Monday, September 03, 2012

Exam grading - a simple one point change to solve much of the problem

I was close with my entry regarding grading on a curve, but like the politicians I was still stuck in the Victorian mind-set of grades to differentiate between students. It took a conversation with my family to suddenly click as to what should be done and how simple it a change it would be. So an big thanks to my mother for allowing me to explain it to her and a big thanks to my father who started me down this path.

The simplest method for removing the problems with grades and year-on-year comparisons is... remove the grades. Rather than create an arbitrary A-Fail set of boundaries mark as a percentage for that exam. If you score in the top 5% of your year then your 'grade' is "5.25% June 2012" if we want big numbers to be better we can switch to "94.75% June 2012"; but I think the former is a simpler concept.

Consider how much easier this will make for comparisons particularly for application to colleges or universities. Rather than state a requirement for an A-C in English and Maths and two A-C in other subjects simply state that an applicant needs to be within the top 50% of their year for English and Maths and 50% for two other subjects. Get someone with a 49.82% 'grade' and it's easier to see how close they were rather than an arbitrary D grade.

We'll still be able to check the quality of a paper using the standard distribution/normal graphs and both schools and employers will have an accurate gauge of examination results to look at.

Are there any flaws. I can see a couple. First "outliers" If you have a year with some particular intelligent students (or particularly unintelligent ones) a student's actual score on a paper might have meant they'd be in the top 10% of the year had these others not been present, but instead got pushed down a little. However given the quantities taking the exams these should even out.

A second flaw (though not really) is that the politicians would be unable to boast about how their changes to the exam system has resulted in a higher pass rate. The top 5% will always be the top 5% and the numbers only change when there are more or less taking the exams. But then again if they're grading to a curve anyway this is already the case.

The third flaw involves modules, which Education Secretary Michael Grove quite wrongly wishes to do away with. How to mark them? Just use the same percentage to calculate the final percentage. So a student takes four modules and gets 10%, 20% 15% and 22%; and get a 16% mark. They could chose to retake the 20% and 20% modules, but with percentages even if they get a better raw grade they might end up with the same or less score due to more people getting higher marks due to paper differences

So no need to create new methods of A* grades or mess about with how the exams are taken; just use a percentage to indicate a student's place within that year.