Reply to Dare to request reduce setting "very" difficult PS on Mon, 22 Nov 2010 10:43:02 GMT

beepeebful — Mon, 22 Nov 2010 10:43:02 GMT

i absolutely agree with you. this is in fact the case with international examinations such as G.C.E. ‘O’ Levels. just with social stratification, we need a measure to identify the smarter people to groom their talents. at the same time we need to identify those people who need more help.

Reply to Dare to request reduce setting "very" difficult PS on Sun, 21 Nov 2010 10:33:49 GMT

autolycus — Sun, 21 Nov 2010 10:33:49 GMT

Actually, the main problem is inculcating the right mindset in the adult population.

A good test is one which allows us to a) determine how good a person is, and b) determine what order a group of candidates should be ranked in.

That means, for a), that a test cannot allow any candidate to score 100%. Why? Because of any candidate scores 100%, then we don’t know how good that candidate is (i.e., we don’t know if that candidate can do better because the paper might be too easy).

It also means that the paper cannot be too hard or too easy, because that would mean all the candidates would end up clustered towards the low end or high end respectively.

Currently, we don’t know how difficult the paper is relative to the population because detailed PSLE stats are not forthcoming (like what the distribution of raw scores is for a specific paper and for each question in each paper). However, we know MOE has those stats.

What is skewing the stats is that every time MOE opens up a new kind of question, tutors, parents, schools etc will all attempt to ‘mug up’ that question or that kind of question. So MOE has to get more creative in setting papers, while still keeping the syllabus accessible.

The only remaining approach is actually to teach children to use their brains analytically WITHOUT recourse to endlessly doing past papers. This is (I guess) what you mean by learning and understanding. It has been tried by some people with some success. But in this kind of social climate, it is hard to sustain. Some kids don’t have the temperament or the cognitive preferences to do it that way.

So, no I don’t think the papers should be easier. I think they’re OK. In fact, I think the ultimate goal is to make it as hard as possible without causing low-end clustering, and still allowing a reasonably normal distribution.