On Tue, Jan 21, 2014 at 12:47:41AM +0000, Steven D'Aprano wrote: > On Mon, 20 Jan 2014 09:08:28 -0500, Roy Smith wrote: > > > In article <mailman.5735.1390198899.18130.python-l...@python.org>, > > Chris Angelico <ros...@gmail.com> wrote: > > > >> On Mon, Jan 20, 2014 at 4:21 PM, Dan Stromberg <drsali...@gmail.com> > >> wrote: > >> > I did a short time of teaching while I was in school. If three > >> > students all turned in the same assignment, they all got docked > >> > significantly. There was no "who copied off of whom?", it was > >> > "someone shared when they shouldn't have." > >> > >> What a wonderful way to promote an attitude of "my code is MY CODE and > >> should never leave my sight". What a delightful way of thinking to > >> unleash on the world. > > > > That's a little harsh. Working in groups, and sharing code, are > > important parts of how software gets developed today. Those > > collaborative work habits should indeed be taught. But, school is also > > about evaluation of progress. At the end of the class, the teacher > > needs some objective way to figure out how much each student has learned > > and assign a grade. It's hard to do that if people aren't handing in > > assignments done individually.
I agree that it is unfortunate but there's a bit of a balancing act with this. The problem is that there are two sometimes conflicting roles in education: teaching and assessing. When you set assignments the students will usually learn more if they work in groups. However at some point you need to try and assess how much they've individually learned. I find in practice that it's easy to tell when a student has copied someone else without really understanding what they're doing though. Of course if they just pay someone else to do it for them then there's not much you can do... > > An objective way to figure out individual progress is easy. It's called > an "exam" or "test". Admittedly, it's normally only practical for > examinations to last no more than a day for senior students, and an hour > or maximum two hours for junior students, and some subjects are more > easily tested this way than others. But you can still examine a lot in a > couple of hours. If you're interested in accurately measuring the > learning of individual students, there is at least one pretty damning > problem with assignments: just because student X puts his name on the > paper doesn't mean student X wrote the paper. Assignments are effectively > based on the honour system, and we know how well that works. For those > with the money to spend, you need not do a lick of work to get an A. The real problem with exams is that exam conditions are so unrepresentative of real work. How often do you use the internet, or documentation, or text books etc. in your own work? How often would you have to do something without having anyone at least to discuss the idea with? But yes it's absolutely necessary to have some exams or else the whole system is open to abuse. > > Perhaps that's why Harvard has just given up even trying to distinguish > the students who learn things from those who don't? Forget George Bush's > "Gentleman's C", Harvard now practically gives A's away to anyone who > shows up (and pays the fees). I think that's a little harsh. To say that the majority of students get an A- or better does not mean that they give A's to "anyone who shows up". I would expect that the majority of students at Harvard do a lot more than just show up. (I don't know much about Harvard specifically but this is true of most universities). > > http://qz.com/153694/the-most-commonly-awarded-grade-at-harvard-is-an-a/ > > Presumably they're protecting their business model. Students are > customers, and if your customers are paying a small fortune to attend, > they need to get something in return. Knowledge is good, but you can't > put knowledge on a CV or frame it and put it on a wall. > > It would be interesting to think about the incentives which have lead to > an over-reliance on take-home assignments rather than exams, as well as > the pros and cons of one versus the other. Don't get me wrong, there are > advantages to assignments as well, but I think that the total prohibition > on collaboration is misguided. The question in my mind is how to > encourage students to learn from each other rather than to merely > mechanically copy from each other? > > Relevant: > > http://qz.com/157579/confession-of-an-ivy-league-teaching-assistant-heres-why-i-inflated-grades/ I can definitely empathise with what she says. Once I started marking assignments it quickly became apparent that my standards were higher than those of other people. Every now and again I would mark a big assignment and get a deluge of grief from the students who had done badly. If it's a small assignment (say 5 students) then you can build something out of that and spend time preparing them for future assignments. If it's a big assignment (100+ students) then it's just a whole load of grief that no one really wants. The problem of students giving you grief doesn't really happen with exams because in that case if someone complains it's not you (the original marker) who has to talk to them and remark it. Where I work they have to fill out their feedback forms before they take the exam so they can't use that to complain about the exam being too hard or being marked too harshly. But what does happen is that if the average grades are too high you get in trouble for the exam being too easy. If the grade is too low you get in trouble since you've apparently done a bad job teaching. There's a conflict of interest right there (being both the teacher and the examiner) and it basically results in everything adjusting to the ability of the students rather than measuring it objectively. Also I don't know why the Harvard TA says that this isn't a problem in the UK. Here in the UK the government does it with national externally marked exams: http://www.theguardian.com/education/2012/may/01/gcse-alevels-easier-says-ofqual I would estimate that over the past ~50 years school standards for Maths and Physics in the UK have slipped by ~1 academic year. Perhaps that's why we make them do 4 years at university now... Oscar -- https://mail.python.org/mailman/listinfo/python-list
