University 2020 :6: The one with the results at the end

[MillicentMartha]

Hopefully as the title suggests we will actually get up to results on this thread! Only 5 weeks to go.

Old thread www.mumsnet.com/Talk/higher_education/3918392-University-2020-5-Results-day-approaching-and-beyond?pg=40

@Monkey2001

Well done to @Divoc2020's DS!

@Spababe - brilliant reminder about fieldwork stuff - when I did fieldwork I found my Berghaus quick-dry, neoprene, breathable liners invaluable for keeping my tootsies free from damp rot in my boots Guess I should add new waterproof gaiters - her last pair were when she was 8

I'm now off to check on the wet-weather field kit that she may or may not need for her next academic year, depending on Covid-19

I have been Uni 'shopping' today too but only for the dull stuff (tin opener, glasses, mug tree,scissors etc) and only because it was a slow day at work.

It's very real all of a sudden!

Mug tree! We all had those back in the day, so a little forest in our 2nd/3rd year flat. Happy days!

Have not read all of this yet, but an interesting report from the select committee reviewing the process for awarding grades this year. It is great to see that they want OFQUAL to publish the standardisation algorithm details. Another bit that interests me is that OFQUAL have to look at each group to ensure the algorithm has not been unfair - as I have a DS in this who, like most boys including his brother, would have upped his game in the final months, I would be pleased to see them making sure that the approach did not cause a drop in boys' results, although he will just accept it if that is what happens as GCSEs will not matter nearly as much as A levels.

committees.parliament.uk/publications/1834/documents/17976/default/

If DS misses out on his firm choice uni does he HAVE to accept his insurance or can we enter clearing? (He's having 2nd thoughts over insurance choice)

He can enter clearing SouthCoast

@southcoastshell, no, he can get released by his insurance.

Actually he should be able to release himself this year. I'm sure Clearing Plus allows it this year in case applicants aren't happy with what their chosen unis are doing re COVID.

He could start researching alternatives now as clearing has opened.

Thanks everyone, we are going to a open day next week to check out a local uni we overlooked first time around to see if it's a better insurance option

Hi everyone.
Can I ask what is probably a stupid question....with regard to standardisation of A level results, what does ‘prior attainment of cohort’ actually mean? As in, does it take into account the individual GCSE results of each individual within the A level cohort then standardise from there?
Or, does it just use the GCSE data from the school Itself for the cohort, which surely wouldn’t be accurate as the individual students wouldn’t necessarily be the same?
I’m guessing the former, as sixth form colleges wouldn’t have data for the second scenario. I hope so anyway!

@TikkaMoSalah - I think this was discussed on another thread some way back, but as I understand it, it's not about the individual, it's about the cohort, and that can cohort can be accurately constructed because the individual GCSE results can be tracked and used to compose the cohort. So for a sixth form college, the GCSE results of all their Year 13 candidates are gathered together to create the prior attainment data (regardless of their feeder school).

We have to get used to the idea that they are not moderating the individual, ie they are not saying "should Billy Bright really have been allocated an A when he got all 6s at GCSE?", but moderating the overall allocation in a school, ie "should there be 7 As awarded in this subject taken by 25 students given that the cohort at GCSE got 5% 9s, 10% 8s, etc?"

What seems a bit odd is that I think the idea is that for each subject they look at the GCSEs of the whole cohort not just the GCSEs of those sitting that particular subject. But maybe I've misunderstood the intended approach (or maybe it's a part of the mysterious algorithm that we just don't know about ).

Thanks @DadDadDad that makes perfect sense to me now!
I wasn’t sure about individual GCSE results being tracked to form the cohort, if you see what I mean.
At my DS’s school the GCSE results were not particularly great, but the sixth form is small, with the majority of Year 11’s going on to other sixth form centres. So you’ve put my mind at rest a bit!
Now I just have to worry about the fact that the average A level grade achieved at the school over the past 3 years is a D, and my DS needs AAB; that’s a whole different story!
Good luck to your DS/DD, and everyone else’s too.

No problem, Tikka - the previous discussion that I mentioned was prompted by me asking pretty much the same question - I even quoted sixth form colleges as an example! So, others explained it to me and I'm passing that on...

Of course, as you imply, prior cohort attainment is not the only factor - the grading will also look at school performance in the subject.

I didn’t think about Johnson’s speech for this group. School leavers to me is year 11.
We’d only have got pissed off at him anyway!

@TikkaMoSalah and @DadDadDad a little light reading on OFQUAL's approach, although with the standardisation section embargoed is here:

assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/898202/Requirements_for_the_calculation_of_results_in_summer_2020.pdf

The Select Committee report referenced above said they have to release the details on standardisation as a matter of urgency.

The only reason they can have for not releasing it is that they forsee wholesale challenge, from which one would conclude that they realise that the mechanism is flawed .

Thanks, @Monkey2001. Not much point reading it if the interesting bit has been embargoed!

Obviously, the opaqueness is a cause for concern, but I do have some sympathy with OfQual. In an ideal world, you'd take time to introduce a whole new assessment model, maybe trial it with a few schools for introduction the following year, consult on various statistical approaches, get some research done.

Instead, they've had to come up with something in the space of a few months. And if I were them, I'd probably want to test an approach on the actual data, then spend some time sense-checking the results, tweaking the algorithm, maybe even have two different models to compare, before I felt confident about publishing a robust algorithm without leading to too many bad headlines: "massive grade inflation!" v "teachers furious that students have been harshly downgraded" - that said, the reporting on the IB outcomes suggests they might manage to get both headlines anyway, which would be quite an achievement.

D^3 You're sounding worryingly Actuarial there...

There is still some interesting stuff there, like the fact that they are only using 1 year of data for maths cohort progress as there is only 1 year of the linear A level.

I agree that it is impossible to design a statistical system which works every time, but I can't see how any system will work for schools other than the ones with consistently high results. I have said this before in various places, but at DS1's school they got 14% A in Maths A level last year, 3% the year before with similar sized cohorts >70. No idea how you interpret that, but then for Product Design they got 100% As one year (a cohort of 2) and nothing over a C in another year (cohort of 5). Not sure how well GCSEs would map onto a subject like Product Design as you do not have to have done the GCSE.

My main issue with the system they have developed is that you REALLY need to be able to appeal if the results look silly. There were a lot IB students who ended up with silly results - a parent from a school where all 17 people with medicine offers failed to get the grades posted on TSR. The system starts with a personal approach (teacher assessment) and ends with a statistical approach when it should be the other way round, the assessment of the individual should carry more weight than the algorithm. If there was a better system for schools to be able to appeal if the results are clearly wrong, I would have more confidence in the system.

From where we are now, I think that this year may have been more successful if they had turned the process round and done the statistical analysis for the schools and allocated a pot of grades to be distributed across subjects as fairly as possible, then maybe had a second round of statistical approach if the subject grade statistics were unacceptable. A top down approach where teachers who know the students made sure they got as close to the right grades as possible would surely have been better than what we have.

Maybe it will all be fine; the fact that teachers have ranked students will hopefully mean that students can not slip more than 1 grade in each subject unlike the IB.

I know, it's how I build models...

Actually, there is a serious modelling decision for the grade moderation that I haven't seen discussed...

If you had the GCSE results of all this year's A-level candidates, plus some data about A-level results in previous years, then you could build a model of the expected grade distribution this year for the whole country (x% get A*, y% get A etc), and you'd want an outcome of actual grades to be something close (because the population is large enough that there shouldn't be too much statistical variation from one year to the next).

BUT, now imagine you are applying that approach to a subject in a particular school. Your best estimate model might tell you say that you expect 1 A, 3 As, 5 Bs etc. If the teachers' assessment is 2 A, 4 As, 4 Bs,... then here's the key decision: do you adjust the grades to fit your expectation or do you recognise that the teachers' assessment is within the confidence interval of potential variation (a set that has 20 candidates will have far more statistical variation than a set with 20,000 candidates) and leave the grades unchanged?

(That was in reply to Kingscote, cross-posted with Monkey).

Actually, Monkey - I think you make a good suggestion there about making the statistical assessment first then asking teachers to allocate within an allocation "budget" and justify where they choose to exceed (or possibly fall short) of that allocation for a particular student. Although, I guess the haggling over that could be quite protracted.

They are indeed building a nationally expected results too. All the boards have been asked to feed information into this.

My understanding for the individual settings is that with regard to historic data they are standardising it over one year, two year and three year's data and applying which gives the best result for that setting. (I have now read so many papers I am not sure I can quickly find that point to be able to link to this quickly, for which I apologise).

I think that as there are more factors being built in for gcses and A levels hopefully they will be more robust than the IB results.

My understanding is that when results are issued to schools they will be given details of what standardisation has been applied to their set of results. (Again will need to trawl the various papers I have to cite exact reference). When the information came out I just read it. If I had realised there would be so many versions I would have cross referenced and indexed areas!

I've followed the info regarding models and data but the question I have is are our children likely to get a grade that realistically reflects what their teacher predicted they would have got based on all the data they had?
This might be a how long is a piece of string thing.