Oh man, I have to pick just one. I'll pick a lesser-known one. TEAMRAMONYCAJAL PRESENTS: A DIATRIBE ON THE SAD STATE OF UNIVERSITY BIOLOGY EDUCATION 1) Statistics is not seen by still-too-many undergraduate programs as anything other than an elective. Hogwash, I say. My undergraduate program has required me to take one course in biostatistics, and I elected to take what was essentially Biostatistics Part 2 as part of a research semester. A lot of people like to say 'Why require calculus as part of a biology degree? We never use it!' Now I think calculus is damn necessary because it's increasingly the case that we're going to have to incorporate higher-level math into our work (MATLAB is a big thing in parts of neuroscience, for example, and it requires LINEAR ALGEBRA), but if you're going to drop calculus as a requirement you'd better fucking replace it with stats. And yes, that means two semesters. And yes, most biology majors are premeds or stuff like that. If you're a doctor, you had better fucking be able to interpret medical journal articles and keep up with your field, and that means knowing how the fuck statistics work, from the simplest goddamned t-test to a big ol' convoluted case-control thingy or some wackily complicated regression (and know how to talk to a biostatistician if it's maybe a little beyond your ken, which requires some biostatistical knowledge anyway). SOLUTION: REQUIRE SOME FUCKING STATISTICS 1a) If you're going to teach statistics, teach a fucking statistical program, and to the point TEACH IT WELL. I learned statistics using R, and my biostatistics professor was fucking abysmal at teaching us R and used R Commander as an interface. I didn't fucking understand how to use R when my professor of my entirely elective research design class taught us more advanced stuff in R, and forced us to do it at the console. I am a better statistics user for it. SOLUTION: FUCKING LEARN HOW TO TEACH A STATISTICAL PACKAGE AND INTEGRATE IT WELL INTO THE STAT CLASS 1b) I had the dubious privilege earlier in the semester of having to basically teach some poor sophomore huge amounts of statistics in an hour, using R. This was not something I was required to do, but I overheard her having difficulties and decided I'd step in. She had not taken the required statistics class, but yet her ecology and evolution class was requiring her to do even some stuff that wasn't required in the required statistics class. Out of the goodness of my own black little heart, I did for her what her professors neglected to do: teach her how to do a T-test, an ANOVA, a chi-square test, and a linear regression, send her a little 'cheat-sheet' with the tl;dr of the theory, what each test actually tells you, and how to do it in R, and she thanked me profusely and told me I'd explained everything in a crystal-clear manner. I was told by the resident biostatistician in my department that part of the problem with integrating statistics into the rest of the core curriculum is that a lot of scientists are still uncomfortable with statistics. WHAT THE FUCKING FUCK. SOLUTION: HAVE SOME PROFESSORS WHO AREN'T UNCOMFORTABLE WITH STATISTICS TEACHING THE CLASSES, FOR FUCK'S FUCKING FUCK SAKE FUCK IT. 2) At least at my university the core curriculum and all the concentrations are continually in flux and it's fucking stupid. Yes, biology, by its nature as a science, changes with all the new information we get. No, that is not an excuse to change the degree requirements like Italy changes its prime ministers. SOLUTION: PUT SOME FUCKING THOUGHT INTO THE CURRICULUM. 3) Vishnu Roosevelt Muhammad Christ Selassie Guan Yin on a pogo stick, I had the displeasure of having to edit several classmates' papers for a required technical writing class for our major. WHY THE FUCKING FUCK CAN'T HALF OF YOU DUMB SHITS WRITE. SOLUTION: STEP IT THE FUCK UP ENGLISH DEPARTMENT I'll add more rant as I think of it. EDIT: I just edited the group project for the technical writing class I am required to take. SIDDHARTHA AMATERASU H. THOR CERNUNNOS, WHO THE FUCK LET YOU IDIOTS INTO COLLEGE.
Had the same problem in stats for engineering, which was required for my Bio degree. R. Everything in R. All we spoke was R. Professor could barely make an coherent sentence without writing a proof. His wife was an Ayn-Rand acolyte macroeconomics professor who wore those weird round-bottomed medical shoes. They made sense, but I can't express why in words. Regardless, I have all but completely forgotten how to use R. I definitely became less enamored/'scared'/mesmerized about stats and feel much better about criticizing their usage, including arguing with my current boss about why student's t is NOT going to be as sufficient as ANOVA.
What made me learn R in my second semester of statistics was that my professor actually gave us references with the exact command and all the variables and what they meant. He EXPLAINED it. He didn't make us scour the internets for what the hell we were supposed to do for a two-sample t-test. He gave us the commands. He expected us to know how to use them. Did you explain to your boss that Student's T is an ANOVA for only 2 samples and that for ANOVA you have to do Tukey's HSD (which is really not that hard in R)? This is the thingamabobber I typed up for the sophomore, published here just in case anyone else on Hubski wants a mini-statistics and mini-R lesson:
IMPORTING YOUR DATA INTO R Say you have a set of data. Before you do anything with this data at all, make sure each variable is in the columns, not the rows (i.e. the headers are at the top of each column and the data goes down the column). This makes working with the data a lot easier. Let's assume you use Microsoft Excel 2007 or later, because that's common, and that you have this little sheet of data, called dogbones.csv (save it as a .csv, because R hates Excel. Also, save it in your My Documents folder, because this is where R will pull datasheets from). Even though the data's going to be typed out here in rows because this is an email, imagine that it is in columns. To import this data into R, there are a few ways to do it. If you have a small dataset with one IV and one DV, you can type: If you have a huge dataset, you type: dogbones <- read.csv("dogbones.csv", header=TRUE) To get the same sets as 'dogs' and 'fmrlengths', you type: TESTS One-sample T-test: Let's say we're comparing a population and we want to see whether it's significantly different than a mean of 3. Two-sample T-test: Let's say we're comparing heights of men and women, and that these variables are coded as 'hmales' and 'hfemales', and that we're assuming equal variance. All you type is: Pearson's chi-square: For a 2-way table called 'townsmog' where the rows are towns A and B and the columns are 'bothered by pollution' and 'not bothered by pollution': Correlation test: Is 'money' correlated with 'intelligence'? Linear regression test: Assuming 'money' is correlated with intelligence, how much does 'intelligence' determine 'money', and is this significant?
So dog 1 has femur length 14, dog 5 has femur length 18, and so on. Dogs: 1 2 3 4 5
FemurLengths: 14 15 16 17 18
and ignore the more complicated dataset commands. dogs <- scan(1, 2, 3, 4, 5)
fmrlengths <- scan(14, 15, 16, 17, 18)
(as an aside here, before you import a data set in R, it's a lot easier if the variable names are kinda condensed). dogs <- dogbones$Dogs
fmrlengths <- dogbones$FemurLengths t.test(population, mu=3) t.test(hmales, hfemales, var.equal=TRUE) chisq(townsmog) cor.test(money, intelligence) moneyIQreg <- lm(money ~ intelligence)
summary(moneyIQreg)
There were underlying issues that went further than just the type of test to be using, I meant it more as an example. Also, while I oh-so cherish my time spent in R, I'm not doing anything that necessitates that level of control, nor is that particular skill (working in R, not stats knowledge) set something that needs to be maintained. I also spent a bunch of time messing around in LaTeX, thinking that would come in handy later, or at least be interesting enough to merit working in, but in the end I have a fancy-looking CV that is really annoying to update.Did you explain to your boss that Student's T is an ANOVA for only 2 samples and that for ANOVA you have to do Tukey's HSD (which is really not that hard in R)?
I use Word. I'm so glad biology doesn't have a bizarre practice of using LaTeX for everything. What the fuck, physics.
yeah, university really isn't all that, is it? I'm doing computer science, it's kind of scary how many people complain that: they don't need to learn any maths they don't need to learn any computer science theory they don't need to learn how to program (or don't bother), because they'll learn on the job they don't need to learn how to write they don't need to learn how to present shit uhhh
Oh, university is most certainly all that, and if you disagree you haven't put enough into your education, but many of the students aren't. Do half of the students expect to be the stereotypical unwashed code monkey whose only job is typing code and who never gets brought out as a representative of whichever organization, industrial or academic or governmental, that they only got a job at because they can code?
OK fine, "not all that" perhaps wasn't the right expression but yes, I more or less agree with you. I think I had this idea that university would do the job of making people interested in computer science, when in reality, by the time you get to university level that's really up to the student. But then, there aren't any good pre-university computer science courses in the UK. So where are people going to pick up the computing bug? Nerdy parents (in my case it was my father)? There were a few extra-curricular things at my school but then not all places are lucky enough to provide stuff like that. It's a big problem, one that has been in the news in the UK quite a bit as well (the only sensible thing that Michael Gove has talked about, actually). So quite a few people are only here because of the prospect of a decent job, which is fine I suppose but that seems like a pretty grim attitude to have. But then I've met a lot of people who are way more driven than I am. Which is for the best, really. And yeah, I think I did OK because I'm interested in compsci and alright at the subject. So that's nice.