'Marking on the curve; is it the solution to A-levels?' blog discussion

JimmyTheWig

BurningSnowman wrote: »

WRT discouraging entry, in my experience it happens more at the learning stage than the exam entry one. If your GCSE or AS marks suggest it's not going to end well, you'd be discouraged from taking AS or A2 (respectively). I don't think most places would stop you entering an exam at the last minute and sabotage any chance you had for the whole qualification.

Not at the last minute, no, but the school may be more inclined to encourage you to drop a subject half way through the year if they didn't think you would pass if the school was more results orientated than learning orientated.

nomoneytoday

The pass rates were on commutative frequency curves - but for some reason more people pass. Doesn't seem logical to me.

niceguyed

I think the system that Martin is referring to is called "norm referencing". An obvious benefit is that you can tell the chaff from the cream in a particular cohort. There are inherent problems with this system.

1) The big one. You cannot measure standards easily year on year as results are fixed or tied to a certain number of students achieving each grade boundary.
2) No parity between raw marks and what actual grade you will achieve each year. How would you feel if you got a C when the previous year it was worth an A. Ouch.

Overall the current system is fairer to students.

bc3000

In the current system you have to get a certain amount of UMS marks to get a certain grade, I'd prefer that to a random % getting something. I agree that AEAs are good, they'll test people's knowledge much more than getting 95% in something... I did Spanish AEA It was a lot different to Spanish A Level and I think it proved more than getting a higher UMS mark at A Level.

Flobajob

bc3000 - the UMS marks (usually out of 100 per paper) are set by scaling up the actual marks for the paper (which is often a weird number of marks, e.g. 72 for most maths papers). This allows the exam boards to make an A whatever they want, e.g. if very few people get near 72/72, they can make everyone with 48 marks or more get an A (i.e. get 80 UMS or more), and conversely they could let an A correspond to 60/72. The actual boundaries for grades are set by a committee, and what usually happens (or so I was told by a reliable source) is that they just set the boundaries based on the proportions of people getting As last year. As an example, see here: http://www.mei.org.uk/files/papers/Jun07Marks_Grades.pdf , and here: http://www.mei.org.uk/files/papers/Jun06Marks_Grades.pdf

I think one of the main reasons why this problem has occured is because a lot more people are going to university now compared with 30 years ago, so A-levels have to distinguish between candidates near the top and the bottom of the scale (they can't just fail half of them like they used to). This means many A-levels include quesitons at the beginning that are more straightforward (that almost everyone can have a crack at), getting harder as you go through. Of course, this means that it is easier to score highly, and it means that if you mess up a question you are punished very heavily for it (since most good candidates will get the first 1/2-2/3 of the paper without too many problems).

Also, there seems to have been a move from questions that test 'true' knowledge of a subject to questions that ask you to jump through hoops, e.g. in A-level philosophy, you don't tend to get the marks if you don't give the exact examples used by such-and-such a philosopher (this is probably due to markers looking for 'key-words') - i.e. you get marks for regurgitation of facts. Compare this to IB where you only get the marks if you use your own example (so you have to understand the example to some extent). The AEAs and STEP papers, plus the LNAT, BMAT etc seem to be designed to test not knowledge (since only A-level knowledge is required) but ability, and that's why they are used by so many Universities to help with admissions.