Andrew — enjoyed your talk. There were a few questions from the online participants that weren’t read out during the Q+A, and I wondered if you would be willing to answer them here?
Most first courses base inference entirely on p values. What is the best way to remove misapprehensions about them? Should we even try in a first course?
What are your thoughts on including Bayesian concepts/methods in introductory teaching?
Can you say a little more about motivation – how might we measure whether we’re succeeding in motivating students?
How do you engage the non-data-confident people in handling data, before you even get to the statistics?
Any tips on teaching a class of students who are very reluctant to learn statistics?
Thanks for the questions! It was fun to give the talk, and it would’ve been even better if the online audience had been in the room.
1. Our best attempt to remove misapprehensions about p-values in textbooks is in Chapter 4 of Regression and Other Stories, so I’d recommend taking a look at that first.
2. For an introductory course, I’d recommend just avoiding the whole p-value thing and attacking problems directly, as is pretty much done by Llaudet and Imai in their book that I mentioned in my talk.
3. We do some Bayes in Regression and Other Stories. Right now I wouldn’t do Bayes in an intro course, but in the future I might change my view on that. Each year I find myself moving more and more toward the use of informative priors.
4. Motivation is super important. I have no way of measuring it, beyond informal observations of students. I guess it would make sense for us to do some surveys of students before and after the course to get some measure of their motivation levels.
5. I don’t have any suggestions for engaging the non-data-confident people in handling data. The main thing is that we do a lot of computing in the class and not much math. The math we do is of a different sort than typical statistics classes of that level: instead of fake-O derivations of the sampling distribution of the sample mean, we do things like moving from the formula of logistic regression to the graphed curve, or vice-versa.
6. I have no real tips on teaching a class of students who are very reluctant to learn statistics. Nowadays, statistics is so important that I’d think that just about everybody will want to learn it! If I were to offer a tip, I’d say that it would be good to connect to topics that are important to their work or studies.
Dude, I need more warning for a trip to London. Have a great time, and don’t forget to look to the right before stepping off the curb.
Andrew — enjoyed your talk. There were a few questions from the online participants that weren’t read out during the Q+A, and I wondered if you would be willing to answer them here?
Most first courses base inference entirely on p values. What is the best way to remove misapprehensions about them? Should we even try in a first course?
What are your thoughts on including Bayesian concepts/methods in introductory teaching?
Can you say a little more about motivation – how might we measure whether we’re succeeding in motivating students?
How do you engage the non-data-confident people in handling data, before you even get to the statistics?
Any tips on teaching a class of students who are very reluctant to learn statistics?
Robin:
Thanks for the questions! It was fun to give the talk, and it would’ve been even better if the online audience had been in the room.
1. Our best attempt to remove misapprehensions about p-values in textbooks is in Chapter 4 of Regression and Other Stories, so I’d recommend taking a look at that first.
2. For an introductory course, I’d recommend just avoiding the whole p-value thing and attacking problems directly, as is pretty much done by Llaudet and Imai in their book that I mentioned in my talk.
3. We do some Bayes in Regression and Other Stories. Right now I wouldn’t do Bayes in an intro course, but in the future I might change my view on that. Each year I find myself moving more and more toward the use of informative priors.
4. Motivation is super important. I have no way of measuring it, beyond informal observations of students. I guess it would make sense for us to do some surveys of students before and after the course to get some measure of their motivation levels.
5. I don’t have any suggestions for engaging the non-data-confident people in handling data. The main thing is that we do a lot of computing in the class and not much math. The math we do is of a different sort than typical statistics classes of that level: instead of fake-O derivations of the sampling distribution of the sample mean, we do things like moving from the formula of logistic regression to the graphed curve, or vice-versa.
6. I have no real tips on teaching a class of students who are very reluctant to learn statistics. Nowadays, statistics is so important that I’d think that just about everybody will want to learn it! If I were to offer a tip, I’d say that it would be good to connect to topics that are important to their work or studies.
Agree that more warning is needed! Do you know whether it will be uploaded to YouTube?