Technical Notes in Advance of the June 2017 GCSE Results

INFORMATION PAPER
Technical Notes in Advance of the
June 2017 GCSE Results
Sources of information
ASCL Information Papers: GCSE grading
With the first awarding of the 9-1 grades for the reformed English language, English literature and maths GCSEs
only a few months away, ASCL has produced a suite of information papers containing likely FAQs that are
being asked by school leaders, teachers, parents and pupils about this summer’s results.
Ofqual blogs
Ofqual continues to publish a series of helpful blogs, including:
Grade boundaries: the problems with predictions
Setting grade 9 in new GCSEs (which also looks at likely grade distributions for all grades)
This is particularly helpful because it publishes modelling by Ofqual using the 2016 results. While this modelling
is accompanied by strong ‘health warnings’, it gives a useful starting point for comparison.
Principles in setting grade boundaries
‘Anchor points’ (7/A, 4/C, 1/G)
Ofqual have stipulated from the outset that they will be setting ‘anchor
points’ between old A*-G and new 9-1 GCSEs. At a national level, broadly
the same proportion of students will achieve a grade 4 and above as
previously achieved a C and above. Similarly, the proportions achieving
grade 7 and above will be broadly the same as those achieving the old
grade A and above. The proportions achieving grade 1 and above will be
broadly the same as those achieving grade G and above.
This means, for example, that as in 2016, 16% of 16 year-olds achieved an
A or above in English and 20% achieved an A or above in maths, we can
expect these figures to be broadly the same in 2017.
An updated postcard has been published by Ofqual showing this clearly.
Grade 9
Ofqual has responded to suggestions regarding their original proposals for grade 9, and conducted extensive
research and modelling (involving Education Datalab), to produce a good (albeit complex) methodology termed
the ‘tailored approach’. This approach will mean that across all subjects (when all GCSEs are graded 9 to 1),
about 20% of those students achieving grade 7 or above will achieve a grade 9. In particular, this brings English
back into line with maths, and also achieves greater fairness with other subjects.
Arithmetic setting of grade boundaries for
grades 8,6,5,3,2
In old GCSEs, the A/B, C/D and F/G boundaries
were set by the examiners with the assistance of
statistical advice.
As a worked example, suppose in a paper out of
50 marks:
l
A/B
boundary = 42
l
C/D
= 30
l
F/G
= 12
The intermediate grade boundaries would then be
calculated arithmetically, for example:
l
B/C
is half-way between A/B and C/D = 36
l
D/E
is one-third between C/D and F/G = 24
l
E/F
is two-thirds between C/D and F/G = 18
These were also linked to UMS but this no longer
applies.
NB: this does not mean equal numbers of
candidates will get grade B and grade C as varying
numbers of candidates would be getting each raw
mark.
For example, from the Ofqual research on the 2016
results 24% gained B and 30% C in English, whereas in
maths, 21% gained B and 30% C.
We know the raw mark distribution varies from subject
to subject and from year to year for a variety of reasons,
and this process of setting grade boundaries enables
greater consistency in the meaning of a grade.
A similar process will operate with the new GCSEs, with
the same caveat that equal raw mark divisions do not
lead to equal percentages of candidates for each grade.
Note there are clear rules for situations where there
are not whole mark divisions (for example, a ten mark
spread to be divided into three grades).
www.ascl.org.uk
Grade 8
The grade 8 boundary mark will be set halfway between the grade boundary marks for 7 and 9, so the Ofqual
estimate is that in English, 5% will gain grade 8 and 9% grade 7, whereas in maths, 8% will gain grade 8 and
9% grade 7.
Grades 5 and 6
The grade 5 and 6 boundary marks will be set one-third each between the grade boundary marks for 4 and
7, so the Ofqual estimate is that in English, 15% will gain grade 6, 19% will gain grade 5 and 20% grade 4,
whereas in maths, 13% will gain grade 6, 20% will gain grade 5 and 18% grade 4.
Grades 2 and 3
The grade 2 and 3 boundary marks will be set one-third each between the grade boundary marks for 1 and 4,
so the Ofqual estimate is that in English, 23% will gain grade 3, 5% will gain grade 2 and 1% grade 1. However,
in maths, 17% will gain grade 3, 5% will gain grade 2 and 4% grade 1. This reinforces the longstanding pattern
of maths being more severely graded at lower grades than English.
“How confident can we be in these figures?”
Clearly, a major unknown is what the distribution of raw marks will look like under the new ‘challenging’
content and exams. There is some suggestion that raw marks will be lower as a result there could be more
of a clustering of raw marks than currently (although even if the content is more challenging, the questions
themselves need not be).
We can be confident in the ‘anchor point’ percentages because of the statistical linkage (and grade 9 because
of link to grade 7), but the intermediate grades are much more open to change in this first year.
Ofqual notes that: “The figures are different to those in the Education Datalab modelling, because we used a
different method. That’s why any figures should be seen as a guide and not a guarantee of what will happen.”
The Ofqual figures are slightly different but broadly similar to those presented by David Blow at the ASCL Data
conferences in February, and based on the Subject Transition Matrices.
The overall similarity of the estimates gives a useful starting point for when the actual results come out in
August, revealing what impact the raw mark distributions have had on the intermediate grades awarded.
“What will my results be like this summer?”
Although this is what everyone wants to know, it is worth reflecting a little on the question itself. What is within
your control? How meaningful and accurate can an answer be now? And what difference would this answer
make?
As ever, and especially in these uncertain times, it is worth approaching the question in different ways, to see if
they give similar answers and doing a ‘sense check’.
Perhaps the best starting point is to rephrase the question “is there any reason why the performance (in the
broadest sense) of my students will be different this year from last year?”
www.ascl.org.uk
Is their ability profile similar?
You should be aware of the national KS2 changes
from 2011 to 2012 as the raw KS2 results for your
students may be misleading otherwise.
Y11 in June 2016 took their KS2 test in June 2011
Y11 in June 2017 took their KS2 test in June 2012
Note the jump in national KS2 score, so great care is
needed regarding Progress 8.
Cumulative frequency of KS2 scores 2011 - 2015
Although the lines look close together, there is a significant shift
from 2011 with one-off partial reversal for 2013. Thus, for a fine
score of 4.6, there is nearly a 10% cumulative difference!
A pupil scoring 4.6 in 2011 would score 4.75 in 2012
A8 for 4.6 = 49.8
A8 for 4.75 = 52.7
A difference in A8 of nearly 3, so P8 changes 0.3
Key to success
The information for KS2 results (including the fine score) can be
checked by entering UPN numbers into Key to Success, but there
is no way of finding out who on Key to Success is allocated to
your school. For that you need to wait until the important…
www.ascl.org.uk
June DfE tables checking exercise
This is carried out in June of Year 11. It provides an opportunity:
l
to
check the correct pupils will be used in the unvalidated Performance Tables in October
l
to
see the official DfE data for each pupil, for example their KS2 score, did they join in the last two years,
SEN status, and ethnicity
“But they have performed poorly on the sample tests?”
This is where it is so important that Ofqual has created the ‘anchor points’. One question you can ask yourself
is what grade would you have expected those pupils to get on old GCSEs (you will have data as they have
progressed through KS3). You can then set the points of comparison using the ‘anchor points’, and go further
and use Ofqual’s figures from 2016 and estimates for 2017 to interpolate for the intermediate grades. However,
these are more uncertain as they depend on the RAW MARK distribution and no-one knows what that will be
like.
David Blow, ASCL member and Head, The Ashcombe School, Surrey
Association of School and College Leaders
130 Regent Road, Leicester LE1 7PG
T: 0116 299 1122 E: [email protected] W: ascl.org.uk Tw: @ASCL_UK
April 2017