of philosophical reflection.
Semesters offered: Spring 2025
Course developer: Jonathan Livengood, Associate Professor, Philosophy; teaching in LAS for 15 years
How would you describe the course to someone unfamiliar with the subject?
We talk about some interpretive issues (such as how to understand probability) and some logical issues (such as whether, when, and how statistical reasoning is rational and/or reliable) and some ethical issues (such as whether, when, and how the use of statistics is morally permissible). We connect some foundational disputes in statistics to analogous disputes in philosophy in order to learn something about both.
What made you want to create this course?
I have been thinking about scientific method since I was an undergraduate. Puzzlement about scientific method attracted me to philosophy in the first place. Statistics is the current common language of scientific method. Statistical techniques are used across the whole range of scientific disciplines, from fundamental physics to economics and political science. So, if we want to understand how science works and whether it is rationally justified, we need to understand statistics. Moreover, statistics get used and abused by governments and other policy makers. So, a substantial part of our lives is driven by statistics, even if we're not aware of it. My hope is that thinking explicitly about the foundational issues from a philosophical perspective will help us to do scientific and political work -- that our lives will be improved as a result of philosophical reflection.
Were there any challenges you faced while designing or teaching the course? How did you overcome it?
The main challenge is finding the right "level" at which to pitch the material. Some students are technically sophisticated. Some students aren't. Statistics is a mathematical discipline. So, we need to have some common ground with respect to the technical details. We can't really do philosophy of statistics without knowing something about the machinery of probability and statistical models. The difficulty is finding a way to adequately introduce statistical ideas without boring the technical sophisticates or losing (or misleading) the uninitiated. I didn't feel like this part of the course went as well as I would like the first time around, so I have ideas about how to do it better next time.
Now that you've offered the course at least once, what do you hope students took away from it the most?
The first time teaching, I had two thematic ideas that I'm hoping the students either internalized or explicitly rejected after thinking hard. The first has to do with the most abstract problems that statistics is trying to solve. I argue that statistics is a branch of logic. It's a formal logic of induction: statistics is telling us how to test our theories and how to measure the degree to which our theories are confirmed or disconfirmed by our evidence. The second has to do with uncertainty. Some statisticians have said that statistics is the science of uncertainty. The view I like is an eclectic statistical pragmatism similar to the view defended by Robert Kass, according to which statistics is about building and using mathematical models in the confirmation phase of inquiry. In that way, we can find common ground between historically opposed philosophies of statistics.
View more new course spotlights from LAS Educational Excellence.