I'm afraid the job market for stats teachers is not great - most university/college departments I know of don't hire external instructors to teach stats. Instead they're often taught by lecturers/asst professors whose research specialties also involve some level of statistical sophistication (that is, "ability to teach stats" as a secondary selection criterion rather than primary). In the US system especially there also tends to be a relatively large pool of advanced PhD students who are much cheaper to employ than tenure track faculty, but who also have the requisite skills to teach stats. Teaching fellows and adjunt faculty (ie people hired just to teach) tend to be hired from within.
Teaching at community college level is a possibility, and strangely enough the pay tends to be much higher than at the university level (shockingly low, if you didn't already know). But this comes at the cost of a very high workload, and specifically related to your area of interest, you are likely to come across students who lack even the most basic numeracy skills - multiplication and division.
But then it comes to the big question - can one actually teach Bayesian statistics at the undergraduate level? This depends on the field. In psychology, for example, I think the answer at the moment is a definitive "no". A telling example is John Kruschke at Indiana - one of the main proponents of Bayesian statistics. Here you can see his open letter to the field calling for an end to frequentist approaches. Yet in his introductory undergraduate stats course he explicitly avoids Bayesian statisics (see his course notes here ) and even at the graduate level he sticks mainly to frequentist approaches, albeit with a healthy dose "take the next course which explains how to do things right). Although it's clear that the transition is coming in this field (for example the message being delivered by EJ Wagenmakers and many others) it has not yet trickled down to undergraduates - who need to know frequentist statistics, if for no other reason than to understand this otherwise mystical jargon that appears in all the research articles they are likely to read in their undergraduate careers.
Sorry, I've started to feel like I'm rambling ... research methods instruction is one of those things I am passionate about, and your post compelled me to register and post for the first time after years of lurking.
I did send in an application to the Center for Modern Rationality but I haven't heard back
Please email me (elcenia@gmail.com) and tell me which types of work you wanted. There has been a spreadsheet-tracking issue, and I'm not sure who I have and haven't reached yet. (I'm considering just mass-mailing everybody on the list with "if you haven't heard from me before, please let me know and I can give you sample work". Thoughts on whether this would be more obnoxious than helpful?)
(I'm considering just mass-mailing everybody on the list with "if you haven't heard from me before, please let me know and I can give you sample work". Thoughts on whether this would be more obnoxious than helpful?)
Seems like from the number of people which seem to be still waiting for replies, much more helpful than obnoxious.
Email sent!
I think people would appreciate knowing that they might still have a shot even if they haven't heard anything. You could maybe ask Eliezer to put a note that you're still wading through applications in his next HPMOR author's note. =P Otherwise, I think a mass email would not be too annoying to those who have already heard from you and very much appreciated by those who haven't.
All of the really smart professors I know personally who have an opinion on the topic are Bayesians, Less Wrong as a community prefers Bayesianism, and I prefer it.
Is this how you would respond to a hypothetical student of yours asking why you're not a frequentist? Sorry for the snark, but you are considering a career choice where you would be a voice for Bayesianism to your students and fellow faculty. I think you owe it to those who want Bayesian methods to be treated seriously to acquaint yourself with de Finetti's theorem, Cox's theorem, the frequentist consistency (or lack thereof) of Bayesian procedures, etc.
Also, I'm pretty sure this is not the best way you could get paid to raise the sanity or rationality waterline, given your background. Think about the marginal effect of taking such a job. Best case, you'll either teach a Bayesian stats courses that otherwise wouldn't be taught due to lack of faculty or replace someone who would otherwise teach it (that person will probably be have better or equal academic credentials and might be less enthusiastic about Bayesianism). But students who would choose to take a Bayesian stats course will probably do fine anyways. Teaching intro (frequentist) stats well or doing private tutoring for students stuck in bad intro stats courses will be more likely to make a difference.
I do need to read up on those; Jaynes talks about the implications of Cox's theorem but doesn't go into it directly, so I'm only vaguely familiar. Thank you for the reading suggestions. I did plan to talk about those issues in the introduction of the course. Bolstad has an intro section justifying the Bayesian perspective, as well.
I think I picked that particular set of justifications because educators in general don't care about mathematical proofs, they care about what will be useful for the students to know how to do; in biology, the point of knowing statistics is to be able to read and write scientific papers, and the vast majority of papers are written using frequentist statistics. Proofs will not convince them; the fact that top professors are using Bayesian methods might.
My expectation was that I would replace a mediocre frequentist statistics lecturer with an excellent bayesian statistics lecturer within the same class. The class that I TA is taught by multiple professors, and at least one of them teaches from a Bayesian perspective. Professors have ridiculous academic freedom; one professor covers only basic t-tests, while another professor covers everything from linear regressions to the KS test to chi-squared to two-way ANOVA, and it's still the same course listing. So long as the students aren't complaining about failing, the university does not care. The students can try to sign up for a different professor, and will do so if they hear another prof is easier, but they still have to take the class, so even harder professors still have full sections (especially if their version has a reputation for being very useful/educational).
So, assuming that I would be hired to teach statistics and could choose to teach either frequentist or Bayesian, I see very little point in teaching frequentist. I could also reach vastly more students lecturing than I could via tutoring, probably 80ish vs 10ish.
I think the students that are interested in learning Bayesian stats should have the option available; I think there are probably a fair number of students who are smart, savvy, and motivated enough to sign up for a stronger stats course but aren't quite good enough to teach it to themselves.
I think I would almost rather not teach statistics than teach straight frequentist. I am really sick of teaching kids stuff that I know is suboptimal. I mean, I could do a good job of it, but raising the waterline isn't worth being miserable.
Bayesian statistics was a third year course at the universities I've attended. Keep in mind that Bayesian statistics uses the probability theory that is given in most probability and statistics modules, and requires distributional knowledge, something also taught by frequentist statistics. First year students will not have the knowledge of probability distributions or algebra to be able to find posteriors. And, of course, if you want to apply Bayesian statistics in practice you need some kind of programming skill, as most programs that do Bayesian analysis are not user friendly (a plague on winbugs house..).
Heres what Bayesian statistics courses tend to teach (in my experience) -the basis of bayesian statistics (usually in light amounts) -priors, and obtaining posteriors from them. Usually applied to simple examples with conjugate priors, a tough exam question might be to obtain the posterior of the mean mu and variance sigma of a normal distribution with normal prior and inverse gamma prior. -Jeffrey's prior (so they need to know what an information matrix is) -an introduction to MCMC techniques, particularly gibbs sampling and random walk metropolis hastings.
I'd suggest going into research first, to improve your credentials and also to gain a solid understanding of statistics. If you don't like biology, business schools use a lot of Bayesian techniques.
I haven't read Bolstad, but I think that grokking the first 3-ish chapters of Jaynes seems like a reasonable requirement for being able to explain why probability looks the way it does. The actual "methods" chapters after that aren't really useful/necessart for undergrads, except for the simple stuff like "given some data, what's the likelihood ratio for your hypothesis? How about the null hypothesis?" You could always jump into teaching without re-reading that, but iunno.
Yeah, the early stuff in Jaynes is pretty comprehensible (the ideas are clear if not all the proofs). Intro stats classes tend to be very light on the proofs, though. They're very much "here's probability", not "here's why probability". I'll definitely reread Jaynes again before teaching, but I want to finish Bolstad and work through some of the problems before that.
I don't know about helping you find a job, but as a biology grad student, I wish there were someone teaching a Bayesian course at my university.
What about creating a web application or web-based tutorial that teaches Bayesian statistics? Something like
http://eloquentjavascript.net/contents.html (check out the interactive examples)
or the recent Stanford online classes. (Probably not something you could easily get paid to do though...)
I am considering trying to get a job teaching statistics from a Bayesian perspective at the university or community college level, and I figured I should get some advice, both on whether or not that's a good idea and how to go about it.
Some background on myself: I just got my Masters in computational biology, to go along with a double Bachelors in Computer Science and Cell/Molecular Biology. I was in a PhD program but between enjoying teaching more than research and grad school making me unhappy, I decided to get the Masters instead. I've accumulated a bunch of experience as a teaching assistant (about six semesters) and I'm currently working as a Teaching Specialist (which is a fancy title for a full time TA). I'm now in my fourth semester of TAing biostatistics, which is pretty much just introductory statistics with biology examples. However, it's taught from a frequentist perspective.
I like learning, optimizing, teaching, and doing a good job of things I see people doing badly. I also seem to do dramatically better in highly structured environments. So, I've been thinking about trying to find a lecturer position teaching statistics from a Bayesian perspective. All of the really smart professors I know personally who have an opinion on the topic are Bayesians, Less Wrong as a community prefers Bayesianism, and I prefer it. This seems like a good way to get paid to do something I would enjoy and raise the rationality waterline while I'm at it.
So, the first question is whether this is the most efficient way to get paid to promote rationality. I did send in an application to the Center for Modern Rationality but I haven't heard back, so I'm guessing that isn't an option. Teaching Bayesian statistics seems like the second best bet, but there are probably other options I haven't thought of. I could teach biology or programming classes, but I think those would be less optimal uses of my skills.
Next, is this even a viable option for me, given my qualifications? I haven't taken any education classes to speak of (the class on how to be a TA might count but it was a joke). My job searches suggest that community colleges do hire people with Masters to teach, but universities mostly do not. I don't know what it takes to actually get hired in the current economic climate.
I'm also trying to figure out if this is the best career option given my skillset (or at least estimate the opportunity cost in terms of ease of finding jobs and compensation). I have a number of other potential options available: I could try to find a research position in bioinformatics or computational biology, or look for programming positions. Bioinformatics really makes "analyzing sequence data" and that's something I've barely touched since undergrad; my thesis used existing gene alignments. I could probably brush up and learn the current tools if I wanted, but I have hardly any experience in that area. Computational biology might be a better bet, but it's a ridiculously varied field and so far I have not much enjoyed doing research.
I could probably look for programming jobs, but they would mostly not leverage my biology skills; while I am a very good programmer for a biologist, and a very good biologist for a programmer, I'm not amazing at either. I can actually program: my thesis project involved lots of Ruby scripts to generate and manipulate data prior to statistical analysis, and I've also written things like a flocking implementation and a simple vector graphics drawing program. Everything I've written has been just enough to do what I needed it to do. I did not teach myself to program in general, but I did teach myself Ruby, if that helps estimate my level of programming talent. Yudkowsky did just point out that programming potentially pays REALLY well, possibly better than any of my other career options, but that may be limited to very high talent and/or very experienced programmers.
Assuming it is a good idea for me to try to teach statistics, and assuming I have a reasonable shot at finding such a job, is it realistic to try to teach statistics from a Bayesian perspective to undergrads? Frequentist approaches are still pretty common, so the class would almost certainly have to cover them as well, which means there's a LOT of material to cover. Bayesian methods generally involve some amount of calculus, although I have found an introductory textbook which uses minimal calculus. That might be a bit much to cram into a single semester, especially depending on the quality of the students (physics majors can probably handle a lot more than community college Communications majors).
Speaking of books, what books would be good to teach from, and what books should I read to have enough background? I attempted Jaynes' Probability Theory: The Logic of Science but it was a bit too high level for me to fully understand. I have been working my way through Bolstad's Introduction to Bayesian Statistics which is what I would probably teach the course from. Are there any topics that Less Wrong thinks would be essential to cover in an introductory Bayesian statistics course?
Thanks in advance for all advice and suggestions!