GSA & HMC Report on the 2015 Cambridge International Examinations (CIE) IGCSE First Language English (0500) results in GSA & HMC schools Dr Peter Mason April 2016 CONTENTS SECTION 1 INTRODUCTION & BACKGROUND 1.1 Structure of the report 1.2 Background to the problem 1.3 Executive Summary SECTION 2 ANALYSIS OF SCHOOLS’ DATA Pages 5 – 15 SECTION 3 ANALYSIS OF CIE DOCUMENTATION Pages 16 – 25 3.1 CIE briefing papers for schools on 2015 IGCSE English 3.2 CIE technical paper on 2015 IGCSE English 3.3 CIE Code of Practice 3.4 CIE’s published results on IGCSE English SECTION 4 SCHOOLS’ FEEDBACK & COMMENTS 4.1 Comments from schools’ communications with CIE 4.2 Outcomes and comments from schools’ appeals to CIE SECTION 5 FINAL COMMENTARY & EVALUATION Pages 28 – 29 SECTION 6 APPENDICES Pages 30 – 32 SECTION 7 ATTACHMENTS Pages 33 – 62 GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) Pages 2 – 4 Pages 25 – 28 page 1 SECTION 1 1.1 INTRODUCTION & BACKGROUND Structure of the Report: this report is in seven sections: Section 1 is this introduction Section 2 presents a detailed analysis of the 2015 CIE IGCSE 0500 results data received from GSA & HMC schools. Section 3 presents a detailed evaluation and critique of CIE’s own published documents relating to the 2015 IGCSE English examinations. Section 4 considers schools’ own submissions including their correspondence with CIE and reports of appeals. Section 5 is a summary commentary and evaluation of the evidence. Sections 6 & 7 include appendices and attachments. 1.2 The problem For a number of years many GSA and HMC schools have opted to enter their students for IGCSE examinations rather than the GCSE equivalent, especially in maths, English (including English literature) and the sciences. 2015 was no exception. A majority of those GSA and HMC schools that entered for CIE IGCSE English1 opted for syllabus 0500 which has speaking and listening separately endorsed, rather than syllabus 0522, which has speaking and listening as an integral part of the assessment thus contributing to the overall IGCSE grade. This element of the assessment satisfied Ofqual’s requirement for 0522 to be available to maintained schools in England and the results to be included in DfE performance tables. GSA, HMC and ASCL2 monitor their member schools’ concerns about GCSE and GCE examinations each August. Although the quality of English examinations and standards of marking have been of prominent concern to schools for some time, prior to 2015 they were not noticeably greater for IGCSE English than for GCSE. 2015, however, was very different; a very large number of schools, both maintained and independent, expressed alarm about their CIE IGCSE English results. For GSA and HMC schools this focused principally on results for syllabus 0500 and very largely, though not exclusively, at the higher grades: A*, A & B. For ASCL the focus was on 0522. Independently, CIE must also have been aware of the extent of the alarm since they report3 an increase of almost 200% in Enquiries About Results (EARs) for the two English syllabuses combined: 87% on syllabus 0500 and a staggering 225% on syllabus 0522. HMC’s policy is to write to chief executives of awarding bodies if concerns are received from at least three schools for any given examination. Concern at or beyond this threshold level is common each year but there is no recent precedent for the level of concern expressed last summer over grades awarded by CIE for IGCSE English 0500. In 2015 HMC reported4 that over 80 of its member schools had serious concerns about their results. In the light of the extent of this 1 2 3 4 This is strictly entitled First Language English but for simplicity in this report it will be referred to as English. The GCSE equivalent examination (English Language) will also be referred to as English unless there is any risk of ambiguity. ASCL – the Association of School and College Leaders represents the majority of headteachers in England’s maintained secondary schools and colleges. Data provided to HMC by CIE HMC briefing note to members 05 November 2015 GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 2 unease, and following face to face meetings with CIE staff, GSA & HMC commissioned a detailed report to receive and analyse data from schools in order to answer the questions: was there really a problem? if yes, how did it arise and what was the extent of the problem? if yes, what has been CIE’s response to the problem? If yes, what, if anything, can be done to protect the interests of the 2015 candidates and to ensure that there is no repetition of the problem in 2016? This is the report. 1.3 Executive Summary Cambridge International Examinations (CIE) offer two IGCSE English syllabuses denoted as 0500 and 0522. The syllabuses are closely linked. Syllabus 0500 is intended for overseas centres and UK independent schools; it is not accredited by Ofqual for use by maintained schools in England. Syllabus 0522 is accredited by Ofqual for use by maintained schools. In recent years growth in the number of June entries has been modest for syllabus 0500 reaching just 17,619 candidates in June 2015. Schools in membership of GSA and HMC provided a significant proportion of these candidates. By contrast syllabus 0522 has seen extraordinary growth in entries reaching 194,469 entries in June 2015. Such dramatic growth presented its own problems to CIE particularly in setting the standards for a vastly changed candidature. GSA and HMC schools did not report any major or widespread problems with the June 2014 0500 examination, nor has any evidence been received that schools that entered candidates for syllabus 0522 in June 2014 were dissatisfied with the results. CIE report that, in May 2015, as part of their post results analysis, they judged the 2014 results, in both syllabuses to have been too lenient. The decision was taken to tighten standards in the June 2015 examinations at grades A, C and E. The basis for this decision is complex and not wholly convincing. It was not conveyed to schools but CIE assert that Ofqual was kept informed. There was widespread alarm among schools when results were published in August 2015. This was true for both syllabuses. GSA, HMC and ASCL all conveyed a high level of concern to CIE. HMC records up to 80 of their member schools being affected an unprecedented number for a single examination. CIE itself reports an enormous increase in EARs compared with 2014 for both syllabus 0500 (87% increase) and syllabus 0522 (225% increase), surely sufficient evidence of schools’ widespread concerns. In response CIE published two general briefing documents and a technical document for schools and headteacher associations in an attempt to explain how they had set the standard for the June 2015 examinations. There was no admission that the standard was wrong. However, setting the standard was complex. The two syllabuses (0500 and 0522) are linked by common components: papers 1, 2, 3 and the coursework (paper 4) all of which must have a common standard between the two syllabuses. These different papers provide different pathways through the assessment with the coursework being an alternative to the GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 3 written paper 3. Also, grade C is common between core and extended tiers so the grade C standard must be aligned between these two sets of candidates. Add to this the huge rise in new candidature for syllabus 0522, a large proportion of whom (according to CIE) were C/D candidates and the complexity of all the different factors combined makes reliable standard setting exceptionally difficult. GSA and HMC schools in this study entered candidates only for the extended tier covering grades A*-E. Their concerns centred on grades at the top end, i.e. grades A* and A. Analysis of the detailed data provided by 53 GSA & HMC schools covering around 5,000 candidates entered for syllabus 0500 showed a major discrepancy in the qualification grade profile between candidates entered for papers 2 and 3 and the alternative route of papers 2 and coursework component, paper 4. The 2/4 combination was taken by the majority of candidates and produced a grade profile well below expectations though the 2/3 candidates were also affected but to a lesser extent. CIE should publish national data for each of syllabuses 0500 and 0522 showing how the grade profiles of the 2/3 candidates and the 2/4 candidates differ. Closer scrutiny of paper 2 showed a unit grade profile completely out of line with the abilities of the candidates. The problem encountered by schools could be attributed, in the main, to this paper. CIE revealed in the technical paper that the mark range between unit grade A and unit grade C on this paper was just 4 raw marks. Schools inevitably questioned whether the marking of an English paper could ensure the accuracy and precision required, especially with the large number of additional examiners recruited to mark the large number of additional entries. Schools reported that their candidates had done worse than they had predicted. More in-depth scrutiny of the results revealed that candidates had done far worse in IGCSE English 0500 than in their other subjects combined (on average) and in English literature and in History, two subjects which require similar skills to English. A further detailed comparison with all GCSE candidates nationally, and candidates from a basket of 25 selective maintained schools also showed that the IGCSE English 0500 candidates had performed much worse than would have been expected. CIE’s own Code of Practice provides for a number of checks that must be made before results are released. Schools appear to have little evidence that these checks were done. As already indicated, many schools submitted enquiries about results (EARs) requests resulting in over 2,800 grade changes for the two syllabuses (0500 & 0522) combined. A significant number of schools remained dissatisfied with EARs and therefore lodged appeals. No school’s appeal has been upheld at the date of releasing this report. Despite overwhelming evidence that there was a major problem with the June 2015 0500 examination, CIE maintain that the standards they set were correct. It is recommended that GSA and HMC further engage with CIE about the evidence in this report and, if it is felt appropriate, also with Ofqual, with the aim of achieving justice for the 2015 candidates and ensuring no repetition for candidates in June 2016. GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 4 SECTION 2; ANALYSIS OF SCHOOLS’ DATA Those GSA & HMC schools that felt they had concerns about their CIE 0500 results were asked to submit their results data for analysis. Data requested were: (a) unit grades for each unit entered, and overall (qualification) grade for each candidate (b) unit raw marks and overall raw mark for each candidate (c) estimated IGCSE English grade for each candidate (d) grades for all candidates for all subjects. By 31 January 2016 returns had been received from 53 schools covering almost 5,000 candidates. A wide range of schools were represented from highly (academically) selective city day schools to less selective rural day and boarding schools. A number of schools had chosen to pursue their concerns through CIE’s EAR process and subsequent Stage 1 and Stage 2 appeals. Schools were asked to make available their appeal submission and any subsequent CIE correspondence as part of this analysis. Comments in this report about marking are restricted to information gained from CIE itself or from schools’ EAR feedback. Data tables supporting the analyses that follow are integrated into the text for ease of reading rather than being added as separate appendices where it would necessitate a degree of cross referencing. Various detailed analyses have been carried out which are, in summary 2.1 2.1 candidate entries for the different papers and their combinations 2,2 qualification raw marks and grade profiles 2.3 unit grade profiles 2.4 consistency between the written papers (2 & 3) 2.5 comparison with schools’ estimated grades 2.6 grades compared with candidates’ average grades in all other subjects and with English Literature and History in particular 2.7 comparison with results for all schools in England 2.8 comparison with results of maintained selective schools 2.9 comparison 2.7 and 2.8 separated by different option routes (i.e. papers 2/3 or papers 2/4) ANALYSIS 1: papers (units) and their combinations Candidates entering for CIE 0500 must take papers (units) 1 & 3 or 1 & 4 (core tier, maximum grade C) or papers (units) 2 & 3 or 2 & 4 (extended tier, grades A* – E). All candidates in the schools in this analysis were entered for extended tier thus paper 1 will not be discussed in detail in this report. GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 5 Papers1 & 2 (both written) assess performance on Reading Passages. Paper 3 (written) assesses performance on Directed Writing and Composition Paper 4 is a coursework alternative to paper 3 Of the 4,903 candidates5 for whom data was available 1,202 candidates from 15 schools were entered for papers 2 & 3 3,701 candidates from 39 schools were entered for papers 2 & 4 It is not known how these proportions compared to those in the total entry. For brevity, in this report candidates will be referred to as either 2/3 or 2/4 candidates. Some schools adopted a mixed entry pattern. 2.2 ANALYSIS 2; overall (qualification) raw marks and grades Each paper has a maximum raw mark of 50. Thus the overall raw mark total for the qualification is 100 since candidates must be assessed on two written papers or one written paper plus the coursework component. Raw mark grade boundaries at unit grades A and C are arrived for papers 1, 2 and 3 by a mixture of senior examiner judgement and statistical modelling. Raw mark grade boundaries for paper 4 (coursework) are fixed year on year. The raw mark grade boundaries for the qualification as a whole are thus determined by summing the raw mark grade boundaries of papers 2 & 3 or papers 2 & 4. Table 1 below shows the 2015 raw mark grade boundaries and Table 2 gives the distributions of grades achieved in the schools in this survey, expressed as cumulative percentages. TABLE 1 Papers 2 & 3 Grade A* (notional at unit level) A B C D E F G U 5 Paper 2¥ Max 50 33 31 29 27 24 22 Paper 3¥ Qualification Max 50 Max 100 36 31 26 21 18 15 10 5 69 62 55 48 42 37 Whilst most schools provided all the information requested a small number of returns lacked some data hence the candidature of the more detailed analyses that follow be slightly less than this total of 4,904. GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 6 Papers 2 & 4 Grade A* (notional at unit level) A B C D E F G U Paper 2¥ Max 50 Component 4 Qualification Max 50 Max 100 33 31 29 27 24 22 50 44 38 32 26 20 14 8 83 75 67 59 50 42 ¥ These strictly apply to the papers that are denoted by CIE as paper 2.1 and paper 3.1 which are the papers taken by schools in this report. TABLE 2: actual grade distribution: cumulative percentages Papers/components A* A B C D E U entry 2&3 42.3% 69.1% 88.4% 97.8% 99.5% 99.8% 100% 1,202 2&4 16.3% 50.6% 81.2% 95.2% 99.5% 99.9% 100% 3,701 There is clearly a problem, though the huge disparity, (especially at grades A* and A), between the two separate entries might suggest that the problem is solely that of the 2/4 candidates entered. But that is not the case; schools that entered candidates for papers 2 & 3 also expressed serious concern at their candidates’ outcomes. CIE should be asked to provide the same data for the whole entry of 0500 candidates to see if this disparity is a feature of all entries. 2.3 ANALYSIS 3: unit grade profiles: is any one paper or component responsible for the grade profiles above? Table 3 below provides the details. TABLE 3: paper/component grade distributions: cumulative percentages Papers/components A* A B C D E U 2 (written) 33.1% 48.0% 62.6% 75.9% 90.3% 94.7% 100% 3 (written) 50.6% 78.1% 95.6% 99.4% 99.7% 99.9% 100% 5.8% 62.7% 91.1% 99.4% 100% 100% 100% 4 (coursework) Note: for individual papers/components the A* grade is a notional grade. The (notional) A*/A boundary is derived by extrapolating the grade A/B boundary by the same number of raw marks as the A – B interval. It is worth noting at this point that for component 4 the (notional) A*/A boundary is the maximum raw mark available (50 marks). Brief observations are useful here. The paper 3 profile is more characteristic of the overall grades expected from candidates at the schools which used the 2/3 pattern of entry. A greater proportion of those schools were highly (academically) selective. The full mark range was used and marks above the (notional) A*/A boundary compensated for any (apparent) underperformance on paper 2. GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 7 The component 4 (coursework) profile is not unusual at cumulative grades A*-C. Schools with well-motivated students and largely experienced and stable English staffing will prepare their students well for the coursework assignment. It should be expected that all these candidates would achieve at least grade C. However, the (notional) A* and cumulative A*/A percentages (5.8% and 62.7% respectively) are much lower than would be expected, a result, in part, of candidates needing to gain full marks on this component in order to achieve the (notional) A* grade. The paper 2 profile above is an amalgam of the two different entry populations: the 2/3 candidates and the 2/4 candidates. Disaggregating the two populations gives greater insight as detailed in Table 4 below. TABLE 4: paper 2 grades disaggregated Papers/components A* A B C D E U 2/3 candidates 32.8% 47.5% 61.2% 75.3% 89.3% 93.6% 100% 2/4 candidates 33.2% 48.2% 63.1% 76.1% 90.6% 95.1% 100% There is very close alignment suggesting that the ability profile of the two cohorts is very similar. Although the alignment is close, what is particularly alarming is that only marginally over three quarters of the candidates achieved unit grade C or above on paper 2. Given that the paper 3 profile more accurately represents the expected qualification grade profile it is safe to conclude that (a) (b) 2.4 paper 2 is responsible for lowering the qualification grade profile for the 2/3 candidates paper 2 and component 4 together have a combined effect of significantly reducing the qualification grade profile for the 2/4 candidates especially at grades A* and A. ANALYSIS 4: consistency between papers 2 & 3 Paper 2 caused schools greater concern than paper 3. Some schools reported that (in their view) the marking on paper 2 had been erratic. Whilst scrutiny of marking is not part of this study a simple comparison of paper 2 and paper 3 marks for each candidate will reveal whether there has, in fact, been a problem. Erratic marking usually leads to scrambling of rank order; the best candidates don’t necessarily get the highest marks; there is inconsistency. By contrast severe marking, whilst lowering the marks awarded, tends to do so for all candidates and retains the rank order. How do the marks on papers 2 and 3 correlate? The chart below displays the extent of correlation between candidates’ paper 2 and paper 3 marks. The correlation coefficient = +0.435, i.e. the correlation can be described as fair, at best. If the two papers are assessing similar or related skills and abilities (and it would be peculiar if that were not the case) then a much higher correlation coefficient would be expected. GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 8 2015 CIE 0500: Paper 3 marks v Paper 2 marks 50 45 Paper 3 raw marks 40 35 30 25 20 15 10 10 15 20 25 30 35 40 45 50 Paper 2 raw marks How do marks on paper 2 correlate with those on component 4? The distribution of marks on component 4 is very different from that of paper 2 (see appendices 2 & 3, pages 30 & 31) therefore no reliable correlation can be derived. What role do UMS scores have in determining the overall (qualification) grade? Fortunately, and unlike many unitised examinations, CIE does not convert unit raw marks to UMS scores before determining the (overall) qualification grade. The qualification grade is determined from the total raw mark scores as shown in Table 1 above. That these raw mark totals can be converted to total UMS scores is of nil effect. It seems strange therefore that in providing feedback to centres of changes resulting from EARs, CIE reported mark changes as UMS, not raw marks. Schools were understandably confused; what they really needed to know, and could readily understand, was how the raw mark(s) had changed. What role do paper/component grades have in determining the overall (qualification) grade? Paper grades play no part in determining the qualification grade but the fact that they are reported led (in part) to schools’ alarm and reinforced their view that something was seriously wrong with the 2015 examination. Tables 5 & 6 below give the details. GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 9 TABLE 5 Grade pairs on papers 2 & 3 respectively AA 368 BA 101 CA 100 DA AB 20 BB 20 CB 34 DB AC 3 BC 4 CC 5 DC AD 1 BD 2 CD 0 DD AE 0 BE 0 CE 0 DE AF 0 BF 0 CF 0 DF AG 0 BG 0 CG 0 DG AU 0 BU 0 CU 0 DU 86 35 4 1 0 0 0 0 EA EB EC ED EE EF EG EU 21 21 2 0 1 0 0 0 UA UB UC UD UE UF UG UU 38 18 6 1 1 0 0 0 182 147 82 6 0 0 0 0 EA EB EC ED EE EF EG EU 53 59 26 3 0 0 0 0 UA UB UC UD UE UF UG UU 31 60 47 8 0 1 0 0 TABLE 6 Grade pairs on papers 2 & 4 respectively AA 968 BA 231 CA 188 DA AB 258 BB 123 CB 156 DB AC 36 BC 41 CC 47 DC AD 2 BD 1 CD 3 DD AE 0 BE 0 CE 0 DE AF 0 BF 0 CF 0 DF AG 0 BG 0 CG 0 DG AU 0 BU 0 CU 0 DU Across the two options (2/3 and 2/4) over 200 students were ungraded in paper 2. Many of these students would never have achieved less than grade A in any examination they had ever taken. It is unsurprising that schools, candidates and parents were alarmed and challenged the results. 2.5 ANALYSIS 5: overall (qualification) grades compared with schools’ estimates Schools’ first action when something appears to be wrong with examination results is to go back to their own forecast grades. Such forecasts are not flawless but they do represent the judgement of experienced professionals about the expected performance of their students. They are an important check, so important in fact that CIE’s Code of Practice (CoP) uses schools’ forecast grades as an early indicator of problems with results [see CIE CoP, paras 5.3(f) and 5.8(b)]. From the schools’ returns in this study 3,913 matched actual/estimated grades were obtained covering data from 43 schools. The analysis is shown below in Tables 7a & 7b below. TABLE 7a: actual (qualification) grades compared to schools’ estimated grades. Grade profile A* A*/A A*-B A*-C A*-D A*-E A*-U Av subject pts score Actual (cum %) 24.2% 58.0% 85.2% 97.3% 99.8% 99.9% 100% 6.64 Estimated (cum %) 37.1% 71.7% 93.5% 99.6% 100% 100% 100% 7.02 GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 10 CIE 0500 actual & estimated grade profiles (cumulative percentages) 100% 90% 80% cumualtive % 70% 60% 50% Actual 40% Estimated 30% 20% 10% 0% A* A*/A A*-B A*-C A*-D A*-E A*-U grade range Schools’ concerns are largely focused on the top grades (A* and A*/A), where there is significant discrepancy between estimated and actual grades, though even at A*-B there is a difference of over 8 percentage points. More in-depth analysis helps to identify the main source of the problem. TABLE 7bs Grade difference (actual – estimate) % of results +3 0.1% +2 1.6% +1 11.6% +0 43.0% -1 35.1% -2 8.2% -3 0.4% -4 0.0% Almost 88% of actual grades either match or are within one grade of the estimate. However, it should be noted that there are over three times as many actual grades that are one grade below the estimated, than are above. One in 12 actual grades are two grades below the expected. GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 11 Further analysis breaks these data down so that the accuracy for each estimated grade can be considered. Table 7c gives the details. TABLE 7c Grade difference (act – est) est A* est A est B est C +3 0.84% +2 +1 4.9% 8.37% 15.9% 19.6% 29.71% +0 47.4% 40.0% 40.9% 41.42% -1 40.7% 35.4% 29.3% 19.67% -2 11.2% 8.1% 5.3% -3 0.7% 0.5% -4 0.0% The apparently poor accuracy of estimated grades compared to actuals has been criticised by UCAS though they analysed estimates at A-level submitted by schools as part of students’ UCAS applications, many of which were submitted 6 – 8 months before the students actually sat their A-level examinations. They were not estimated grades in the same sense as analysed here; they are best considered as potential grades. Even so, UCAS estimates from independent schools were shown to have the greatest accuracy, a result no doubt of many independent school students being on target for A* or A grades where accuracy of prediction was much greater than in the middle range of grades. 2.6 ANALYSIS 6: overall (qualification) grades compared with candidates’ grades in other subjects The second comparator that schools use when results in a given subject seem awry is comparison with candidates’ results in other subject. This takes two forms: (a) comparison of the subject’s results with results in all the candidates’ other subjects and (b) comparison of the subject’s results against those in similar subjects. In this analysis the similar subjects used are English literature and History. (a) comparison of schools’ English results with candidates’ results in all other subjects A large body of data was available as shown in Table 8 below. TABLE 8 grade distributions of results in this study expressed as cumulative percentages: English and all other subjects except English. A* English IGCSE (CIE 0500) All subjects except English A*/A A*-B A*-C A*-D A*-E ENTRY Av subject pts score‡ 27.4% 61.0% 86.6% 97.5% 99.8% 100% 4836 6.72 48.7% 76.2% 91.6% 97.8% 99.5% 99.9% 43293 7.14 ‡ for the purpose of this analysis the average subject points score uses A* = 8, A = 7, B = 6, etc GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 12 The residual is defined as the average subject points score for English minus the average subject points score for all other subjects (except English) taken by the same candidates. This equates to 6.72 – 7.14 = -0.42, i.e. on average students’ grades were almost half a grade worse in English than in their other subjects. This, plus the huge disparity at grades A* and A*/A, makes schools’ anxieties fully understandable. (b) comparison of schools’ English results with candidates’ results in English literature and in history TABLE 9: grade distributions expressed as cumulative percentages: English, English literature and History A* A*/A A*-B A*-C A*-D A*-E ENTRY Av subject pts score English IGCSE (CIE 0500) 27.4% 61.0% 86.6% 97.5% 99.8% 100% 4836 6.72 English literature 49.7% 76.9% 92.8% 98.5% 99.7% 100% 4776 7.18 history 49.9% 79.9% 94.4% 98.3% 99.5% 99.9% 2915 7.22 entries in these subjects depend on the schools’ own entry policies. They may be GCSE or IGCSE; no distinction is made here. Again, the data speak for themselves. On average, results in CIE 0500 English are half a grade worse than the same candidates gained in English literature and in history. 2.7 ANALYSIS 7: is achievement in English nationally worse than achievement in all other subjects (on average), worse than in English literature and History in particular, and to the same extent as seen in the above grade profiles? Table 10 (below) provides the details. TABLE 10: all schools in England: 2015 JCQ GCSE data – cumulative percentages A* A*/A A*-B A*-C A*-D A*-E ENTRY Av subject pts score English 3.10% 14.40% 37.10% 65.30% 86.40% 94.40% 459027 4.97 all subjects except English English literature 6.97% 21.69% 43.73% 69.17% 85.18% 92.74% 4376685 5.15 4.80% 21.50% 50.50% 75.10% 89.60% 95.80% 399281 5.35 History 9.50% 28.60% 50.40% 68.80% 81.60% 89.70% 227619 5.22 Whilst it is true that results in English are lower than the other comparators, the differential, especially at the top grades is nothing like that in this CIE 0500 analysis. The problems identified in this study can not be explained by attributing them to a national trend. 2.8 ANALYSIS 8: comparison with results from candidates in selective maintained schools (grammar schools) It can be argued that the ability profile of students in the GSA & HMC schools in this study is skewed significantly towards the more able. Given that English is a subject taken at GCSE (or IGCSE) by the whole population of 16 year olds, the English results shown in table 11 above may GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 13 not be a fair comparison. A more relevant comparison would be the results from candidates in selective maintained schools. These are shown in Table 11 below. TABLE 11: 2015 GCSE (IGCSE) results from a sample of 25 selective (grammar) schools in England (cumulative percentages) A* A*/A A*-B A*-C A*-D A*-E ENTRY Av subject pts score English 24.3% 63.1% 92.7% 99.2% 99.9% 100% 3568 6.79 all subjects except English 35.0% 70.2% 91.4% 98.3% 99.7% 99.9% 35368 6.96 English literature 27.8% 69.0% 93.9% 99.3% 99.9% 100% 3562 6.90 History 34.9% 71.8% 91.0% 97.5% 99.1% 99.8% 2023 6.94 results from this analysis were taken from the schools’ own websites Three key points emerge. (i) (ii) (iii) 2.9 This population of candidates more closely matches that in this CIE 0500 study By cumulative A*-B selective schools’ results in English are very closely aligned with those of all other subjects. The differentials at A* and at A*/A between English and the other comparators are much smaller than for the CIE 0500 students (table 9). ANALYSIS 9 How do the grade profiles in these subject pairs comparisons change if results are analysed separately according to options taken, i.e. the 2/3 candidates and the 2/4 candidates? Tables 12a and 12b, below, have the details. TABLE 12a: grade distributions expressed as cumulative percentages for the 2/3 candidates A* A*/A A*-B A*-C A*-D A*-E ENTRY Av subject pts score English 46.1% 75.5% 91.7% 98.9% 99.6% 99.9% 1194 7.11 all subjects except English 52.8% 77.1% 91.5% 97.9% 99.7% 99.9% 9458 7.19 English literature 52.8% 79.6% 94.1% 99.0% 99.7% 100% 1179 7.25 History 53.5% 83.2% 94.0% 98.0% 99.4% 99.9% 804 7.28 GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 14 TABLE 12b: grade distributions expressed as cumulative percentages for the 2/4 candidates A* A*/A A*-B A*-C A*-D A*-E ENTRY Av subject pts score English 18.6% 54.3% 84.2% 96.9% 99.9% 100% 3437 6.54 all subjects except English 45.4% 74.3% 90.9% 97.5% 99.5% 99.9% 30713 7.07 English literature 47.1% 75.0% 92.0% 98.3% 99.7% 100% 3392 7.12 History 47.5% 78.1% 94.5% 98.4% 99.6% 99.9% 2033 7.18 The differentials in the 2/3 candidates’ results (table 12a) are more consistent with those of the grammar schools’ candidates. By contrast those of the 2/4 candidates (table 12b) are so far out of line at A* and A*/A that a school could almost be accused of being negligent if it was not alarmed. Interim conclusion The analyses above have involved dissecting and objectively comparing the results of several thousand GSA & HMC candidates in the 2015 CIE IGCSE (0500) English examination. On many fronts the results are significantly out of line with legitimate comparators especially for the candidates taking papers 2 & 4. The immediate conclusion is that there appears to have been a major problem with the examination. Before stating anything more definite it is important to assess CIE’s commentary on the examination. This assessment follows in section 3 below. GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 15 SECTION 3: ANALYSIS OF CIE’S DOCUMENTATION How has CIE responded to date? CIE made 3 explanatory documents available to schools: (i) Cambridge IGCSE First Language Briefing for school groups (attachment 1) (ii) Supporting document on IGCSE First Language English – setting the standard, marking and grading (attachment 2) (iii) Technical Briefing Paper: IGCSE First Language English in June 2015 (October 2015)ψ (attachment 3) accessed from www.ascl.org.uk/2015exams . These papers were produced following issues surrounding the quality of marking with the 2015 summer CIE 0522 English GCSE exam, ASCL had been in discussion with CIE about the problems experienced by schools in 2015 as well as seeking reassurance that similar issues will not recur in 2016. As a result, CIE produced these papers outlining their awarding process for the summer exam. Ψ document sent to HMC following discussions with CIE about IGCSE English (0500) There are three further sets of relevant documents published by CIE and available from the CIE website (www.cie.org.uk) (iv) CIE’s Code of Practice (see relevant extracts at Appendix 4) (v) CIE’s grade thresholds (June & November 2014 & 2015) (vi) CIE’s results documents (June 2012 – June 2015) CIE’s Code of Practice (para 1.59a) states that CIE will publish a report to centres from the senior examiners for each externally examined component after each examination session. However a search of the CIE website on 10 March 2016 revealed only the June 2014 report for specification 0500 and a statement that there are no examiner reports for syllabus 0522. [see http://www.cie.org.uk/programmes-and-qualifications/cambridge-igcse-english-first-language0500/past-papers/ and http://www.cie.org.uk/programmes-and-qualifications/cambridgeigcse-english-first-language-uk-0522/past-papers/ ]. The apparent absence of the June 2015 senior examiners’ report for syllabus 0500 in which a number of the concerns expressed by schools as recorded in this report could have been addressed is an opportunity missed by CIE. It also appears to contravene their Code of Practice. (vii) EAR data for syllabuses 0500 & 0522 (CIE private communication to HMC) – attachment 4 Key points made by CIE in their documents It is relevant to note that CIE took the very unusual step of producing 3 documents to try to address schools’ concerns about CIE IGCSE English in 2015. That they did this confirms that they were aware of a problem at a relatively early stage and that they took steps to head it off. 3.1 Documents (i) and (ii) (see attachments 1 & 2 on pages 33 - 39): although not specified, these documents refer principally to CIE 0522 which was Ofqual approved for use by maintained schools in England. Key points to note are CIE produced the documents in response to “enquiries received from schools on a number of different issues”. This suggests that there was an increased volume of enquiries from schools in 2015. GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 16 CIE emphasise that they had ensured the standard of IGCSE syllabuses (does this imply syllabus 0500 also?) are comparable with the GCSE equivalents. CIE state that the same grading procedures had been employed as in previous years. CIE state that in ensuring standards comparable with previous years a small tightening of the standard at grade A and at grade C was needed. “We tightened by about 1 mark at grade A and by less than a mark at grade C”. Assuming this refers to raw marks rather than UMS, it is not clear what about 1 mark means or how the grade C standard can be tightened by less than a mark. CIE assert that the decision (to tighten) was proposed in May 2015 and approved (by whom?) in June 2015 based on retrospective comparison with the 2014 results. There is no evidence that this decision was conveyed to schools. CIE assert that 2.3% fewer candidates were awarded grade A (in 2015) than would have received grade A in 2014. At grade C the equivalent percentage was 1.7%. CIE assert that most schools’ results were “broadly similar to last year” but acknowledges that some schools achieved much higher results, some much worse results. This statement seems to suggest that 2015 results presented a similar pattern to previous years which is at odds with the evidence. 2015 obviously generated much greater concern than previous years otherwise there would not have been a challenge by ASCL and by GSA & HMC. CIE acknowledge the significant (60%) growth in the entry (for 0522) which resulted in the need for recruiting, training and monitoring a large number of new examiners. Whilst they assert that these processes were carried out to a high standard, this is bound to raise suspicions. Poor, inconsistent and erratic English marking has been the concern of schools for many years. To suggest that recruitment of such a large number of new examiners had no impact on the quality of marking stretches credibility. CIE report an increase in EARs but that the proportion of grade changes was in line with previous years. The increase in EARs is not quantified but is indicative of the level of schools’ concerns. The number of grade changes (obviously increased if the proportion has stayed the same) may simply reflect a very tight marking review régime imposed by CIE which mitigates against changing marks and hence changing grades. For example, one school reported a student being judged to have achieved one mark extra on paper 2 following an EAR which would have resulted in the student gaining overall qualification grade A* rather than A. However, the one mark was within the tolerance set by CIE so the original mark stood; the student was denied the A*. Given that A* in English at (I)GCSE is a requirement for some very competitive university courses this could have far reaching consequences for that student. With respect to subject (syllabus) pair comparisons, something that schools are quick to refer to, CIE state that they ensure consistency at cohort level but there will be cases where individual candidates achieve different grades in English language and English literature. But it is not clear how CIE accessed all candidates’ English literature results in order to do such comparisons. The evidence from the GSA & HMC schools’ results (collectively) in this study suggests that there is a big discrepancy between the candidates’ English language (CIE 0500) and their English literature results. That was the cause, in part, of schools’ concerns. GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 17 CIE refers to volatility in individual schools’ results as being entirely normal. It is true that Ofqual’s research: Variability in GCSE Results for Individual Schools and Colleges 2012 to 2015 (Ofqual/15/5767; August 2015) analyses such variability (volatility) but that doesn’t justify such variability. HMC has, for a number of years, presented results data to Ofqual from schools with very stable staffing and student cohorts where results have inexplicably varied massively from year to year. That Ofqual have found this to be true does not justify its happening. CIE reports that Ofqual were kept informed of the intention to tighten standards, and Ofqual duly reported (see attachment 8) that they had monitored CIE IGCSE English, partly because of the large increase in entry from maintained schools in recent years and especially in 2015. Ofqual’s report, though not specific about CIE 0522 must surely refer only to this specification since it is the only one that comes within Ofqual’s regulatory jurisdiction. However, Ofqual’s brief report is not totally consistent with the CIE documents. In particular Ofqual states that CIE informed them in early 2015 (my emphasis) that their (CIE’s) routine analysis had identified some leniency in grading IGCSE English in summer 2014. But CIE report that it was May and June 2015 (see above). Ofqual states that CIE told them, as a result, it intended to tighten its grade standards at Grade C and, to a lesser extent (my emphasis), at grade A in summer 2015. But we know from the CIE reports that grade A was tightened more than grade C. Ofqual acknowledges (somewhat sympathetically) that CIE faced particular challenges in 2015, not just because of the large increase in entries, many from centres that had not entered for IGCSE English before, but also because many of the additional entrants were grouped disproportionately around the C/D borderline. But Ofqual is the regulator and should be concerned with how awarding bodies ensure the maintenance of standards when such perturbations in entry occur. Ofqual required CIE to provide evidence on how they had come to their grading decisions and report that CIE had used cohort predictions based on KS2 results (socalled comparable outcomes) and comparisons of the results of “benchmark centres”, i.e. schools that had results in both 2014 and 2015. Ofqual further report that CIE informed them that evidence from the benchmark centres suggested that tightening 2015 standards to the extent that KS2 predictions required would have been too severe. But the CIE reports make no reference to benchmark centres, and give the reason for the tightening of standards as post hoc comparison with 2014. Finally Ofqual report that not only was there bunching of candidates around the C/D borderline but there was a significant bunching of marks around the C/D boundaries. That is a result of the examination and mark scheme, not the candidature. Ofqual report that on one paper (it is actually paper 2), the difference between grade C and grade A was only 4 (raw) marks. It is astonishing that Ofqual seemingly fails to challenge this or analyse it further but says, rather meekly: “we concluded that Cambridge International had carried out its grading appropriately. Cambridge International has carried out further analysis since the summer and is confident that their grading was appropriate”. We should note that this Ofqual report was being written at a time when there was unprecedented concern about the CIE English IGCSE results from schools, supported by three national headteacher associations. GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 18 3.2 Document (iii) (attachment 3) is a longer and more detailed technical briefing paper which, according to the paper itself, was made available to headteacher associations and to Ofqual. It covers the general ground of the two shorter documents (i) and (ii) and adds more quantitative detail. Important additional points are: A variety of analytical tools were deployed to try to align the 2014 and 2015 standards. These included (a) determination of grade profiles from matched KS2 results (comparable outcomes) and (b) post-award screening which uses the average grade achieved by a candidate in all of their GCSE subjects taken in the same exam series. CIE reports that the screening method is usually considered (by whom?) to be more robust than the KS2 method. The screening method plus a third analysis (subject pairs) suggested that June 2014 had been leniently graded in 2014 especially at grade C. The report attributes the differences between the KS2 predictions and the screening (in the 0522 entry) to an atypical ability profile skewed towards grades C and D. CIE informed Ofqual in May 2015 that they judged KS2 predictions as being systematically lenient and would therefore aim to be 2% more severe (than matched candidate predictions) at grades A, C and E. In the event this was revised to 1% for grade E but retained at 2% for grades A and C. The report describes two further preliminary analyses: one (for 0500) estimating the effects of any changes in the entries from different countries, the second using forecast grades supplied by schools which the report says was of limited value for 0522 (but presumably was of use for 0500). The report does not give details of these analyses In addition to the significant change in candidature for 0522, the report makes reference to the difficulties associated with standardising the multiple routes through the qualification (an option to do written paper 3 or coursework component 4) but aligning standards at the common grade C through papers 1 and 2 and ensuring that standards in the inextricably linked 0500 and 0522 are aligned when their candidatures are so different. Even a mildly informed observer might consider this an impossible task; it is certainly exceptionally difficult. Further, since grade boundaries are fixed for the coursework component (paper 4) year on year the only way of influencing outcomes is the use of papers 1 or 2 and since the primary objective seems to have been tightening the grade C standard (because of the large influx of candidates presumably viewing grade C in CIE English as being more accessible than on other GCSE English examinations), there was always bound to be a knock-on effect at grade A. The report helpfully discloses the difficult task of fixing the raw mark grade boundaries for grade C on papers 1 and 2 in order to tighten standards. Raw mark C/D boundaries of 34/50 (paper 1) and 27/50 (paper 2) were agreed; at grade A (on paper 2) the A/B boundary was set at 31/50, resulting in a marginally smaller tightening of the 2014 standard than had been planned. However, the report continues (para 6.5.6): “the agreed grade threshold still produced large drops in outcome in the benchmark baskets for the options including components 2 and 4 and these were unexpected in the sense that they were not matched by drops in forecast grades. There was also an appreciable drop in the overall percentage of UK candidates achieving grade A or above on syllabus 0500; this was not seen among non- GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 19 UK centres.” Unfortunately, the term “appreciable drop” is not quantified but this is absolutely crucial importance to the issues behind this report. The latter comment re non-UK centres may be a result of paper 3 being popular outside the UK. In fact, given the tightening on Paper 2 at grade A, CIE decided not to tighten the grade A on paper 3 “to avoid creating an unnecessarily large adjustment to last year’s standard for those (mostly overseas) centres which had taken papers 2 and 3 together.” The report comments that paper 3 continues to be slightly easier than paper 4 at grade A. Grading decisions were formally approved (presumably by the accountable officer) in August 2015. The grade outcomes are shown graphically (p10 in the report; see page 49) and confirm that 0522 grades were heavily concentrated around grades C and D as previously mentioned. However, the report states that the entry for 0500 was heavily concentrated at the top grades “so that a movement of a single mark in the grade A threshold can have a significant impact on the percentage of candidates achieving the grade”. This statement is again of critical importance when considering the problems encountered by the GSA and HMC schools in this study. EARs and appeals: the CIE report correctly points out that schools have the opportunity to make an enquiry about results (EAR) if they have concerns and, if still dissatisfied, schools may appeal. The level of unease at this year’s results, already commented on above is clear from the CIE report which says: “we have seen an unusually large number of EARs as well as contacts from centres that feel that their results were lower than expected”. That the CIE report does not express alarm about this is astonishing; instead it attributes the dramatic rise in schools’ concerns, in part, to the increase in entries alongside a wider trend for increases in EARs. HMC reports an unprecedented level of concern among its member schools which entered candidates for 0500. The further option of appeal is open to centres who remain dissatisfied with the EAR process but, as the CIE report explains, such appeals are only for situations where schools believe that CIE has not followed its processes properly or has failed to abide by its own Code of Practice. Appeals based on a disagreement with CIE’s judgement are rejected. Schools in this study report very few changed grades resulting from an EAR; not one reports a successful appeal. Section 9 (p11) of the CIE report lists the findings of further analyses by CIE as part of their post-results checks. The initial bullet point in this section: “we see no evidence of any reduction in the reliability of the components compared to June 2014” hardly inspires confidence in centres who feel that their candidates have been treated unfairly. Schools would argue that there is plenty of evidence as shown in this report. The post-results checks show that candidates entered for a combination that included coursework (paper 4) suffered a greater tightening of standards. However the most revealing statement comes in bullet point 5 of this section: “the effect of the tightening at grade A has been greater on 0500 candidates from the UK because of the concentration of candidates close to the grade A threshold. Changes of a single mark in a threshold can have a disproportionate impact on schools where the ability profile is untypical of the wider cohort (specifically, where many of the candidates are to be found close to the threshold. As such, the combination of the tightening at GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 20 grade A and the ability profile of the schools largely explains the drop in the percentage of candidates achieving grade A in 0500 within large stable centres in the UK” This is exactly the experience of schools in this study, but it is a post hoc discovery by CIE expressed with almost a degree of surprise. There is no disclosure of the extent to which the tightening has had a greater effect on 0500 UK candidates, no attempt to assess whether that tightening was fair on those students, no attempt to explain whether this greater tightening effect was intended and consistent with CIEs attempts to better align the 2015 standards with those of 2014, no details or confirmation that the prospective 0500 results of any of the schools in this study were scrutinised and checked before results were issued. Given that CIE admit that the schools at high risk of disproportionate tightening at grade A were the 0500 UK schools, of which the schools in this study must have made up a large proportion, it must have been apparent before results were released that the prospective results differed significantly from those forecast by the schools. CIE’s Code of Practice (para 5.8(b)) requires CIE to check a sample of results from such centres before results are issued. CIE should produce written evidence that this was done. 3.3 Enquiries about Results (EAR) data (attachment 4) CIE provided HMC with the number of EARs for candidates in the June 2014 and 2015 examinations. The data make alarming reading showing an increase of 125% across all subjects from 20,394 (2014) to 45,984 (2015). Of these, syllabuses 0500 & 0522 combined accounted for 30,768 EARs (an increase of 197% on 2014), 67% of the 2015 all syllabuses total. Such was the volume of EAR requests that 7,700 (17% of the total EARs received) could not be completed within the planned 30 days. Whilst it is accepted that there was an increase (66%) in candidature (0500 & 0522 combined) between 2014 and 2015 this does not explain the enormous increase in EAR requests. These data alone must have indicated to CIE that there had been a serious problem with the results but very surprisingly, the percentage of grades that changed following an EAR was not significantly greater than that for all other subjects (9.3% compared to 9.0%). Contrast this with Ofqual’s data where the percentage of GCSE grades changed following EARs was 18.0% (see the reference in the footnote on page 54). 3.4 The CIE Code of Practice – what does it require? – see the extracts in attachment 5 CIE’s own Code of Conduct (CoP) sets both the standards and procedures it must follow in administering its examinations. (see CoP para 1.2(i)). Failure to comply with the CoP would present strong grounds in any centre appeal. Each paragraph in the CoP is important but comments here will be reserved to those that are most relevant to the issues in this report. Paras 1.3(e) and 1.3(f) give assurance about transparency and a clear audit trail of decisions taken, why, by whom, and to whom those taking the decisions were accountable. Schools that took their EAR through to Stage 2 appeal do not report that this was their experience. In fact they report that they had to be persistent to get answers to routine questions. Para 4.7(d) confirms that marks will not be changed following a marking review if the senior examiner performing the review of marking is within the pre-established tolerance. Whilst this is common practice by all awarding bodies there is a particular issue with the tolerance on paper 2 where the reviewing senior examiner’s mark had to exceed 2 raw marks before the original GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 21 mark was changed. Setting aside any instructions that may have been given to senior examiners in this process, the implications for paper 2 are profound. It has already been shown that the paper grades A to C on paper 2 are separated by only 4 raw marks. The preestablished tolerance of 2 raw marks means that the senior examiner must differ in his/her judgement by at least 1½ unit grades before a raw mark is changed. Put another way: if a senior examiner judges a candidate’s paper to be top grade B standard, but the original examiner thinks it only low grade C, the original examiner’s mark will remain. Can CIE really defend this as excellence and best practice in assessment (CoP para 1.2(a))? Para 4.7(e) commits CIE before the issue of results and after the grading process (my emphasis) to “targeted re-marking by senior examiners of the work of candidates who are most at risk of receiving inappropriate results because they are close to a grade boundary”. Only one school that appealed, either at Stage 1 or Stage 2, reports evidence that CIE actually did this for some of their borderline candidates. Declaration by CIE in defence of such a process having happened would be a strong defence in an appeal. Since only one school reports such, it can only be assumed that it was not in fact done routinely though the transparent audit trail (see paras 1.3(e) and 1.3(f) above) should be a swift way of confirming this. Further evidence that this was not done to any widespread extent is clear from the example quoted on page 17. The candidate was one raw mark below qualification grade A*, that mark was awarded by the senior examiner on review of paper 2 but denied because of the tolerance rule. That candidate must definitely have been re-marked (note re-marked, not reviewed) by a senior examiner before results were issued. If so did the senior examiner at that stage not think the candidate had achieved the grade A* mark? No explanation was provided to the centre on this stage of the marking process. One final observation on this para is relevant. The wording does not specify what is meant by most at risk of receiving inappropriate results. To any reasonably informed teacher, candidate or parent this carries the immediate message of being very close to a qualification grade boundary. CIE has already confirmed that in their massively increased candidature for 0522 a large number of candidates were at the C/D borderline. Again, in the absence of any clear evidence to the contrary, it is most unlikely that CIE would have had unallocated senior examiner capacity to re-mark this volume of borderline scripts, even if they prioritised those candidates who fell just below a (qualification) grade boundary and gave those who were just above the benefit of the doubt. This aspect of CIE’s processes appears to be far from transparent. Paras 4.9(a) and 4.9(b) commit CIE’s principal examiners to producing and CIE to publishing examiners’ reports. This has not been done for syllabus 0522 and does not appear to be available for syllabus 0500 for the June 2015 examination. See http://www.cie.org.uk/programmes-and-qualifications/cambridge-igcse-english-first-languageuk-0522/past-papers/ and http://www.cie.org.uk/programmes-and-qualifications/cambridge-igcse-english-first-language0500/past-papers/ The paragraphs in section 5 of the CoP deal with the crucial aspect of the grading process. CIE’s own technical paper (attachment 3) gives details about how grade boundaries were finally arrived at by a mixture of judgemental and statistical means. GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 22 Para 5.2(f) requires the endorsement of the grading process by CIE’s Chief Executive, “if necessary after the production of further evidence to support it”. Given the problems with paper 2 grade boundaries identified in this report and in CIE’s own technical paper, and the fact that Ofqual commented on the narrow raw unit grade A to C mark range (4 raw marks) it would be extraordinary if the Chief Executive did not question the proposed grading and ask for further evidence. No confirmation of such was reported by schools at appeal stage. Para 5.3(b) commits CIE to seeking the views of teachers about the difficulty of question papers and consider them “when making grading decisions”. It is not clear how or if this was done. The only practical way is via teacher associations but neither GSA nor HMC report having been canvassed in this way. CIE needs to clarify whether this process was actually done as required by the CoP. A number of paras in section 5 promise further safeguarding checks to confirm accurate of results before results are issued. Para 5.3(e) wherever possible an independent measure of the ability of the cohort will be used to compare with previous years’ cohorts. There is evidence from CIE’s technical report that this was done for matched pairs on syllabus 0522 but no reference to candidates on syllabus 0500. CIE need to confirm whether this was done. Para 5.3(f) linked to 5.3(e) above commits CIE to considering changes in the aggregates of forecasts from centres as an indicator of whether centres perceive their candidates to be of the same ability as those in previous years. Again, it is crucial to ask if this was done, if so was it done for all centres (of which many would be new to CIE) or was it done for those centres which had previously entered candidates with CIE. And if it was done, where is the evidence for that and the impact it had on grading decisions? Para 5.8(a) commits CIE to further checks on the likely accuracy of results by reference to benchmark centres, a control group of centres or other cohorts of candidates. The technical paper gives no information about whether any of these checks were done. CIE needs to specify which and provide details. Para 5.8(b) commits CIE to yet another pre-release check where a centre’s results differ markedly from previous years. A sample of candidates’ scripts will be checked before results are issued. Given that all the many centres who registered their alarm at their 2015 IGCSE 0500 results fall into the category mentioned, yet not one reports CIE as having confirmed that their results were, in fact, checked before their results were released, it is important to ask (yet again) whether this was done for any centres. In summary CIEs CoP sets out clear procedures to establish standards year on year. It is especially detailed in setting out several measures to safeguard against incorrect grading, in fact it goes well beyond the equivalent Ofqual CoP and could therefore be viewed as a model of best practice. It is particularly alarming therefore that with all the safeguards in place, as required by the CoP, that the 2015 CIE IGCSE English (both syllabuses 0500 and 0522) generated such a storm of complaints, i.e. that so many centres were so outraged by the results awarded to their candidates. It is further alarming that CIE should appear to be so dismissive of centres’ concerns and complaints when their CoP commits them to close engagement with centres. GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 23 These points, and the questions raised above, deserve further discussion between CIE and GSA & HMC. 3.4 June 2015 grade thresholds compared to June 2014, November 2014 and November 2015 CIE publishes grade thresholds on its website so it is instructive to see how the raw mark grade thresholds for June 2015 compare with those of syllabus 0500 in June 2014 and also the November 2014 & 2015 papers. The data are shown below in Table 13. All raw marks are out of 50. Table 13: raw mark grade thresholds A* (notional) Paper 2.1 A B C D E June 2014 31 29 27 24 21 18 November 2014 34 30 26 21 18 15 June 2015 33 31 29 27 24 22 November 2015 31 28 25 23 20 18 June 2014 40 35 30 26 22 19 November 2014 35 32 29 27 23 19 June 2015 36 31 26 21 18 15 November 2015 36 32 28 24 21 18 Paper 3.1 One particular point to note is that only 4 raw marks span grades A to C in June 2015 for paper 2.1 compared to 5 raw marks in each of June 2014 and November 2015, and 9 raw marks in November 2014. The C/D boundary (27/50) in June 2015 was also much higher than in the other three comparator years, 3 marks higher than in June 2014 (though CIE’s technical paper quoted a smaller differential. 3.6 CIE IGCSE English June 2015 grade profiles (0500 and 0522 combined) compared to June 2012 to June 2014 % A* %A*/A %A*- B %A*- C %A*- D %A*- E Entry numbers June 2012 16.0% 37.0% 57.8% 80.2% 91.8% 97.1% n/a June 2013 8.4% 20.9% 37.5% 65.8% 86.7% 95.6% 65,441 June 2014 6.8% 18.2% 36.2% 66.0% 87.5% 95.8% 121,981 June 2015 4.3% 14.1% 33.4% 64.4% 86.4% 95.0% 212,028 The grade profiles above are consistent with a changing cohort population which has a decreasing proportion of very able (A*/A) students which is what CIE assert, hence the dramatic reduction in the %A* and %A*/A across the four examination years. And whilst it is certainly true that the %A*-C (66.0%) in CIE IGCSE English in June 2014 was greater than in GCSE English (61.4%, England only) – which led CIE to judge their 2014 standards as being too lenient – it is notable that the %A*-C in GCSE English returned to a higher level (65.3%) in 2015. The volatility at A*-C is with the GCSE results. Further, the other large scale GCSE English (AQA) with over 200,000 entries records an A*-C percentage of 70.2% for 2014, well above that GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 24 of CIE: why then did CIE consider they had been too lenient in 2014? Was this due to Ofqual or DfE6 pressure concerned about the growing popularity of IGCSE as an option to GCSE? GCSE English June 2012 – June 2015 (taken from JCQ published results for England) % A* %A*/A %A*- B %A*- C %A*- D %A*- E Entry numbers June 2012 3.4% 15.0% 35.5% 64.2% 85.1% 94.4% 611,996 June 2013 3.3% 14.2% 34.5% 63.7% 85.2% 94.4% 666,288 June 2014 3.6% 14.3% 34.3% 61.4% 84.7% 93.8% 453,350 June 2015 3.1% 14.4% 37.1% 65.3% 86.4% 84.4% 459,027 SECTION 4: SCHOOLS’ FEEDBACK AND COMMENTS This section considers outcomes of Enquiries about Results (EARs) and comments from individual schools, including any comments from Stage 1 or Stage 2 appeals or other ad hoc communications. All the schools are members of GSA and/or HMC but their identities have not been revealed in this report. 4.1 EARs EARs were reported from 44 schools for 1,068 candidates resulting in 176 (16.8%) of changed grades. It is probable that the remaining 9 schools also engaged with the EAR process but this could not be confirmed from the data submitted. Of the 44 schools 10 had no grades changed at all; the highest proportion of grade changes from any of the schools was 33%. In the light of the data set out in section 2 above, it is unsurprising that most schools requested EARs on paper 2 results. 4.2 Stage 1 & Stage 2 appeals 14 schools reported going to appeal with a further three schools that entered candidates for CIE IGCSE English (0522). Not one reports having their appeal upheld. CIE publishes guidance to centres on Stage 2 appeals (see attachment 7). At all stages of the appeal every effort will be made to establish whether CIE has used procedures which were consistent with its Code of Practice and whether it applied its procedures properly and fairly in arriving at its judgements. The hearing of the appeal will take the form of a re-examination of the evidence, documents, comments and reports seen at Stage 1 of the appeals process, presented in writing by the appellant and the CIE representatives. While it is not usual for the appellant to attend the hearing they may do so Two observations; 6 See letter from Nick Gibb, Education Minister, to examination boards (16 January 2015) [attachment 6] in which there is the clear message about IGCSE (level1/2 certificates) being a threat to the new (more demanding) GCSEs. “Including level 1 / 2 certificates in performance tables risks undermining the Government’s national curriculum and could lead to a less demanding curriculum for some students. Simply put, the Regulator believes that the biggest market opportunity for awarding organisations would be to create level 1 / 2 certificates that are less demanding than new GCSEs”. GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 25 Without full access to all CIE documentation para 1 above makes it virtually impossible for an appeal to succeed. It would be interesting to know how many CIE Stage 2 appeals (across all subjects) have been upheld this year and last. It is clear that CIE does not expect the appellant to appear to present their case. This might explain some schools’ comments following their Stage 2 appeal. CIE Stage 2 appeals documents also states: The CIE Appeals Committee is composed of independent members, who are not CIE employees. which gives the impression that all members of the appeals panel (committee) are independent. However from the Stage 2 appeal reports submitted by three schools in this study only one member was categorised as independent. Does this contravene CIE’s own regulations? Further, and to add to the confusion, one panel member is listed as being independent on 2 occasions, but a committee member on another occasion. Clearly he can’t be both. 4.3 Schools’ own comments Many schools provided detailed commentary about their attempts to rectify (what was in their view) wrong results awarded to their candidates. A number provided detailed exchanges between themselves and CIE. It is not possible in this report to attach all the comments which are inevitably lengthy, detailed and individual and apply to the school’s own context, but the common themes have been identified. (a) The Quality of Marking: Inevitably the quality of marking dominated most comments. Equally inevitably CIE rejected such challenges largely on the grounds that (i) they recruited, trained and monitored examiners (including the large number of additional examiners) to a high standard and (ii) whilst schools were entitled to re-mark retrieved scripts themselves, their teachers had not been part of the standardisation process and therefore the school’s own marking had no authority in the examination process. Some schools, in fact, went into considerable detail, to defend the high calibre answers provided by their candidates which had not been identified by the original examiner or allowed by the mark scheme. In some cases it appeared that sophisticated answers, at an intellectual level well above IGCSE were simply not recognised. There was also criticism from some schools that marking was inconsistent, a feature identified in Section 2.4 above. To quote from one school: We also recalled eight scripts from CIE to study the marking process and our findings were that the marking of the papers was inconsistent and seemed, in some cases, to have been completed by non-specialists. This latter charge of marking having been done by non-specialists is something that CIE will want or need to respond to; i.e. how many of the large number of examiners employed by CIE in June 2015 to mark IGCSE English 0500 and/or 0522 were non-specialist English teachers? (b) Detailed critiques of the question papers Schools used their expertise to evaluate and analyse the questions identifying weaknesses in style or failure to adhere to the specification. Their comments appear to have been GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 26 rejected though the CoP does state (para 5.3b) that “CIE will seek the views of teachers about the difficulty of question papers and consider them when making grading decisions”. (c) Frustration with the Appeals process There was extensive detailed correspondence between schools and CIE about appeals, both at Stage 1 and Stage 2. Schools’ concerns and CIE responses followed a general pattern as outlined below. School’s complaint/challenge CIE’s standard response Poor exam questions Paper was subject to usual quality assurance Poor or inconsistent marking or lack of adherence to the mark scheme Examiners were recruited, trained and monitored to a consistently high standard Procedures not followed correctly All procedures followed correctly Failure to adhere to the Code of Practice All aspects were Code of Practice compliant Although CIE generally appeared to respond in reasonable detail and occasionally sympathetically to concerns expressed: “despite your slightly lower forecast grades at grades A*, A and B, I can see that the results you obtained at grades A* and A have dropped by more than you expected, and to this extent I can understand your concern”, schools uniformly remained dissatisfied and frustrated by the appeals process. Many schools gave up after Stage 1, one school writing: “despite the scale of our issues with CIE, we decided not to appeal as we deem it a pointless process, which they will ultimately conclude on the point that they followed their processes so can’t possibly be wrong!”. A significant number of schools did, however, proceed to Stage 2 such that CIE fell behind their required schedule of arranging and hearing the appeals. In more than one case CIE wrote to a school apologising for the delay saying: “This delay has been caused by an unexpected volume in Stage 2 appeals”. The fees were waived as a result. That there has been such an “unexpected volume” must surely cause CIE to reflect and review the reasons behind it. Frustration was not just with the outcome but, as in the case of the school quoted above, with schools’ feeling that there were no grounds on which a Stage 2 panel could or would uphold an appeal. The standard response following a Stage 2 appeal appeared to be The panel reviewed the evidence submitted and heard the cases presented by Cambridge and the centre. Following the hearing, the panel discussed the appeal and found that Cambridge had (1) used procedures consistent with the Code of Practice, and (2) applied these procedures properly and fairly in arriving at their judgements. The appeal was therefore REJECTED. Thus despite there being evidence of eminently poor marking (one centre had a candidate’s paper 2 mark raised from 24/50 to 36/50: the original examiner was identified as 03.05) CIE would not acknowledge poor or inconsistent marking to exist and it was GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 27 clearly outside the remit of the Appeals Panel to evaluate it. It is not know whether CIE reviewed all scripts marked by examiner 03.05 as a result of this gross error. Failure to adhere to the Code of Practice was a major line of challenge from many schools but one school which prosecuted this line in detail at a Stage 2 appeal reports that: “The panel seemed to think it (i.e. scrutinising whether CIE had adhered to its CoP) was just far too much to deal with at the hearing”. The result of all this is a large number of schools feeling aggrieved that their candidates have been denied their correct grades and a major loss of confidence in CIE as an examination board and in its procedures. The one good news post script is that two schools report the results of candidates having re-sat in November: in one, four students who had previously been awarded grade D in summer all gained grade C, with nil or minimal preparation; in another their one re-sit candidate’s grade improved from C to A, again with no additional preparation. SECTION 5: FINAL COMMENTARY & EVALUATION The detailed analysis and commentary above attempts to answer four questions about the 2015 CIE IGCSE First Language English (0500) examination results 1. was there really a problem? 2. if yes, how did the problem arise and what was its extent? 3. if yes, what has been CIE’s response to the problem? 4. If yes, what, if anything, can be done to protect the interests of the 2015 candidates and to ensure that there is no repetition of the problem in 2016? On 1 the evidence is indisputable: there was a major problem with the CIE 0500 results in 2015, especially at the top grades of A* and A. CIE’s own reports acknowledge this but not in sufficient detail. The detailed analysis in this report shows this to be largely, but definitely not exclusively, among those candidates who were entered for papers 2 and 4. On 2 there were a number of contributory factors including: CIE’s judgement that the 2014 standards were too lenient especially at grade C but also at grades A and E CIE’s attempt to correct this for 2015 which also saw a significant increase in entry especially in the tied syllabus 0522 The need to recruit, train and monitor a very large number of additional examiners for a subject in which marking quality has been of serious concern for some years. It is no surprise that schools reported poor and inconsistent marking. The exceptionally difficult task of adjusting the grade C standard between 2 tied syllabuses, between core and extended tiers and between candidates who had taken different options (i.e. papers 2/3 or papers 2/4). This had a major knock-on effect for grade A* and A which CIE themselves have identified. The very small grade A-C raw mark range that ensued on paper 2 The capping of raw marks on paper 4 (coursework) such that a notional unit A* could only be achieved by candidates gaining full marks. GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 28 However, even when these contributory factors are acknowledged the question remains as to why CIE undertook such a realignment of standards in the first place. One very plausible reason is that they were under pressure to do so from Ofqual (for syllabus 0522) who themselves were under pressure from the DfE to take action on what was perceived as the easier standard of IGCSE (level 1/2 certificates) at a time when large numbers of maintained sector schools were opting for it. GSA & HMC might consider testing this by making a Freedom of Information (FoI) request to Ofqual to gain access to communications with CIE about IGCSE English between January and September 2015. On 3 that CIE produced explanatory documents once the scale of the schools’ concerns became apparent is to be welcomed but we suspect that this was done only because of the volume of complaints about the results. It would be extraordinary if CIE did this routinely for every subject. However, aspects of CIE’s documents played down the scale of the problem and failed to provide the necessary detailed breakdown of grades between option routes that would have shown the nature of the problem. Their documents only hint at problems associated with the top grades; they should publish the full details so that the extent of the A* and A problem is clear to everyone, including Ofqual as regulator. In their Code of Practice, CIE commit to being fully transparent; this is their opportunity to make that commitment a reality. On 4 CIE should, as a matter of urgency, consider what action can be taken to protect the interests of the 2015 candidates who have been subject to the IGCSE English 0500 problems detailed in this report. GSA and HMC could consider sending this report to Ofqual both to inform them of the report’s findings and to ask them to monitor any retrospective action taken by CIE. For the June 2016 examination CIE should ensure that none of the fundamental flaws seen in June 2015 are repeated, and Ofqual should monitor them. Finally, it is essential to return to the impact all of this has had on the candidates themselves, large numbers of whom will have an IGCSE English grade that does not reflect their ability or performance, and the impact that this could have on their career aspirations. It is also important to remember their teachers and parents, whose confidence in the examination system will have further decreased as a result. GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 29 SECTION7: APPENDICES Appendix 1: distribution of raw marks on paper 2 2015 CIE 0500: Paper 2 raw mark scores in HMC & GSA schools 400 350 Frequency 300 250 200 150 100 50 0 <10 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 Paper 2 raw mark score (max 50) Appendix 2: distribution of raw marks on paper 3 2015 CIE 0500: Paper 3 raw mark scores in HMC & GSA schools 80 70 Frequency 60 50 40 30 20 10 0 <10 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 Paper 3 raw mark score (max 50) GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 30 Appendix 3: distribution of raw marks on paper 4 2015 CIE 0500: Paper 4 (coursework) raw mark scores in HMC & GSA schools 450 400 Frequency 350 300 250 200 150 100 50 0 <10 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 Paper 4 (coursework) raw mark score (max 50) Appendix 4: distribution of raw marks on papers 2+3 2015 CIE 0500: Paper 2 + 3 raw mark scores in HMC & GSA schools 60 50 Frequency 40 30 20 10 0 <40 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 85 87 89 91 93 95 97 99 Total GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 31 Appendix 5: distribution of raw marks on papers 2+4 2015 CIE 0500: Paper 2 + 4 raw mark scores in HMC & GSA schools 250 Frequency 200 150 100 50 0 <40 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 85 87 89 91 93 95 97 99 Total raw marks (Paper 2 + Paper 4) GSA & HMC report (April 2016) v2: 2015 CIE IGCSE English (0500) page 32 SECTION 7: ATTACHMENTS Attachment 1: CIE briefing paper for schools (i) GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 33 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 34 Attachment 2: CIE briefing paper for schools (ii) GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 35 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 36 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 37 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 38 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 39 Attachment 3: CIE’s Technical Paper GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 40 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 41 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 42 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 43 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 44 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 45 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 46 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 47 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 48 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 49 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 50 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 51 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 52 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 53 Attachment 4: CIE EAR data June 2015 All syllabuses June 2015 45,984 0522: regulated IGCSE English Language June 2015 26,858 0500: non-regulated IGCSE English Language June 2015 3,910 0500/0522 combined June 2015 30,768 Grade changes (June 2015) Enquiries closed 20,394 June 2014 8,263 June 2014 2,090 June 2014 10,353 EAR Grade changes % % increase 125% % increase 225% % increase 87% % increase 197% EAR grade changes as % of subject cohort All syllabuses 45,984 9.0% 0.4% 0522 26,858 9.4% 1.3% 0500 3,910 8.6% 0.7% 30,768 9.3% 1.2% 0500/0522 combined Ofqual report on 2015 examinations7 7 June 2014 18.0% See https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/512899/2016-03-31-enquiriesabout-results-summer-2015.pdf GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 54 Attachment 5 CIE Code of Practice – extracts relevant to this Report [The full Code of Practice can be downloaded from http://www.cie.org.uk/images/7881-code-of-practice.pdf Para 1.2 (a) CIE will remain committed to excellence and best practice in assessment Para 1.2 (d) CIE will ensure that all candidates receive the results that their performance merits when judged against the relevant syllabus content and assessment criteria. Para 1.2(g) CIE will use expert judgements and statistical evidence to set and maintain internationally recognised performance standards. Para 1.2(h) CIE’s assessments will be criterion-referenced or standards-referenced rather than norm referenced. Para 1.2(i) CIE will comply with its customers’ regulatory and procedural requirements and act in accordance with this Code of Practice to ensure that assessment standards are maintained. Para 1.2(j) CIE will take steps to encourage the appropriate use of the outcomes of its assessments. Para 1.3(a) CIE will use customer feedback and self assessment to target areas for development, innovation and continual improvement. Para 1.3(e) CIE will ensure that for each process in its assessment system there is an audit trail that sets out the key evidence that supports the decision taken. Para 1.3(f) CIE’s assessment system will be transparent, in that the evidence on which decisions are based will be clear, it will be possible to audit the decision-making process and it will be possible for those making decisions to be held accountable. Para 1.4(e) CIE will assess all candidates for what they show that they know and can do, not for what they might have achieved had circumstances been different. Para 1.5(a) To help teachers prepare students for future examinations, CIE will publish a report to Centres from the senior examiners for each externally examined component after each examination session. Para 4.2(b) Principal Examiners will be responsible to the Product Manager, who will ensure that the examination as a whole meets the requirements of the syllabus and maintains standards year on year. Para 4.7(d) When a senior Examiner awards a mark that is different from the original Examiner’s, it will be the senior Examiner’s mark that prevails, subject to any pre-established tolerance within which it is agreed that marks will not be adjusted. Para 4.7(e) Before the issue of results and after the grading process (Section 5) has been completed, there will be a targeted re-marking by senior Examiners of the work of candidates who are most at risk of receiving inappropriate results because they are close to grade thresholds. Paras 4.9(a) & (b) Principal Examiner’s Report GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 55 Para 5.2(b) CIE’s grading processes will use a combination of professional judgement and statistical evidence. The statistical evidence used will relate to more than one previous examination session. Para 5.2(d) CIE’s grading processes will ensure that the standard of a qualification is maintained from one year to another, so that CIE’s results will be standards-referenced. Due consideration will be paid to the need for alignment with any equivalent qualification taken in the UK. Para 5.2(e) The Product Manager will be responsible for the initial grading of a syllabus, but at least one other person will be involved. Where appropriate, representatives of Ministries or other partner assessment organisations will be involved in the grading process. Para 5.2(f) The outcome of the grading process for each syllabus will be endorsed by CIE’s Chief Executive, if necessary after the production of further evidence to support it. Para 5.3(a) If the assessment tasks for a component are unchanged or are believed to be of equal inherent demand, the grade thresholds will remain unchanged from one year to another. Otherwise grade thresholds will be raised or lowered from one session to another. Para 5.3(b) CIE will seek the views of teachers about the difficulty of question papers and consider them when making grading decisions. Para 5.3(e) Wherever possible, CIE will use an independent measure of the ability of the cohort (e.g. their results in previous examinations or in control/ reference tests) to confirm any belief that they are better than some other cohort and should therefore do better in an examination. Para 5.3(f) Consideration will be given to changes in the aggregates of forecasts from candidates’ Centres as an indication of whether teachers perceive the candidates to be of the same ability as those in previous years. Para 5.4(c) The aim in the determination of grade thresholds will be that alternative options within papers, alternative papers within syllabuses, alternative syllabuses in the same subject, and alternative subjects within the same qualification will be equally demanding of candidates. Para 5.5(b) Each Principal Examiner will make a report to the Product Manager for their paper stating and explaining the minimum mark that they recommend should be taken for each of those thresholds designated as ‘key’ thresholds. They will do so at the end of the marking period and before the grading process itself. Para 5.6(c) Those responsible for decisions about grade thresholds will sign a record of the outcome. Para 5.7(a) A candidate’s syllabus grade will be calculated directly from the total of their marks on the components that they took (weighted in accordance with the syllabus), not from the component grades calculated for them. GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 56 Para 5.8(a) Checks on the likely accuracy of prospective results will be conducted by reference to benchmark Centres, a control group of Centres or other cohorts of candidates. Para 5.8(b) Centres where the prospective results in a subject differ greatly from the previous year or from the forecast grades will be identified, and a sample of their results will be checked before issue. Para 5.8(c) Particular attention will be paid before the issue of results to candidates around critical borderlines that are commonly used for the determination of progression to the next stage of education or where failure negates or reduces candidates’ success in other examinations (e.g. for a Group Award). GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 57 Attachment 6: DfE letter to awarding bodies GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 58 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 59 Attachment 7: CIE’s guidance on Stage 2 Appeals GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 60 GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 61 Attachment 8: extract from the Ofqual Report on the 2015 examinations Monitoring Cambridge International IGCSE First Language English (0522) Entries for this qualification have increased in recent years and in summer 2015 they almost doubled to just over 200,000. As a result we have monitored this qualification closely in recent years. Cambridge International informed us early in 2015 that their routine analysis had identified some leniency in grading IGCSE First Language English in summer 2014. Cambridge International told us that, as a result, it intended to tighten its grade standards at Grade C and, to a lesser extent, at Grade A in summer 2015. Cambridge International had particular challenges in setting standards in this qualification. This was partly due to the increased entry: about half of the increase was from schools new to the syllabus, while the other half was from existing schools entering more students. Cambridge International told us that the increase in entry was also disproportionately focused on students who might be expected to achieve C/D, potentially exacerbating a clustering of students around the C/D borderline. We asked Cambridge International to provide us with the evidence for how they had come to its awarding decisions in 2015. In setting standards in IGCSEs, Cambridge International use very similar evidence to the GCSE exam boards. Cambridge International considered predictions based on KS2 prior attainment and comparisons of the results for 'benchmark centres' - schools with stable entries for this syllabus in 2014 and 2015. Evidence from the benchmark centres suggested that to tighten grade standards as far as Cambridge International had intended (in relation to KS2 predictions) would have been too severe. The other factor that made awarding more challenging was the bunching of marks, particularly around the C/D boundaries. On one paper, the difference between C and A was only 4 marks. We concluded that Cambridge International had carried out its grading appropriately. Cambridge International has carried out further analysis since the summer and is confident that their grading was appropriate. GSA & HMC report (April 2016): 2015 CIE IGCSE English (0500) page 62