2 ms·
The problem with trying to teach statistics, to anyone really, is that once you get past the very basics, you have to stop every time the lecturer/textbook invo
by eli_gottlieb 10y ago
The problem with trying to teach statistics, to anyone really, is that once you get past the very basics, you have to stop every time the lecturer/textbook invokes measure theory and figure out how to explain the concepts without just making everyone sit through two to three semesters of real analysis.
- hilop 10y agoYou just use discrete scenarios, not continuous. Same concepts, easier math. same way you can do linear geometry without calculus.
- eli_gottlieb 10y agoIn my experience, just knowing discrete probability and statistics is profoundly unuseful, especially when you walk out of class thinking you know the subject, sign up for a next course that involves real-valued quantities, and bomb everything.
- thearn4 10y agoAgreed, there is definitely a huge gap between doing, for example, a t-test and deriving the t-test (or giving any substantial justification for its form). You need a class in analysis, a class in measure-theoretic probability, then a mathematical statistics class to bring you back to the topic. It's a tough square to circle. Most who teach stats at the undergrad or hs level have little choice but to defer a lot of things to a later course.