A Guide to Double Marker for GCSE English Language and Literature Mock Examinations

Welcome to Double Marker, an assessment system designed by No More Marking to resolve the workload crisis of marking secondary English mocks while maintaining the rigorous alignment required for GCSE preparation.

This guide explains how the system works, the underlying architecture, the step-by-step collaborative process, and our commitment to data security and GDPR compliance.

1. The Core Challenge & Our Solution

Marking GCSE English mock examinations is exceptionally time-consuming for secondary teachers. However, because mock exams serve as key preparation for national assessments, schools understandably require students and teachers to engage with the detailed, criteria-based mark schemes produced by official exam boards.

Traditionally, schools have faced a trade-off:

1. Holistic Assessment (Comparative Judgement): Proven to be highly reliable and accurate but does not naturally employ a traditional mark scheme.
2. Traditional Marking: Aligns with the exam board criteria but is notorious for being highly inconsistent and incredibly demanding on teacher workload.

Double Marker resolves this tension by combining both approaches into a unified "two-marker" system.

2. The Two-Marker Architecture

Double Marker uses a hybrid, multi-layered grading system to achieve both maximum reliability and strict mark scheme alignment:

Marker 1: AI Comparative Judgement

All pupil scripts are uploaded and processed using Comparative Judgement (CJ). Rather than looking at a rubrics-based mark scheme, the AI evaluates papers holistically to establish an incredibly reliable relative rank order and relative scale.

Marker 2: The Human Teacher (Targeted Sampling)

Instead of marking their entire cohort's papers, school teachers mark a small, carefully selected sample of student work. This is done on an online marking platform where teachers are completely blind to the initial marks assigned during the Comparative Judgement process. Teachers mark these specific questions using the standard exam board mark scheme.

The Rescaling & Optimisation Engine

Once teacher marks are collected, Double Marker's optimization routine calibrates the AI Comparative Judgement scale against the human marks. This mathematical routine accomplishes two critical tasks:

• Eliminating Bias: It ensures that the AI is not universally too harsh or too generous compared to human standards.
• Minimising Distance: It aligns individual AI scores with human marks to map the relative scale onto the actual, national GCSE mark scheme (e.g. mapping a raw comparative score onto a 0-40 exam board scale).

3. Scale Alignment & Quality Assurance

To ensure that mock results are valid both within a single school and across a wider multi-academy trust or national cohort, the system employs a two-tier moderation process:

School-Level Scale

First, each participating school's scripts are judged independently to place them on their own relative scale (for instance, ranging from 0 to 40).

Unified Cohort Scaling (Centralised Moderation)

To allow direct comparison between different schools, No More Marking extracts a 20% sample of scripts from every participating school. These scripts are loaded into a separate, centralised moderation pot where they are judged directly against one another. This 20% sample acts as an anchor, allowing the system to place all schools on a single, unified scale. For example, a stronger school’s marks might map from 4 to 39, while a school with weaker performance might see marks run from 0 to 36.

Quality Control Flags & Overrides

If an individual teacher's mark for a script differs wildly from the AI's comparative score, the platform automatically flags it for review. In all cases of discrepancy, a Senior Marker has the final authority to review the script and override any AI-derived mark, supported by our targeted flagging system.

4. The 12-Step Collaborative Process

Double Marker is structured as a 12-step collaborative workflow divided into four distinct phases.

Phase 1: Preparation

• Step 1 – Paper writing: No More Marking writes the assessment papers.
• Step 2 – School contact: The school designates a single "Super User" to serve as the sole liaison for logistics.
• Step 3 – CSV upload: The school uploads pupil data via a provided CSV template into a secure Dropbox folder.
• Step 4 – Booklet creation: No More Marking generates personalised, print-ready PDF booklets where questions and response areas are interleaved.

Phase 2: Administration & Scanning

• Step 5 – Administration: The school prints and distributes the booklets. Pupils complete their mock exams directly in them.
• Step 6 – Scanning: The school slices off the booklet spines and scans the completed papers in small, manageable batches (typically 4 pupils per scan file).
• Step 7 – Scan verification: The school uploads the scans, and No More Marking verifies receipt and provides an absolute confirmation check.

Phase 3: AI & Marker Setup

• Step 8 – AI processing (2 working days): No More Marking runs transient AI transcription, automated judging, and structural checks.
• Step 9 – Marker setup: The school Super User adds all marking teachers' emails to the platform. Teachers receive invitations to set up accounts.
• Step 10 – Allocation creation: No More Marking creates item-level allocations and notifies teachers that marking can begin.

Phase 4: Marking & Delivery

• Step 11 – Marking period (5 working days): Teachers mark their allocated questions online. Because marking is distributed at the item level (specific questions) rather than the paper level, the total workload is massively reduced to approximately 6 to 7 scripts per teacher instead of the entire cohort.
• Step 12 – Results and feedback (2 working days): No More Marking applies the Comparative Judgement scaling, calibrates against teacher marks, conducts QA checks, and delivers the final scores and feedback.

5. Rich Diagnostic Feedback

Within two weeks of the mock examination date, schools receive a comprehensive suite of data-driven reports:

• Pupil Level: Standardised marks and extensive diagnostic AI feedback on individual question responses.
• Class & School Level: Aggregated metrics showing strengths and areas for development.
• Cohort & Trust Level: Broad overview reports and comprehensive Examiner Reports summarizing performance across all participating schools.

6. Data Protection, Security & GDPR Compliance

No More Marking treats pupil data with the highest standards of security. Under UK GDPR, the school acts as the Data Controller and No More Marking acts as the Data Processor (satisfying Article 28 requirements).

Data Storage & Residency

• All persistent pupil data and scanned script images are stored securely on Amazon Web Services (AWS) S3 servers in Ireland (EU).
• The primary database storing task, candidate, and school information is hosted on MongoDB Atlas in Ireland (EU).
• Staff in the Philippines may remotely access scripts for quality checks, but data is never transferred or stored there. These staff are directly employed under standard UK contracts and undergo rigorous security screening (equivalent to the UK's enhanced Disclosure and Barring Service checks).

AI Sub-processors & Privacy

Our advanced AI capabilities are delivered through secure, enterprise-grade sub-processors bound by identical GDPR obligations:

1. Google Cloud Vision (EU): Used for initial document processing to separate handwriting from pupil-identifying details (e.g. names). Data is restricted to EU endpoints (eu-vision.googleapis.com) and processed entirely in-memory (not persisted to disk).
2. OpenAI (US endpoint - transient): Used for transcribing audio comments, automated judging, and generating summaries. Under our enterprise agreement, OpenAI encrypts all data in transit (TLS 1.2+) and at rest (AES-256), retains transient data for a maximum of 30 days solely for abuse monitoring, and is strictly prohibited from training models on our input or output data.
3. Google Gemini: Utilised for media-related AI processing under strict GDPR-compliant sub-processing terms.
4. Clerk & Resend: Utilised securely for user authentication/authorisation and customer email delivery respectively.

Data Retention

Schools retain complete control over their data. Any task deleted on the platform immediately removes all associated student data and images. Additionally, schools can request complete data erasure at any time by emailing support@nomoremarking.com (completed within 5 working days).

Updated on: 17/08/2026