-
Asked by amelie to Chris, Dave, David, Fiona, Jack on 19 Jun 2013.
Question: Physicists repeat their experiments hundreds of times due to natural variation. Do you think it is statistically unfair that we only get to do our GCSES once?
- Keywords:
-
Jack Miller answered on 19 Jun 2013:
This is a brilliant question! In short, yes I do think it’d be better to measure your ability lots of times.
The reason physicists repeat their experiments lots of times is because each measurement is uncertain, and by repeating the same measurement lots of times we can start to understand the variation within those measurements, and work out what the ‘true’ value is, together with an estimate of how confident we are of that. We can also try to work out if our method of measuring is biased, and how accurate it is.
In many ways, GCSEs are a measurement too — they’re a measurement of your understanding and knowledge about certain things at a particular point in time. You’re quite right, in that we should really get you to sit lots more exams, and then take an average, and get some degree of confidence in what your ‘true’ mark should be. Whether or not you’d like to sit six long hard exams instead of one, however, is another matter!
-
Chris Mansell answered on 19 Jun 2013:
It is a bit unfair. However, you are probably taking more than one GCSE so if you have a bad day in one exam, maybe you will have a good day when sitting a different exam.
A-levels consist of modules. I think you are allowed to retake some of the modules if you want. So that’s a bit more fair.
Anyway, I hope you get the results you deserve, amelie, or that natural variation works in your favour rather than against you. Fingers crossed. (The best way to be on the safe side, though, is to work extra hard.)
-
David Freeborn answered on 19 Jun 2013:
Hi amelie,
This is an awesome question. The short answer is yes, I think you’re right!
Our performance varies a lot day-to-day, and so measuring it on just one single day, on one single exam is very statistically unfair. The good news is that it’s actually very easy to mathematically model how much random variation we would expect, and find out how unfair the exam system is.
Some mathematicians did precisely this, and ran a series of computer simulations to see what would happen.
Their results were quite shocking: they think that on any particular exam, about 15% of students will be given the wrong grade, just due to random variation. And that’s not taking into account other problems, like errors in marking, or the changes in the difficulty of the papers.
That’s a really, really big margin of error: In particle physics, we normally demand margins of error less than 0.00005% before we allow any important result to be published!
I think a lot needs to be done to change the exam system, to make it a lot fairer for students.
-
Fiona Coomer answered on 19 Jun 2013:
Ideally we’d do each experiment an infinite number of times in order to minimise the error (due to human error, how accurately we can measure anything and also to account for small changes in experimental conditions). Unfortunately, this isn’t possible (due to time, money, human patience), so we repeat measurements as often as we need so that we believe the result. I bet you’re complaining about the number of exams you have already, without having to do them many more times. Also, for each subject I guess you have more than one exam paper and also some coursework? This is a bit like doing an experiment more than once. Good luck though!
Comments
amelie commented on :
Thank you so much for all your interesting responses. In response to Fiona, I didn’t ask because I hated doing too many exams, but because my marks vary so much on practice papers that I couldn’t help but think the system relies quite heavily on luck (as well as hard work). Perhaps exam boards should consider making people sit the same module twice (but different exam papers). It would be annoying but there would be less error so marks would be more realistic.