The move towards Computer Marking from Image+ (CMI+) and the impact of sequential numbering on GCSE question papers

Oct 20, 2026

Uploaded by Angel

Study examines whether switching GCSE question papers for long-form answers from a nested alphanumeric numbering convention to a purely numerical two-digit code benefits candidates in computer marking with image-based systems (CMI+). It addresses machine-reader segmentation requirements and incompatibilities with lettered and Roman-numeral part-questions, while considering risks to candidate clarity and performance. Measures include completion time, question scores, rubric infringements, and candidate impressions; the experiment alters only the numbering system.

LanguageEnglish

CategoryExam

FormatPDF

Pages9

Reading Time23 min

AccessFree to download

Introduction

Computer marking and numbering incompatibilities

Study focus and evaluation measures

Method

Participants

Materials

INTRODUCTION The move towards Computer Marking from Image+ (CMI+) 1 for long form answers, such as essays, has raised an issue concerning the way that question papers are numbered. One of the challenges for introducing the CMI+ system for components with long form answers is the segmentation of answer booklets. The system needs to detect not only where an answer to a question starts and finishes but also what question is being answered. Some of AQA‘s questionnumbering conventions are not compatible with the capabilities of the machine readers used for CMI+, at least not to the degree of accuracy that makes automation worthwhile. For example, the use of letters (a, b, c) and Roman numerals (i, ii, iii) to denote part-questions, such as 1(a)i, cannot be accommodated by the system. It has, therefore, been decided that a purely numerical system will be adopted for these papers, such that any item can be identified using a two digit code (0 1 – 99) . Question papers currently in the format of combined question paper/answer booklets, and all question papers that can be put into a combined question paper/answer booklet format for June 2010, may continue to use the nested numbering system, e.g. 1(a)i. 1 An online system where examiners mark scanned images. Nonetheless, in the long run, maintaining two parallel question paper numbering conventions where one will suffice is at odds with the goal of promoting a consistent―house style‖ with which candidates and teachers are familiar across specifications and subjects. The current study investigates whether there is any measurable benefit in maintaining the nested numbering system. GCSE Geography and GCSE Mathematics are two subjects that make extensive use of the nested numbering system because they both tend to use stimuli and scenarios to generate a series of thematically related questions. In some cases, the answer to one part-question may be required to produce the answer to a subsequent part-question. Several senior examiners for GCSE Mathematics expressed concern at the prospect of losing the alphanumeric system although, at that stage, the proposed replacement was a numerical equivalent to the alphanumeric system, where Arabic numerals would replace the letters and Roman numerals, e.g. 1(a)i would become 1-1-1 and 2(c)iv would become 2-3-4. Although (perhaps mercifully) this system was not introduced, the simple sequential numbering system might make the relatedness of these part-questions less obvious to candidates than the alphanumerical system (and even its Arabic equivalent) . Presumably, it was this clarity that the examiners were concerned about losing. Unfortunately, owing to specification development commitments, it was not possible to recruit to this study the senior examiners from GCSE Mathematics who had vocally opposed the change; however, a Principal Examiner and his Assistant Principal Examiner were recruited from the GCSE Geography A legacy specification to collaborate on this study. In terms of how the sequential numbering system might affect candidates in comparison with the alphanumeric system, there are three objective measures that this study focuses on: the time candidates take to complete the questions, their question scores, and the incidence of rubric infringements. Candidates‘ subjective impressions of using the sequential numbering system were also collected, to complement the objective measures . For example, candidates may like the new numbering system but, because they are unfamiliar with it, appear to perform worse when using it; alternatively, the objective measures might show that the numbering system had no affect on performance, but candidates may express strong dissatisfaction with it. So that the worst case scenario was tested , only the numbering system was altered on the papers used in the study. In practice, it seems likely that alterations to the format of the question papers could, in many cases, be used to compensate for the cha

FAQs

Why was a purely numerical question paper numbering system adopted for CMI+ long-form answers?

Some existing AQA numbering conventions use letters and Roman numerals for part-questions, which the machine readers in CMI+ cannot accommodate with sufficient accuracy for automation.

Which objective measures does the study use to assess the numbering system change?

The study focuses on completion time, question scores, and the incidence of rubric infringements.

How were participants and materials organized in the study?

Four centres provided mock exam paper 3031/2 to 485 candidates, with half receiving the unmodified June 2009 papers using alphanumeric numbering and the other half receiving a modified version where only the numbering system changed.

Oct 20, 2026

Uploaded by Angel