Evidence / Case studies
Evidence before claims: our case-study approach
PaperGrader does not publish unnamed or unverifiable institutional success stories. When an institution permits a public case study, it should explain the assessment context, workflow, review method and the limits of the evidence rather than relying on a headline claim.
For institutions planning a measured pilot.
A pilot can help an examination team understand whether PaperGrader fits its own script formats, rubrics, reviewers and operating requirements before it makes a larger deployment decision.
Case studies are only useful when their evidence can be understood.
A percentage without the subject, script type, review method or baseline does not establish value for another institution. PaperGrader's public evidence policy is therefore conservative: we publish only approved, attributable material with enough context to interpret it.
How the evaluation workflow operates
01
Set a pilot question
Agree what the institution wants to evaluate, such as review time, question mapping or score agreement.
02
Establish a baseline
Record the current assessment workflow and define reference grading or operational measures before the pilot.
03
Run a representative batch
Include the script types, subjects and scan conditions relevant to the intended use.
04
Review outcomes
Analyse errors, educator overrides, workload and process observations with the authorised academic team.
05
Publish only with permission
A public case study requires institutional approval, attribution and an explanation of its methods and limitations.
No public case study is currently presented as a universal benchmark.
Results from a pilot should not be generalised to every subject, handwriting style or assessment type. We will publish public case-study results only when the participating institution authorises the material and the methodology is clear enough to evaluate.
Operational benefits
- A pilot designed around the institution's own assessment context
- Evidence that distinguishes quality from speed claims
- A baseline for review workload and operational change
- Clear treatment of limitations and exceptions
- No invented customer names, ratings or performance statistics