Shopping for textbooks, I’ve stopped in at three bookstores on campus. Just for fun. No, really — I always like to stroll up and down the aisles and get a feel for what all the people in all the various colleges and their departments are assigning. Now and then I even pick up a neat-looking Machine Learning, Molecular Dynamics, or History of Science book. So sue me; I’m a pan-geek.
My wife was with me this last time. She pointed out, as she often does, something I missed entirely. There are, like, seventeen different “Introduction to Probability and Statistics” textbooks, scattered in the Biz School, Engineering, Medicine, Biology, Psychology, Social Sciences, Applied Math, and for all I know Art History shelves [I did see Guns, Germs and Steel as an assigned reading in one very interesting Art History class].
Many of these are, by implication, remedial Probability and Statistics — for graduate students, in many cases, but also for “advanced” undergraduates. And that is so wrong, it hurts. Practical, modern probability and statistics courses should be required at the entry level in all curricula. Period. I say quit futzing with calculus for anybody but mathematics majors, and get everybody tight with R.
Oh, and as I recall there was not one Probability and Stats books in the Elementary Education shelves.
I confess: I _should_ have redesigned the introductory probability and statistics class I’m teaching, so it uses R and not (may Heaven have mercy on me) Minitab. I should also have redesigned it so it’s more modern, but I was a little constrained by what the departments I’m providing a service for expect the students to emerge knowing. Still, I must enter a plea of “guilty, by reason of laziness”.
Actually, I was wondering why they were all by different authors (and of course, mostly all new editions — but that’s a different topic).
But I also found it interesting that each department felt it necessary to provide the class for *their* students. Having had basic P&S classes in engineering and in business (separated by 18 years), I can state with certainty that the normal distribution didn’t change in that time, and neither did the method of teaching it. [Well, except the first time was Minitab, and the second Excel (business, you know).] This seems to be one area where it makes sense to have a centralized course, doesn’t it? Much like freshman chemistry.
But then Cosma’d lose out on creating a new class
By the way, Cosma, it’s not “laziness,” it’s “exploring the landscape as it is.”
Heh. May Heaven have mercy on us all: Barbara and I both learned statistics with Minitab. In the Old Days. When it was all there was.
Not like Kids These Days. In our day, you Learned Statistics, by gum, or onto the ice floes you went. And they beat us with sticks, too. Both ways uphill.
Ah well, now you know statistics you at least have something that will help you make angry everytime numbered are mentioned in the papers or on TV.
Argh! s/numbered/numbers/
Anyway, I came back because of Barbara’s question; I guess it is a simple case of staking a claim. If you can give Statistics lectures in your own department, it means you can (sort of) justify one more teaching position. There was a lot of that going on at my university. At least I think there was; my suspicions were never acknowledged of course. (Me being a lazy newsman did not help either when it came to digging up the truth.)
Minitab is a great multiple regression program, according to my husband, but he’s just a nut for that kind of stuff. We still have a book somewhere on it collecting dust in the attic.
We just had a very interesting conversation about whether you can truly ever teach probability and come away with definate understanding, because that would indicate that probability doesn’t change over time. Which he says, probability is the same in any given instant. But I say, it can’t be. Otherwise, wouldn’t a book on probability be as modern today as it was 20 years ago? You’d only ever need one.
And you wouldn’t believe the sentence I just backspaced over. It started something like, “The probability of probability never changing is probably…” Gah!
Ok, what stats books would you recommend for someone with an honours degree in Pure Mathematcs and Computer Science and a reasonable amount of general stats picked up from broad general science reading?
Ideally I’m looking for something abstract enough to satisify the mathematician in me, but with real case studies applications to make motivation easier. I’d like to understand epidemiology papers, for example, (say the Lancet Iraqi deaths study) and be able to read the bits of Cosma’s PhD thesis that went over my head when I looked at it…
Danny, I wish I knew. I wish I knew.
More at the top, this morning.
Danny: “A Guide to Econometrics” by Peter Kennedy is your man. It is the only textbook worth shit on the subject.