UKC

I was discussing my daughter's progress at school with my wife, who is herself a teacher here in France, and we got around to talking about grade boundaries.

In France, they mark our of 20 for some bizzare reason and a mark of 12-14 is deemed as fairly acceptable, but to ke 60-70% seems quite low. I was convinced that in the UK, the grade boundary for a C grade was 70%, but was shocked to find that it's actually 60%. I don't know if this is after they have normalised the grades, because that could mean someone gets a C grade pass when in reality they got less than 50% in the actual exam.

Now to my question. Has this always been the case for grade boundaries? Maybe my memory is failing me, but I seem to remember that when I did my GCSE exams that the boundary for C was 70%. Can anyone point me to a table where I can see how these boundaries have evolved over the years, if indeed they have changed.

Cheers.

In reply to Phil Payne: no, but I think you've got a case of wishful thinking. I think a pass at O, A level was 45%, 60 for C seems right

In reply to wbo:
I just found this though and it says pretty much exactly what I think about modern exam results. I also happened to sit my GCSE's in 1996, the same year as the author, so am happy to see that I wasn't wrong in thinking that there was a more normal distribution back in my day.

http://www.soylentdave.com/2011/09/are-exams-too-easy/

In reply to Phil Payne:
You might be amazed at how high a grade you can get at GCSE with a mark far below 60%.

In reply to Phil Payne:
The grade boundaries vary from paper to paper, subject to subject and year to year.

Here is a random selection https://qualifications.pearson.com/content/dam/pdf/Support/Grade-boundaries...

When I write exam papers (for internal use in my school) I make the grade boundaries not on a distribution or a set percentage, but by looking at how many marks are available in the particular paper at each grade and then making the boundaries match this. This means we can fairly and accurately give a grade that reflects the difficulty of the paper. This means we can make the exams easier or harder for different abilities so that everyone gets a paper that they can answer, and that will stretch them. So for the top set, the % to get a C might be only be 30% but for a lower set it might be much higher.


Now that we are all used to letter grades at GCSE and what they represent they have made them numbers

In reply to Phil Payne: why do you think the results should have a normal distribution ? I think that's a bad assumption to start with. I'd expect it to look like it does. Datasets that have a lot of criteria constraining them often don't trend to normality

Also the idea of normalizing around the averages is pretty stupid as it means if you had a particularly lazy year followed by a year of shots, you'd equalize them incorrectly. Hopefully the data population is big enough to avoid that, but still, in smaller subjects.

In reply to wbo:

Statistics was never my strong point, so I can't tell you why I expect to see a normal distribution of results other than the fact that it feels right that it should.

If it was possible to set exams at the exact same level year after year, then any gradual change in either direction would indicate if todays kids are getting more or less intelligent. Any large jump in either direction would show up when the examiners got it wrong and set an easier or more difficult exam.

I looked at the pdf that marsbar posted and was shocked to see just how low you could be in mathematics and still get a grade C in GCSE mathematics. Surely getting below 50% in an exam should be a fail! Why change the grade boundaries by so much so that 70% achieve A-C grades because that just muddies the water and doesn't allow the truly bright kids to stand out.

In reply to wbo:

You could argue that a paper producing a normal distribution (of either grades or scores) is a badly written one. As for numbers, if you want a high mark / percentage for a given grade, make the paper easier (or at least those questions most relevant to that section of the ability profile), and vice versa.
But agree with point above, great idea to change system from grades to numbers - many, many parent-teacher minutes to be invested over the coming years on that one...

In reply to Phil Payne:

On your final point - have no fear, the new system is doing its best to simultaneously muddy the waters and at the same time help the brightest stand out!
While 'exam performance' distribution would reasonably be expected to be normal, an exam paper has the job of separating that into equal-sized bands. That's done through both good writing of the paper and the allocation of raw marks to grade boundaries.

In reply to Phil Payne:

You missed my point completely. There are 2 ways to get a C in maths, around 40% on the hard paper or around 80% on the easy paper.

Why should getting 50% be a fail when the paper is harder. That's totally illogical.

In reply to marsbar:
Having a hard and easy paper is a totally rubbish idea. It's just like the old system of CSE's and O levels. All pupils should be given the opportunity of getting an A grade and it's wrong that they should be put forward for an exam where the highest mark that they can hope to achieve is a C. If they are really capable of getting a C in the easy paper, then they should be able to get a C in the hard paper and may even get a B or even A if feeling suitably inspired on the day.

By your logic, it's harder to get a C on the easy paper than it is on the hard paper.

In reply to Phil Payne:

But it's harder to meet your aspiration of letting the brightest shine in a single paper that also tries to separate all sections of the ability spectrum. It's all about separating out that normal distribution you referred to into a nice, clear set of discrete and objectively established groups.

In reply to Phil Payne:

In theory it should be the same to get a C. Maybe a little easier on the harder paper.

It's impossible to sensibly enter the full ability range for one paper. Some people are not in any way capable of getting an A or B no matter how hard they study, they just don't have the ability to learn and understand the algebra needed. It would be cruel to make them sit the harder paper. Meanwhile the questions at the start of the easy paper would be a ridiculous waste of time for an A grade student.

There is an overlap, usually A* to D on the hard paper and B to G on the easier one. It's very unlikely that someone who could in any way get an A would be entered for the easier paper.

In reply to sg:

I started off not liking the French system, but now I see its merits. If we do away with the whole grading system and just have a percentage mark we can accurately see how everyone did in an exam. Being able to get a grade C (which is considered an ok pass and a prerequisite for some further education/employment) with as low as 40% is just a joke if you ask me.

I have recently sat exams where the pass mark is 75% and the only 2 marks are pass or fail. If the putlrpose of an exam is to test knowledge and understanding then being able to get a pass with less than 50% is pretty pointless.

In reply to Phil Payne:

And I thought from the title we were talking about climbing grades! Must have thought I was on a climbing forum.

Martin

In reply to Phil Payne:

I could easily set a Maths exam where only the very brightest dead cert grade A pupils would achieve 20%. But it wouldn't differentiate at all between the vast majority of pupils who would score near enough 0%. Equally well, I could easily set an exam where only the dimmest and laziest would fail to achieve 80%. There is nothing magic about 50% (I have to regularly point this out to pupils).

In reply to marsbar:

You're a teacher aren't you? I find it shocking that you think it's acceptable to decide what someone is capable of and limiting the possibility of them achieving their maximum.

There will always be some pupils that are capable of more, but will choose the supposedly easier option because that's what teenagers do. Removing that option stops them from being able to do that.

They could always put the easier questions at the start of the exam so that those that are maybe not capable of achieving the higher grades are not daunted or become flustered by seeing the harder questions until later in the exam.

In reply to Phil Payne:


You're not a teacher are you? Teachers do not decide what a pupil is capable of. The pupils demonstrate what they are capable of and are entered for appropriate exams accordingly.

In reply to Robert Durran:

That's where the examiner has to do their job properly to set a reasonable exam. Allowing the manipulation of grade boundaries after marking means that the exam boards can be lazy in exam setting because they can just adjust them later if they mess up and set an exam that is too easy or difficult.

In any case, aren't most subjects marked mostly on coursework, so that pupils can be helped out by parents and google and can achieve a C grade with 0% in the exam. I have a GCSE in Latin which I got almost entirely based on my coursework because I know for certain that I didn't have a clue what I was doing in the exam.

In reply to Phil Payne:

Ffs. I don't limit anyone's maximum ability. It's my job to decide what someone is capable of, and then to push them as hard as possible to achieve that. This ridiculous idea that it's impolite to point out or even believe that some people are not as clever as others is nonsense. Your shock is an emotional response to a logical approach.

I can assure you that as a teacher I can tell the difference between a kid thats not trying and a kid that can't do it despite trying their best.

My GCSE class last year (set 8) all got grade G except one who got Entry Level 2. I'm extremely proud of each and every one of them, despite various learning difficulties,epilepsy, Down's syndrome etc they all passed their exams. They worked very hard to do that. So did I. Any that had been in that set but were capable of more were pushed and moved up to set 7.

If you think I limited them by preventing them from getting an A by giving them the easier paper then you are beyond clueless.

In reply to Phil Payne:


An examiner sets an exam for a purpose. If it is tease out the very brightest it might have a median mark of 20%. If it is to tease out those really struggling, it might have a median mark of 80%. Many exams will be something in between and do neither of these things. But there is definitely nothing special about 50%.

In reply to Robert Durran:

I don't actually know how it works in the UK. Who decides which paper the kids sit? Teacher? Pupil? Parent?

If it's the teacher then all I can say is that I think they are failing their pupils by not letting have the opportunity to achieve the maximum. Likewise as a parent.

I always try to encourage my kids to have a go at things even if they are hard and knowing that they might fail. What sort of message are we sending if we suggest they take the easy option because there's less chance of failure?

In reply to Phil Payne

Well, when you have your maths PGCE and decades of experience teaching maths in secondary school to the ability range from A* to severely brain damaged and can't manage anything beyond counting on fingers then you can tell me and the examiners how we are wrong.

In reply to Phil Payne:
I haven't taught GCSE for a few years but the exam papers did start with easy questions and then got progressively harder.
Some subjects never used to have separate higher and lower papers and all pupils sat the same exam.
Some pupils will never gain a C grade no matter how hard they try.

In reply to marsbar:
Maybe it's insensitive, but is there even any point in having a set 8 in maths and pupils getting grade G. Maybe at some earlier stage in their life they could have been steered down an alternative path in education in something that might actually be useful for them. What the f*ck use is a g grade GCSE in maths apart from wasting a couple of years at school.

In reply to Phil Payne:

The assumption is that most pupils will sit the harder paper. Low ability pupils with no chance whatsoever of achieving a D will sit the lower paper. Those in the middle will have the decision carefully made on an individual basis taking into account mock results and all other information with discussion with the parent and pupil. Last year 1 parent disagreed with our decision, and requested against our advice that their child sit the harder paper. This request was agreed to by the headteacher against the wishes of the teacher and the head of maths. The child failed the exam.

I can't speak for how other schools decide.

In reply to Phil Payne:

It is insensitive. Actually most people already thought they wouldn't pass. They passed. So don't you tell me I'm limiting kids when I got kids who you think we shouldn't bother with a pass.

If they have a grade G they can do some of the basic maths that we need in every day life. How dare you say that's a waste of 2 years when you were the one telling me I'm limiting children? How dare you say they aren't worth teaching? You are shocked at me for daring to point out the reality that not everyone can get a C and then you come out with this.

I have nothing more to say to you on this.

In reply to baron:

I can imagine that that might work with, say, an English essay, where, with the same title, answers might vary from a few poor sentences to a brilliant work of litertaure. But, by age 16, attainment in Maths is so massively diverse that it would be ludicrous for all to sit the same paper. By sitting a paper appropriate to their attainment, pupils will be able to demonstrate that attainment.

In reply to Phil Payne: well you'd certainly not get a normal distribution then. If you lump the two papers into one big one theesults would be bimodal as good kids would simply blast the easy questions, most of the others won't dent the hard questions and there'll be a hole in the middle. And that wouldn't really kaibosh letter grades, so while I see your point, in practice it's tough to implement.

Re. Normal distribution - it might start normal, but adding teaching into the picture messes it up. Because almost all kids have maths teaching, and most learn a bit, the left hand tail (failure) is curtailed and shortened, with data (pupils) moving right to varying degrees and stretching that tail. There is a random spread, but the bigger picture is impacted by a basic level of teaching. The data looks right to me

In reply to marsbar:

To be fair, I don't really think he understands some of the key points which several of us have now tried to explain.

In reply to marsbar:

You're not listening to what I'm saying. I asked if there was a point in doing GCSE maths and getting a g grade, not whether or not it was valuable to teach them skills which will be useful in everyday life. You could still do that combined in with another vocational course, I was questioning the usefulness of a GCSE grade G.

In reply to Robert Durran:
Agreed, it wasn't a good system. It's changed now. Who'd have thought that - poorly thought out ideas being introduced into schools and then needing to be changed?

In reply to wbo:

Marsbar is the one talking specifically about GCSE maths because they have ' a PGCE and decades of teaching experience' in the subject. My original question was more general across all GCSE subjects and I'm speaking from the standpoint of a parent wanting the best for my children and wanting to know where I should expect them to be based on modern marking criteria.

In reply to Phil Payne:
Unfortunately schools no longer operate solely on what's best for the pupil's future after they leave education. (This doesn't mean that teachers don't do their best for the pupils only that they're hamstrung by the system).
There are a variety of predictors that schools use to show progress and based on a child's early tests their future is almost mapped out.
Pupils must make progress each year or the teacher is answering to the headteacher. If a child can't make a whole level of progress we sub levelled them even though those sub levels shouldn't have existed.
If a primary teacher inflates a child's SATs score the secondary teacher is stuck with it even if the score is obviously nonsense.
Even special needs pupils fall into the same system.
A school is judged on how well its pupils do based on these prediction systems.
It makes little sense to the teachers so must be baffling to anyone not directly involved in education.

In reply to Phil Payne:

As I said in my first post "modern marking criteria" is not something based on the same percentage for each subject or each paper. Getting fixated on 50% won't help you understand where they are. If your children are doing GCSE in England then the new grades are 1 to 9. The marks for each individual exam to get grades 1 to 9 will be set depending on how difficult the paper is.


https://www.cgpbooks.co.uk/gcse_grades_9_1_explained

In reply to Phil Payne:




Can anyone see a contradiction here?

In reply to Phil Payne:

What's wrong with the raw score? Then everyone knows exactly how well you've done.

In reply to Phil Payne:

When GCSEs were 'O' levels (prior to 1988), level 1 would be equivalent to GCSE grades A, B and C. A star wasn't introduced until 1994. But to get a level 1 you would need to get 70%. The exam boards felt that it was unfair to deduct grades for poor grammar etc so that was part of the reason they scrapped the O levels and introduced GCSEs and GCSEs were also introduced to incorporate non-academic based and more vocational based qualifications like GNVQs, NNEB, NVQ and other BTECS.

I teach psychology and we have a system where although we allow students with GCSE grade C in maths and English onto our A level courses, we try to make sure that within the grade transcripts students actually achieved 70% or more even though 60% is the pass rate. Quite a few schools do this, particularly grammar schools.

"because that could mean someone gets a C grade pass when in reality they got less than 50% in the actual exam.

Now to my question. Has this always been the case for grade boundaries? Maybe my memory is failing me, but I seem to remember that when I did my GCSE exams that the boundary for C was 70%. Can anyone point me to a table where I can see how these boundaries have evolved over the years, if indeed they have change"

I think it also depends on the exam subject. I.E. it's easier to get a C in English because of the marking system (and get 70 odd %) than in maths.

If you look at AQA, National Curriculum or Bite Size you can see different tables but the best one is to look at different LEAs or the office of national statistics for grade tables.

In reply to Pete Pozman:


No they don't (unless they also know how hard the exam was).

In reply to Phil Payne:

When I did O levels we were told that the exact boundaries of each grade were determined by the examining board once the results of an exam were known. I believe this went something like the % obtained by the the peak number of examinees was just a fail (or a pass...I forget). Didn't matter if it was an exactly normal distribution, and the other grade boundaries were worked out from the % distribution curve.
If this is correct then there would have been a built in allowance for variation in difficulty of the paper from year to year. It is a reasonable assumption that the average ability/laziness/intelligence of pupils across the country doesn't vary from year to year.
I do fail to see why we have so many exam boards for the same subject, though keeping questions secret in advance might be even more difficult if there was only one board.
Sorry if this has been said before as I haven't read all the responses on the thread.

In reply to oldie:


Not necessarily. Its well-known that performance on IQ tests has increased a lot since the 1930s. Google "Flynn effect" for details.

In reply to Phil Payne:

The problem as I see it is school exams in the uk exist to to serve four basic ojectives:

To meaningfully differentiate person A from Person B who sat similar exams in the same year.
To meaningfully differentiate person A from person C who sat similar exams in different years.
To facilitate a meaningful determination of system performance over time and space.
To allow exam boards to compete.

But the system we have, indeed any practical system that exists in a changing world will inevitably struggle to do this without making some compromises in one or more of those.

Personally I've never really understood why we don't just rank people's results relative to their peers in each exam year then give the results as a percentile or in percentile blocks but that's probably a comprehension failure on my part.
jk

In reply to Bob Hughes:

A trend over ***several generations*** rather backs up the assumption that the average ability/laziness/intelligence of pupils across the country doesn't vary from ***year to year***. Otherwise the "Flynn effect" would be referring to changes detected from year to year.

It would be interesting to know if the increase in IQ test results is driven by historical factors such as going from mostly leaving school at 14 to increasingly going to university, education that familiarises children with IQ-like testing, improved nutrition, pre-natal health care, parents valuing education more as manual jobs no longer available or 101 other things that have happened since the 1930s.

In reply to jkarran:


In effect that is what does happen.

In reply to jkarran:

Probably because most people don't understand percentiles? I blame the teachers