Algorithms or algorithm?

Discussion in 'Education news' started by aypi, Aug 22, 2020.

1. aypiSenior commenter

The Scottish algorithm produced a similar result to the English, Welsh and Northern Ireland algorithms.
Did all the exam boards go to the same algorithm shop and buy the cheapest one they could find?
Did one exam board do the work and the others copy? Not knowing that they had got the answer wrong?

It is a serious question, how did all the boards use a similar if not the same algorithm?

ajrowing likes this.

All authorities used algorithms which operated rse-about-face.

Normally students will sit the exams, gain an individual grade (important point, that) which is then grouped with other individual grades to allocate a grade for the school which then is used in the process of ranking schools in league tables.

All the algorithms operated in reverse, starting with the school's position in previous league tables and attempting to adjust the teacher-assessed grade for individual students.

Didn't work!

The other problem is that teachers tend to grade according to the cohort, allocating top grades to the best students in that cohort, rather than allocating grades according to the national picture. The best in a class is unlikely to be anywhere near the best in the country.

Sadly, the decision to accept CAGs almost certainly means that a good number of students will now have elevated trades, will use those to enter universities where they will struggle.

ajrowing, corgie11, dleaf12 and 2 others like this.
3. ScienceGuyEstablished commenter

I expect the end of this year will be brutal at some Universities - rather than pass the year to continue, I expect that Universities will be told to reduce numbers back to the usual number in a cohort

Some universities are asking students to delay going until next academic year. If they do that, next year's cohort will be larger than usual too.

phlogiston likes this.
5. offhegoesNew commenter

Surely universities already use NRTs to determine pass marks in modules? I can’t remember there every being an objective definition of what a pass looked like in a university module.

But yes, I imagine that the threshold for a pass just got higher than usual.

6. colpeeStar commenter

Perhaps some might struggle, but some always do. University is such a step change from the average A-Level offering that students can blossom or wilt at the new situation. I am not sure a grade here or there is what makes the difference though.

7. gainlyStar commenter

I think a bigger concern will be that most year 13s won't have done any work since March and often won't have finished the A level course. For example, one local school hadn't even started integration in A level maths, which is a very major part of the course and probably the most difficult. It will be difficult for them if they are going to study maths or a mathematical subject such as physics or engineering. Perhaps the university will have to introduce some remedial courses.

dleaf12 and strawbs like this.
8. aypiSenior commenter

There was a medical school in Texas perhaps that had to dramatically expand their intake, something like 20-40 percent, they took students that they had earlier rejected. The rejects did as well as the original selection, so there may not be a clearout at end of first year uni.

To re-route my thread, the original Q was:
It is a serious question, how did all the boards use a similar if not the same algorithm?

9. gainlyStar commenter

I don't know, but I would think they all used different algorithms but all with the same objective, i.e. to adjust the teacher predicted grades so that nationally the overall grades were in line with previous years' results.

The odd thing to me is that nowhere did anyone appear to have looked at the individual grades which were being produced by the algorithm before they were released. They had months to do this. What were they doing all that time?

ajrowing and dleaf12 like this.

I think dignifying the process they used with the term "algorithm" is to be rather generous. Centrally "Calculated Grades" was a doomed endeavour from the very start.

As any fule noes, statistical processes are largely non-comutative - you just CAN'T run them in reverse (see @nomad's post up above) - its just wrong maths.

Even so, SLT's up and down the country regularly do just that when they produce an individual pupils "expected grades" from Fischer Family Trust data and similar (don't even get me started about RaiseOnLine or whatever its called now).

There was a very telling comment which escaped into the public domain from, I think, the Office for Statistics Regulation, who were beginning an investigation into what went wrong, to the effect that many of the OfQual officials "simply did not understand the maths they had just typed in" - I was only half listening to the news at the time and wish I could find it again, but I remember being impressed both by the unusually honest lack of weasel-speak and that the hunt for the guilty was underway so fast... The interview was not repeated in the next bulletin...

No one looked at the individual grades because to have done so would have been to question this whole idea of reversing out "data" that is supposed to be meaningful for an individual student from nationally aggregated data - and with it the whole edu-data business that underpins league tables, performance related pay, performance managment and all the other paraphenalia that has been used to suck the joy out of Teaching.

11. aypiSenior commenter

I have a spreadsheet that I put the class marks into.
Class sizes change, the exam questions change. Sometimes the copying of cells does not go the way I expect (missing \$).
I always look at the results and see if they "look right" and if not I poke about, and the spreadsheet gets there.

I only started the spreadsheet when I realised how parc I was at adding up the marks of an exam paper.
The functionality grew with the years.

Yep, many of us use something in this line to analyse what is causing difficulties and thus what we need to give support with a class. But I bet you never looked at what caused issues for the class three cohorts ago and applied it to this year’s class! - and that is sort of what Ofqual did.

13. aypiSenior commenter

In the same class I had two B students go to an A, so I am marking too hard , and two c students go to a D so I am marking too lenient.
I cant be both.
This was a class of 15, obviously something wrong.

Wait, what? A school didn't teach any integration for 18 months and expected just to squeeze it into the last 6 weeks before the exam? That sounds... incredibly stupid.

15. gainlyStar commenter

They'd done the year 12 (AS) integration but hadn't started year 13 integration.

16. nervousnedSenior commenter

6 weeks before the exam, the only thing left should be revision.

That's slightly better but still not great.

18. gainlyStar commenter

It was actually 11 weeks from when schools closed to when the first A level maths exam should have been.

19. nervousnedSenior commenter

I would have expected to finish teaching the content by March.

20. gainlyStar commenter

I only tutored a couple of year 13 students from the school, or at least I did until exams were cancelled. Although it's a grammar school, the maths department doesn't seem to be very good.