Talk:Ofqual exam results algorithm

Schools Start‑class Mid‑importance

	Schools portal This article is related to WikiProject Schools, a collaborative effort to write quality articles about schools around the world. If you would like to participate, you can edit the article attached to this page, or visit the project page.SchoolsWikipedia:WikiProject SchoolsTemplate:WikiProject Schoolsschool
Start	This article has been rated as Start-class on Wikipedia's content assessment scale.
Mid	This article has been rated as Mid-importance on the project's importance scale.

This article was nominated for merging with 2020 UK GCSE and A-Level grading controversy on 20 August 2020. The result of the discussion (permanent link) was Not merged.

The algorithm

This is very clearly written, subscripts would be nice. I wonder how many people would understand that CAG and k both depend on student, that is, if the students are numbered in rank order 1,2,3,...,n then CAG(1), CAG(2),..., CAG(n) are the corresponding centre assessment grades, but also k(1),k(2),...,k(n) in increasing order would be values of the random variable with distribution given by the formula on the right (for each fixed value of j).

A question is, how are the values k(1),k(2),..,k(n) meant to be chosen? If we know that the desired distribution (the averaged and corrected distribution on the right side of the equation for each j) gives a particlar proportion of each grade A*, A, B, .... then we can allocate the k(i) in order to those grades. I suppose in general as the area under the curve is 1, we would choose k(1),k(2),... so that vertical lines drawn on the graph with those horizontal coordinates divide the area into n+1 parts each with equal area 1/(n+1). So k(n+1-i) is the inverse image of i/(n+1) under the function which is the integral of the distribution. In other words, the k(i) chosen so that the cumulative distribution function evaluated at k(1),k(2),...,k(n) would be each of n/(n+1), (n-1)/(n+1),..., 1/(n+1). So 0<k(n)<k(n-1)<...<k(2)<k(1)<1.

I guess the unfairness was that grade inflation is controlled *for each individual student* and in *each individual course*. While the method of controlling grade inflation doesn't matter much for the final statistics, from the standpoint of fairness for individual students, it isn't right to say 'Here's the maximum grade you could ever have received based on previous performance of your school'.Createangelos (talk) 21:46, 23 August 2020 (UTC)[reply]

@Createangelos: thanks for your kind remarks. You are correct we need to expand on CAG s and how it was implemented to create unfairness. Superscripts are easy^3.. Where do you _want? ~~Start over~~ So starting as we always do- what are the references we should use. I need some clear referenced text to work with. Can you draft up something here, and I will work it in under the algorithm box for you to correct.ClemRutter (talk) 23:28, 23 August 2020 (UTC)[reply]

OK, have to be careful as it is efficient and clear writing the way it is. I just worry that almost no-one reading it will understand it. Very condensed and intelligent writing but the notation is unusual. There were descriptions which ofqual had put online which could serve as references, they were unreadable. Createangelos (talk) 23:38, 23 August 2020 (UTC)[reply]

Links to Controversy

As the articles were not merged, it would be good to review if some content should be in the other article. The material about the Royal Statistical Society seems to be about the controversy. And it does not describe side effects of the algorithm. Ca3tki (talk) 15:35, 25 August 2020 (UTC)[reply]