Regulation

Summer 2023 Data Exchange Procedures in England

Updated 24 August 2023

Applies to England

The following procedures outline the data that must be provided by awarding organisations during the summer 2023 data exchange for each qualification forming part of the data exchange process.

The requirements in this document apply to GCE and GCSE qualifications and to Project Qualifications at Level 3.

This is a Regulatory Document under Condition B7 of the General Conditions of Recognition: Compliance with Regulatory Documents.

In addition –

1. a) in relation to GCSE qualifications, this document sets requirements under Condition GCSE3.1 and GCSE9.1 of the GCSE Qualification Level Conditions;

2. b) in relation to GCE Qualifications, this document sets requirements under Condition GCE3.1 and GCE9.1 of the GCE Qualification Level Conditions; and

3. c) in relation to project qualifications, this document sets requirements under Condition Project3.1.

GCE A level

GCSE outcomes that are used in A level predictions must combine A* to G grades and 9 to 1 grades onto a common scale.

Candidates who took their GCSE assessments outside England should be excluded from the predictions.

Predictions for A levels in England

For all subjects (except mathematics), each awarding organisation must create for each specification its own prediction for its specific cohort of 18-year-old matched learners. The prediction must be based on the relationship between A level outcomes in England in that subject for 18-year-old learners in 2019 and the GCSE outcomes for those learners in 2017.

Data for matched learners against predictions must exclude partial absences.

Data for all learners must include partial absences.

Predictions for A level mathematics in England

For mathematics, each awarding organisation must create for each specification its own prediction for its specific cohort of 18-year-old matched learners, excluding learners that are also certificating A level further mathematics in the same series. The prediction must be based on the relationship between A level outcomes in England in mathematics for 18-year-old learners in 2019 and the GCSE outcomes for those learners in 2017, excluding learners that also certificated A level further mathematics in 2019.

Data for matched learners against predictions must exclude partial absences.

Data for all learners must include partial absences.

GCE A level reporting

The agreed rules for setting the A*/A boundary are included in this document (Appendix 1A).

All awarding organisations must provide Ofqual with a schedule of dates (Mondays, Wednesdays and Fridays) by Monday 26 June 2023 to show when they expect to submit A level award outcomes and grade boundaries to Ofqual. Submission should be no later than 5 days after completion of the award, but outcomes can be submitted sooner where available (for example, if the award is completed on a Thursday, the outcomes and grade boundaries should be submitted no later than the following Wednesday). There may be exceptions, for example when a suite of specifications need to be reviewed together before the awards are finalised, but any such exceptions must be discussed with Ofqual in advance.

In all specifications, where a subject level boundary is chosen that does not result in a matched outcome that most closely meets but still exceeds prediction at grades A*, A and E, and there are more than 500 matched learners, awarding organisations must provide an explanation (making reference to Ofqual’s ‘Reviewing GCSE, AS, A level and Level 3 Project outcome data received from awarding organisations as part of the data exchange procedures, Summer 2023’), using Template A3 and according to the schedule provided to Ofqual.

Templates A1 and A2 must be updated each Monday, Wednesday and Friday, according to the schedule provided to Ofqual, to show interim A level results and grade boundaries to date.

All GCE A level outcomes and grade boundaries must be reported, using Templates A1 and A2, by 12 noon, Wednesday, 2 August 2023.

For GCE A level art and design, the outcomes from the suite of options must be reported as a whole.

In a few cases awarding organisations may need to carry out further analyses post-award in order to establish the security of the outcomes. If there is to be a delay in submission, awarding organisations must notify Ofqual via the data exchange mailbox dataexchange@ofqual.gov.uk as soon as possible and certainly by the date shown on the schedule.

GCE AS

GCSE outcomes that are used in AS predictions must combine A* to G grades and 9 to 1 grades onto a common scale.

Candidates who took their GCSE assessments outside England should be excluded from the predictions.

Predictions for AS in England

For all subjects, each awarding organisation must create for each specification its own prediction for its specific cohort of 17-year-old matched learners. The prediction must be based on the relationship between AS outcomes in England in that subject for 17-year-old learners in 2019 and GCSE outcomes for those learners in 2018.

Data for matched learners against predictions must exclude partial absences.

Data for all learners must include partial absences.

GCE AS reporting

All awarding organisations must provide Ofqual with a schedule of dates (Mondays, Wednesdays and Fridays) by Monday 26 June 2023 to show when they expect to submit AS award outcomes and grade boundaries to Ofqual. Submission should be no later than 5 days after completion of the award, but outcomes can be submitted sooner where available (for example, if the award is completed on a Thursday, the outcomes and grade boundaries should be submitted no later than the following Wednesday). There may be exceptions, for example when a suite of specifications need to be reviewed together before the awards are finalised, but any such exceptions must be discussed with Ofqual in advance.

In all specifications, where a subject level boundary is chosen that does not result in a matched outcome that most closely meets but still exceeds prediction at grades A and E, and there are more than 500 matched learners, awarding organisations must provide an explanation (making reference to Ofqual’s ‘Reviewing GCSE, AS, A level and Level 3 Project outcome data received from awarding organisations as part of the data exchange procedures, Summer 2023’), using Template AS3 and according to the schedule provided to Ofqual.

Templates AS1 and AS2 must be updated each Monday, Wednesday and Friday, according to the schedule provided to Ofqual, to show interim AS results and grade boundaries to date.

All GCE AS outcomes and grade boundaries must be reported, using Templates AS1 and AS2, by 12 noon, Wednesday, 2 August 2023.

For GCE AS art and design, the outcomes from the suite of options must be reported as a whole.

In a few cases awarding organisations may need to carry out further analyses post-award in order to establish the security of the outcomes. If there is to be a delay in submission, awarding organisations must notify Ofqual via the data exchange mailbox dataexchange@ofqual.gov.uk as soon as possible and certainly by the date shown on the schedule.

GCSE

All GCSE predictions must exclude independent and selective centres.

For GCSE double award, references in this document to grades 9, 7, 5, 4 and 1 refer to grades 9-9, 7-7, 5-5, 4-4 and 1-1, respectively.

Predictions for GCSE qualifications in England (full course and short course)

For all subjects, each awarding organisation must create for each specification its own prediction for its specific cohort of 16-year-old matched learners. Predictions must be based on the relationship between GCSE outcomes in England in that subject for 16-year-old learners in 2019 and key stage 2 outcomes for those learners in 2014.

Data for matched learners against predictions must exclude partial absences.

Data for all learners must include partial absences.

For each of GCSE French and GCSE German, awarding organisations must adjust their prediction as follows: +2pp at grades 9 and 7, and +1pp at grade 4.

An awarding organisation must adjust its prediction for a relevant qualification in line with any requirements specified by Ofqual to reflect the results of the National Reference Tests. Where Ofqual specifies such requirements, these will be set out in Annex A.

Tiered GCSE specifications

Each awarding organisation must use the outcomes of test equating (chained equi-percentile or nominal weights method) to support comparability between tiers at grades 4 and 5, where there are sufficient entries on each tier. The chained equi-percentile method should generally be used where there are more than 250 candidates at each tier. Where there are 250 or fewer candidates at one or both tiers, the nominal weights method could be used. For specifications with very small entries, it may not be possible to use test equating.

The agreed principles for moving the higher tier grade 3/U or 4-3/U boundary are included in this document (appendix 1B).

GCSE reporting

The agreed rules for setting the 9/8 boundary are included in this document (Appendix 1C).

Where Ofqual has specified an adjustment to predictions for a relevant qualification, including in Annex A, an awarding organisation must report against the adjusted prediction.

All awarding organisations must provide Ofqual with a schedule of dates (Mondays, Wednesdays and Fridays) by Monday 26 June 2023 to show when they expect to submit GCSE award outcomes and grade boundaries to Ofqual. Submission should be no later than 5 days after completion of the award, but outcomes can be submitted sooner where available (for example, if the award is completed on a Thursday, the outcomes and grade boundaries should be submitted no later than the following Wednesday). There may be exceptions, for example when a suite of specifications need to be reviewed together before the awards are finalised, but any such exceptions must be discussed with Ofqual in advance.

In all specifications, where a subject level boundary is chosen that does not result in a matched outcome that most closely meets but still exceeds prediction at grades 9, 7 and 4, and there are more than 500 matched learners, awarding organisations must provide an explanation (making reference to Ofqual’s ‘Reviewing GCSE, AS, A level and Level 3 Project outcome data received from awarding organisations as part of the data exchange procedures, Summer 2022’), using Template G3 and according to the schedule provided to Ofqual.

Templates G1 and G2 must be updated each Monday, Wednesday and Friday, according to the schedule provided to Ofqual, to show interim GCSE results and grade boundaries to date.

All GCSE outcomes and grade boundaries must be reported, using Templates G1 and G2, by 12 noon,Tuesday, 8 August 2023.

For GCSE art and design, the outcomes from the suite of options must be reported as a whole.

In a few cases awarding organisations may need to carry out further analyses post-award in order to establish the security of the outcomes. If there is to be a delay in submission, awarding organisations must notify Ofqual via the data exchange mailbox dataexchange@ofqual.gov.uk as soon as possible and certainly by the date shown on the schedule.

Project Qualifications

These requirements shall apply only to Project Qualifications at Level 3.

GCSE outcomes that are used in predictions must combine A* to G grades and 9 to 1 grades onto a common scale.

Candidates who took their GCSE assessments outside England should be excluded from the predictions.

Predictions for Project Qualifications

Each awarding organisation must create for each specification its own prediction for its specific cohort of 18-year-old matched learners. The prediction must be based on the relationship between Project outcomes in England for 18-year-old learners in June 2019 and the GCSE outcomes for those learners in 2017.

Data for matched learners against predictions must exclude partial absences.

Data for all learners must include partial absences.

Project reporting

All awarding organisations must provide Ofqual by Monday 26 June 2023 with the date when they expect to submit Project award outcomes and grade boundaries to Ofqual. Submission should be no later than 5 days after completion of the award, but outcomes can be submitted sooner where available (for example, if the award is completed on a Thursday, the outcomes and grade boundaries should be submitted no later than the following Wednesday).

Where a subject level boundary is chosen that does not align with the subject level boundary in the winter 2022/2023 series^{[footnote 1]} (where available), awarding organisations must provide an explanation (making reference to Ofqual’s ‘Reviewing GCSE, AS, A level and Level 3 Project outcome data received from awarding organisations as part of the data exchange procedures, Summer 2023’), using Template E2 and according to the schedule provided to Ofqual.

Templates E1 and E2 must be updated according to the schedule provided to Ofqual to show Project outcomes and grade boundaries.

Project outcomes and grade boundaries must be reported, using Templates E1 and E2, by 12 noon, Wednesday, 2 August 2023.

In a few cases awarding organisations may need to carry out further analyses post-award in order to establish the security of the outcomes. If there is to be a delay in submission, awarding organisations must notify Ofqual via the data exchange mailbox dataexchange@ofqual.gov.uk as soon as possible and certainly by the date shown on the schedule.

Sending data

Awarding organisations must confirm their data exchange contacts with Ofqual before the process begins. All data are to be uploaded to the secure Teams group.

Where data are re-run, awarding organisations must supply the re-run data to Ofqual on each appropriate template. Awarding organisations should clearly highlight any rows where data have changed.

All returns are subject to the final ratification of the awards by the awarding organisation.

On receipt of the information required by this document, Ofqual:

1. will review the data on the basis of its published procedure

2. may contact awarding organisations to ask for additional information or clarification, and will aim to do this within 24 hours of receipt

3. will hold teleconferences in July with awarding organisation technical colleagues, as necessary, to review data and explore any implications

4. will upload data to the secure Teams group in July (Fridays) showing aggregated outcomes across awarding organisations

5. will discuss outcomes with awarding organisations at the Maintenance of Standards meetings on 3 August 2023 (AS and A level) and 9 August 2023 (GCSE)

Appendix 1A

Rules for setting the A*/A boundary in A level specifications^{[footnote 2]}

1. 1. In specifications with only one entry option, and in each entry option of specifications with multiple options, the A* boundary (at subject level or at entry option level) is set so that the outcome for matched candidates at grade A* most closely meets but still exceeds the predicted outcome, provided that, in the specification or entry option:

   1. (i) there are more than 500 matched candidates and
   2. (ii) the cumulative number of matched candidates at grade A is more than 100^{[footnote 3]}.

2. 2. If, in a specification with multiple entry options, some (but not all) options fail to meet one or both of the criteria in paragraph 1 above (ie there are some Group 1 options and some Group 2 options), the following process is followed.

   1. Step 1: Calculate the weighted (by matched entry size) average of the differences between the grade A* and grade A boundaries, for each of the Group 1 entry options.

   2. Step 2: Add that average to the grade A boundary of each Group 2 entry option to obtain the grade A* boundary for that option^{[footnote 4]}.

3. 3. If, in a specification with a single entry option, or in a specification with multiple entry options, the matched entry of every option fails to meet one or both of the criteria in paragraph 1 (ie all entry options are Group 2 options), the grade A* boundary is set in each entry option as follows.

   1. (i) Where the mark width from the grade A boundary to the maximum mark is more than twice the width from grade A to grade B, grade A* is the same width above grade A as grade B is below grade A.

   2. (ii) Where the mark width from the grade A boundary to the maximum subject mark is less than or equal to twice that from grade A to grade B, grade A* is halfway between grade A and the maximum, rounded down where necessary to the nearest whole number below.

4. 4. If, in a subject with multiple entry options, the overall matched entry meets both of the criteria in paragraph 1:

   1. • the aggregate subject-level matched outcome at grade A* must be compared with the subject-level prediction;

   2. • if necessary (e.g. if outcomes are below prediction), some or all of the grade A* boundaries, as appropriate, for the Group 1 options may be adjusted to bring the subject-level outcome closer to expectation while maintaining equitable outcomes for candidates in different options;

   3. • the boundaries for Group 2 options should not normally be adjusted independently (ie they should be moved only as a consequence of adjusting Group 1 boundaries);

   4. • if there are no Group 1 options, the grade A* boundaries for the Group 2 options (provisionally set arithmetically, as in paragraph 3) should be adjusted consistently to bring the subject-level outcome closer to expectation.

For art and design see paragraph 5.

1. 5. In A level art and design, some or all endorsements may have common judgemental (A and E) boundaries. The grade A* boundary is set as follows^{[footnote 5]}.

   1. (i) For endorsements which have common judgemental boundaries (and provided that the criteria in paragraph 1 are met for the aggregate matched entry), the grade A* boundary is set so that the aggregate outcome for matched candidates at grade A* most closely meets but still exceedsthe predicted outcome for the aggregate of those endorsements based on whole-subject outcomes in the reference years.

   2. (ii) For any de-coupled endorsement(s) (ie where one or both judgemental boundaries are unique to that endorsement), and provided that the criteria in paragraph 1 are met for that endorsement, the grade A* boundary is set so that the outcome for matched candidates at grade A* most closely meets but still exceeds the predicted outcome for that endorsement based on whole-subject outcomes in the reference year.

   3. (iii) If neither the aggregate matched entry for endorsements with common boundaries nor the matched entry for any de-coupled endorsement meets the criteria in paragraph 1, the process in paragraph 3 above should be followed.

   4. (iv) If, in (ii), the criteria in paragraph 1 are not met for a de-coupled endorsement (but the aggregate matched entry for endorsements with common boundaries and/or the matched entry for another de-coupled endorsement meets those criteria), the process in paragraph 2 above should be followed, with ‘entry option’ replaced by ‘aggregate or de-coupled endorsement’.

2. 6. In all cases, if a review of statistical and/or technical evidence suggests that a grade A* boundary should be set at a different mark, the report on the award ^{[footnote 6]} must provide evidence to justify the final boundary mark recommended.

3. 7. Where the awarding data need to be re-run post-award, and the boundary recommendations reviewed, the criteria in paragraph 1 are applied to the new data; the grade A* boundary is re-calculated on that basis.

Appendix 1B

Principles for moving the grade 3/U or 4-3/U boundary on the higher tier of tiered GCSE specifications

1. 1. In tiered GCSE specifications, there is an allowed grade on the higher tier. For single award GCSEs there is an allowed grade 3, and for double award GCSEs there is an allowed grade 4-3.

2. 2. As set out in the GCSE 9 to 1 qualification level conditions, the higher tier grade 3/U or 4-3/U boundary is provisionally set arithmetically. For single award GCSEs, the higher tier grade 3/U boundary is provisionally set by subtracting half the mark interval between the 5/4 and 4/3 boundaries (rounding up half marks) from the 4/3 boundary. For double award GCSEs, the 4-3/U boundary is provisionally set by subtracting half the mark interval between the 5-4/4-4 and 4-4/4-3 boundaries (rounding up half marks) from the 4-4/4-3 boundary.

3. 3. For both single and double award GCSEs, awarding organisations must consider moving the higher tier grade 3/U or 4-3/U boundary where a review of the statistical and technical evidence leads the awarding organisation to judge that the grade 3/U or 4-3/U boundary should be set at a different mark, and the reasons for this must be recorded.

4. 4. The remainder of this document sets out some broad principles for awarding organisations to follow when the statistical and technical evidence suggests that the grade 3/U or 4-3/U boundary should be set at a different mark to the arithmetically calculated position. These principles are intended to facilitate a common approach across awarding organisations.

Principles

1. 5. When awarding tiered GCSE specifications, awarding organisations should provisionally set the higher tier grade 3/U or 4-3/U boundary arithmetically, in accordance with the qualification level conditions. This should be the recommended grade boundary unless there is compelling statistical and technical evidence to set the boundary at a different mark.

2. 6. There are several indicators that might suggest that the arithmetic position of the higher tier grade 3/U or 4-3/U boundary should be reviewed. Examples of such indicators (though there might be others) include instances where:
   1. a. There is a higher than anticipated proportion of candidates that would otherwise be ungraded on the higher tier
   2. b. The position of the higher tier grade 3/U or 4-3/U boundary raises concerns

3. 7. Where there are indicators that it might be appropriate to review the position of the higher tier grade 3/U or 4-3/U boundary, awarding organisations should investigate this further and consider the possible reasons. In doing so, consideration should be given to whether there is an issue at another grade that is impacting on the grade 3/U or 4-3/U boundary or outcomes.

4. 8. Awarding organisations should raise any emerging issues in relation to the higher tier grade 3/U or 4-3/U boundary immediately with Ofqual for discussion at the data exchange teleconferences with the other awarding organisations. If appropriate, Ofqual will schedule additional teleconferences to facilitate this. It is expected that any issue will be raised with Ofqual while the preparations for the award are taking place, rather than after any grade boundaries have been recommended.

5. 9. Ofqual will facilitate discussions between awarding organisations of any emerging issues. This might require Ofqual to collect, collate and share additional data from each awarding organisation. It is intended that, through these discussions, a consensus on the most appropriate approach will be reached.

6. 10. It is unlikely (though not impossible) that an issue would affect one awarding organisation’s specification in a given subject in isolation, and it would be difficult to justify different approaches being taken by awarding organisations for the same qualification and subject. Thus, there is an expectation that, unless there is a good reason not to, the same approach will be taken across awarding organisations in a given qualification and subject. However, a different approach for different subjects might be appropriate. In any instance, recommending a grade 3/U or 4-3/U boundary that is lower than a full width grade 3/U or 4-3/U boundary is unlikely to be appropriate.

7. 11. Where an awarding organisation recommends setting a grade 3/U or 4-3/U boundary that is not the arithmetically calculated half-width boundary, it must first discuss this with Ofqual before confirming the final grade boundaries. When considering these grade boundary recommendations, Ofqual will consider the consistency of the approach taken across awarding organisations and qualifications in the current examination series.

Appendix 1C

Rules for setting the 9/8 boundary in GCSE specifications

1. 1. In specifications with only one entry option, and in each entry option of specifications with multiple options, the grade 9 boundary (at subject level or at entry option level) is set as indicated below, provided that, in the specification or entry option:

   1. (i) there are more than 500 matched candidates and
   2. (ii) the cumulative number of matched candidates at grade 7 is more than 100^{[footnote 7]}.

Subjects first available in summer 2020^{[footnote 8]}.

For matched 16-year-old learners, the percentage of those achieving at least grade 7 (in that specification or entry option) who should be awarded grade 9 = 7% + 0.5*(percentage of candidates awarded grade 7 or above in that specification). This is known as the tailored approach.

Subjects first available prior to summer 2020

For matched 16-year-old learners, the percentage achieving grade 9 should most closely meet but still exceed the prediction.

1. 2. If, in a specification with multiple entry options, some (but not all) options fail to meet one or both of the criteria in paragraph 1 above (ie there are some Group 1 options and some Group 2 options), the following process is followed.

   1. Step 1: Calculate the weighted (by matched entry size) average of the differences between the grade 9 and grade 7 boundaries, for each of the Group 1 entry options.

   2. Step 2: Add that average to the grade 7 boundary of each Group 2 entry option to obtain the grade 9 boundary for that option^{[footnote 9]}.

2. 3. If, in a specification with a single entry option, or in a specification with multiple entry options, the matched entry of every option fails to meet one or both of the criteria in paragraph 1 (ie all entry options are Group 2 options), the grade 9 boundary is set in each entry option as follows.

   1. (i) Where the mark width from the grade 7 boundary to the maximum mark is more than 3 times the width from grade 7 to grade 6, grade 8 is set so that it is the same width above grade 7 as grade 6 is below grade 7 and grade 9 is set so that it is twice the width above grade 7 as grade 6 is below grade 7.

   2. (ii) Where the mark width from the grade 7 boundary to the maximum mark is less than or equal to 3 times the width from grade 7 to grade 6, grades 9 and 8 are set by dividing the width between the maximum mark and grade 7 by 3. Where there is a remainder of one or more marks, one extra mark is added in turn to successive intervals, starting with the highest interval (ie the maximum mark to grade 9).

3. 4. If, in a subject with multiple entry options, the overall matched entry meets both of the criteria in paragraph 1:

   1. • the aggregate subject-level matched outcome at grade 9 must be compared with the outcome suggested by the tailored approach applied at subject level, or with the subject-level prediction^{[footnote 10]};

   2. • if necessary (e.g. if outcomes are below prediction) some or all of the grade 9 boundaries, as appropriate, for the Group 1 options may be adjusted to bring the subject-level outcome closer to expectation while maintaining equitable outcomes for candidates in different options;

   3. • the boundaries for Group 2 options should not normally be adjusted independently (ie they should be moved only as a consequence of adjusting Group 1 boundaries);

   4. • if there are no Group 1 options, the grade 9 boundaries for the Group 2 options (provisionally set arithmetically, as in paragraph 3) should be adjusted consistently to bring the subject-level outcome closer to expectation.

For art and design see paragraph 5.

1. 5. In art and design, some or all endorsements may have common judgemental boundaries. The grade 9 boundary is set as follows^{[footnote 11]}.

   1. (i) For endorsements which have common boundaries at grades 7 and 4 (and provided that the criteria in paragraph 1 are met for the aggregate matched entry) the grade 9 boundary is set so that the aggregate outcome for matched candidates at grade 9 most closely meets but still exceeds the predicted outcome for the aggregate of those endorsements based on whole-subject outcomes in the reference year.

   2. (ii) For any de-coupled endorsement(s) (ie where one or more judgemental boundaries are unique to that endorsement), and provided that the criteria in paragraph 1 are met for thatendorsement, the grade 9 boundary is set so that the outcome for matched candidates at grade 9 most closely meets but still exceeds the predicted outcome for that endorsement based on whole-subject outcomes in the reference year.

   3. (iii) If neither the aggregate matched entry for endorsements with common boundaries nor the matched entry for any de-coupled endorsement meets the criteria in paragraph 1, the process in paragraph 3 above should be followed.

   4. (iv) If, in (ii), the criteria in paragraph 1 are not met for a de-coupled endorsement (but the aggregate matched entry for endorsements with common boundaries and/or the matched entryfor another de-coupled endorsement meets those criteria), the process in paragraph 2 above should be followed, with ‘entry option’ replaced by ‘aggregate or de-coupled endorsement’.

2. 6. In all cases, if a review of statistical and/or technical evidence suggests that a grade 9 boundary should be set at a different mark, the report on the award^{[footnote 12]} must provide evidence to justify the final boundary mark recommended.

3. 7. Where the awarding data need to be re-run post-award, and the boundary recommendations reviewed, the criteria in paragraph 1 are applied to the new data; the grade 9 boundary is re-calculated on that basis.

Appendix 1D

Guidance for setting the grade 1/U boundary in GCSE specifications

1. 1. The approach to setting the grade 1/U boundary in GCSE specifications will be the same as other key grades. However, given that the statistical evidence at grade 1 might be less reliable and there might be limited script evidence to review, one or more of the following methods should, where necessary, be used to support setting the mark used as a starting point for the awarders’ consideration of the grade 1/U boundary:

   1. a. Use the mark which gives a subject outcome that most closely meets but still exceeds prediction. (If there are very few candidates in the relevant part of the mark range, eg if the cumulative percentage just below grade 4 is very nearly 100%, an alternative means of determining the starting point mark may be needed, eg cognate subject information – see (b) below.)

   2. b. Consider cognate subjects. For example, the starting point for grade 1 in a separate science (where there may be few candidates) could be placed at a mark which has the same relationship to the grade 4 boundary as the grade 1-1 boundary has to the grade 4-4 boundary in combined science^{[footnote 13]}.

2. 2. Ask the awarders to review script evidence around the boundary which is used as a starting point. When reviewing the evidence awarders may wish to identify a range within which the grade 1 boundary would be ‘broadly acceptable’ and/or recommend a specific mark within that range.

3. 3. Where evidence is limited, review the proposed boundary with the other awarding organisations to ensure consistency.

Appendix 2

Principles for using the secure Teams group

1. A secure Teams group is to be used for sensitive data relating to the data exchange.

2. Email should be used as usual where no sensitive data is included or attached. All email correspondence should be sent to dataexchange@ofqual.gov.uk.

3. Where appropriate, notification alerts should be set up by individual users to notify them when folders or documents are added or changed.

4. In terms of version control, all templates must be uploaded using the date as a prefix, for example 01072023-WJEC-Template 1. Therefore, whenever a revised data template is uploaded, please create this as a new document using the date as a prefix and the time as a suffix where necessary, for example 01072023-WJEC-Template 1-8.30.

5. Old versions of templates must not be deleted from the Teams group.

6. Awarding organisation users must not create extra folders within the data exchange library of the Teams group without consulting Ofqual first.

Appendix 3

Principles for using statistical and judgemental evidence when awarding GCSE, AS, A level and Project qualifications in summer 2023 in England

Prior to the awards of GCSE, AS, A level and Project qualifications in summer 2023, an awarding organisation must ensure that persons involved in the setting of specified levels of attainment are adequately briefed in line with the requirements set out below and the relevant ‘Briefing document for awarders’ prepared by Ofqual.

Principles

GCSE, AS and A level

1. Predictions will be used to generate statistically recommended boundaries (SRBs) at subject level for the key grades (A*, A and E for A level, A and E for AS, 9, 7, 4 and 1 for GCSE), as in a typical exam series. The basis for these predictions will be the same across awarding organisations and will provide a common starting point for awarders to review candidate scripts, since this is the best mechanism for supporting alignment between awarding organisations. At all key grades, the SRBs will be the boundary marks that most closely meet but still exceed the prediction. The same approach should be taken for GCSE, AS and A level, although awarders for AS should be reminded that the cohort has changed significantly in recent years.

2. Awarders will use their expert judgement to determine whether the quality of work demonstrated at the subject-level SRBs^{[footnote 14]} is acceptable at each grade where script evidence is reviewed, taking into account the context of the summer 2023 awards. In doing this, awarding organisations should pose the following question to their awarders: ‘Is the quality of work at the subject-level statistically recommended grade boundary acceptable for the grade, given the context of this series?’. Archive scripts from summer 2019 should be used to support awarders’ judgements, where available (for phase 4 reformed subjects limited archive evidence might be available).

3. Awarders should be clear that the quality of student work at the SRBs may not align with performance from prior to the pandemic due to disruption, and that allowances should be made where performance is lower than prior to the pandemic. This is to ensure that students aren’t disadvantaged.

4. Where awarders are content that a subject-level SRB is acceptable (based on having reviewed the quality of student work within the context of the series), that boundary should be recommended. However, if the awarders are not content with the balance of outcomes across the components, they may adjust the component boundaries while keeping the subject-level boundaries at the statistically recommended positions.

5. In a subject with components with scaling factors that are unequal, adjusting one component up by one mark and one component down by one mark might not retain the same subject SRB. In this case, the change to component thresholds would be permitted where such changes ensured that the overall subject- level boundaries remained the same (ie by adjusting component boundaries by different numbers of marks), or where awarders were not content that the overall subject-level SRB is acceptable.

6. Where awarders are not content that a subject-level SRB is acceptable (based on having reviewed the quality of student work), additional script review will be required. In instances where awarders judge the quality of work indicated by the statistical prediction to be insufficient to support public confidence in the qualification, consideration should be given to scripts at marks above the SRBs. In instances where awarders judge the quality of work indicated by the statistical prediction to be observably better than prior to the pandemic, awarding organisations will need to consider whether there is additional technical evidence to consider the reliability of the predictions, the context for that subject (ie, phase of reform), and the presence (or otherwise) of effects across other awarding organisations.

7. If an awarding organisation wishes to set boundaries other than those suggested by the subject-level predictions (ie the boundaries that are as close to but also exceed the predictions), and there are more than 500 matched learners, it will need to provide additional judgemental and/or technical evidence to Ofqual to support this.

8. Ofqual will consider the outcomes of awards based on the judgemental and/or technical evidence provided, bearing in mind the aim of awarding in summer 2023 and of aligning grade standards across awarding bodies.

Project

9. Awarding must seek alignment, as far as possible, with pre-pandemic standards (this is the same approach as the winter 2022/2023 series). To facilitate this, the starting point for script scrutiny should be the grade boundaries from the winter 2022/2023 series (where available), unless there is compelling evidence to suggest otherwise. In instances where there were no awards in the winter 2022/2023 series, the starting point for script scrutiny should be the summer 2019 boundaries (where available), unless there is compelling evidence to suggest otherwise. Archive scripts from winter 2022/2023 or summer 2019, respectively, should be used to support awarders’ judgements, where available.

10. Ofqual will consider the outcomes of awards based on the judgemental and/or technical evidence provided, bearing in mind the aim of awarding in summer 2023 and of aligning grade standards across awarding bodies.

Appendix 4

Summary of data exchange templates

Template	Description
A1	A level outcomes (all candidates, matched candidates and prediction)
A2	A level grade boundaries (including by option)
A3	A level awards not as close as possible but exceeding prediction
AS1	AS outcomes (all candidates, matched candidates and prediction)
AS2	AS grade boundaries (including by option)
AS3	AS awards not as close as possible but exceeding prediction
G1	GCSE outcomes (all candidates, matched candidates and prediction)
G2	GCSE grade boundaries (including by option)
G3	GCSE awards not as close as possible but exceeding prediction
E1	EPQ outcomes (all candidates, matched candidates and prediction)
E2	EPQ grade boundaries (including by option)

Annex A

An awarding organisation must adjust its prediction for a relevant qualification in line with any requirements specified by Ofqual to reflect the results of the National Reference Tests. Where Ofqual specifies such requirements these will be set out in this Annex.

Ofqual has specified no requirements.

1. Or summer 2019 boundaries (where available) in instances where there were no winter 2022/2023 awards. ↩

2. If a specification with multiple entry options is treated as unitised for the purposes of aggregation and grading, some of the provisions below will apply to units rather than to entry options. However, as for other specifications with multiple entry options, the overall subject-level matched outcome at A* must be checked against the prediction, and adjustments made if necessary. Therefore, there will be no material difference in the final outcomes where a specification is treated as unitised. ↩

3. Entry options meeting these 2 criteria will be termed Group 1 options. The entry options not meeting the criteria will be termed Group 2 options. ↩

4. Normal rounding rules apply, except that if the calculated grade A* boundary is yy.5, it is rounded down, eg 78.5 is rounded down to 78 (but 78.51 is rounded to 79). ↩

5. This procedure should also apply in any other subjects where there is an assumption that all options will share the same judgemental boundaries. This is in contrast to the situation where options happen to share the same judgemental boundaries but there is no a priori assumption that they should do so. ↩

6. This may be a Chair’s report or a technical report. ↩

7. Entry options meeting these 2 criteria will be termed Group 1 options. The entry options not meeting the criteria will be termed Group 2 options. ↩

8. Get the facts: GCSE reform. ↩

9. Normal rounding rules apply, except that if the calculated grade 9 boundary is yy.5, it is rounded down, eg 78.5 is rounded down to 78 (but 78.51 is rounded to 79). ↩

10. The tailored approach for subjects first available in summer 2022 and the subject level prediction for subjects first available prior to summer 2020. ↩

11. This procedure should also apply in any other subjects where there is an assumption that all options will share the same judgemental boundaries. This is in contrast to the situation where options happen to share the same judgemental boundaries but there is no a priori assumption that they should do so. ↩

12. This may be a Chair’s report or a technical report. ↩

13. This means that if the grade 1-1 boundary in combined science is at the mark which is x% of the grade 4-4 boundary, then the starting point for the grade 1 boundary in the separate science should be positioned similarly. In the absence of cognate subject information, 20% may be used. ↩

14. Although many of these principles are framed around subject-level boundaries, in practice awarders may review work at component level. ↩