Skip to content

The evidence exists before the appeal does

Six records, written as the exam ran

An appeal asks a short list of questions, and StudyDrome Exam Manager answers each from a record written at the time. Six records carry it: the publication snapshot, the item statistics, the reliability figure, the pass-mark record, the conduct report, and the grading history. None is assembled afterward.

Book a pilot

What does an appeal actually ask?

An appeal is not an open-ended inquiry. It asks a short list of questions, and each one has a record that answers it.

The question

The record that answers it

Written

Is this the paper we approved?

The publication snapshot

When the exam is published

Was this question fair?

The item statistics

Once the responses are in

Was the paper precise enough to decide on?

The reliability figure and its error band

With the statistics

Where did the line come from?

The pass-mark record

When the pass mark is set

What happened during the sitting?

The conduct report

While the sitting runs

Who changed this mark, and when?

The grading history

At each change

Six records, six questions. A panel that has all six rarely needs a seventh.

When is each record written?

Read the right-hand column downwards, and the point of the page appears. Not one of those records is written after an appeal begins.

The snapshot is taken at publish, before anybody sits the paper. The conduct record is written while candidates work. The grading history gains a row each time a mark changes. The statistics and the reliability figure follow the responses. The pass-mark record is set when somebody sets the pass mark.

So a challenge does not start a process. It opens records that were already there.

Each one is created at the moment it describes. That is what makes preparing a case a retrieval job rather than an investigation.

Why does the order matter?

The order in the table is not presentation. Each record means something because the one above it is already fixed.

Statistics computed on a paper that could still change would prove very little. Nor would a pass mark set against a moving target. Publishing freezes the paper first, and everything after it is computed against that frozen version.

What freezes at publish covers the mechanism. For an appeal, the consequence is the useful part. The paper in the evidence is the one the candidate sat on.

Where does the pass mark's own record live?

Where did the line come from? is the question that catches institutions out, because the answer usually lives in somebody's inbox.

An exam carries its standard-setting record with it. The record holds the method that set the line and the per-station decisions that shaped it. The regression report goes one step further: it states its own method on the page it prints, and cites the published reference behind it. A panel reads where the mark came from instead of taking it on trust.

Standard setting covers the four methods and how each derives a mark. What matters here is narrower — the record exists, and it travels with the exam.

How long does the evidence stay good?

Appeals arrive late. A result challenged in the spring can turn on a paper sat the previous autumn.

Every export is generated from the published exam, so the figures do not drift. An export produced a year after the sitting carries what the one produced the week after it carried. Reporting and exports set out what each report produces.

A bundle assembled at appeal is therefore a reproduction, not a reconstruction.

For an exam office preparing a case

Assembling a file is mostly a matter of knowing which record answers the point at issue.

  • Each record exports on its own, so you assemble the case the panel actually asked about.

  • A results bundle can be left with candidate names masked, allowing a reviewer to read the scripts without names.

  • The conduct record is per attempt, so one candidate's file stays one candidate's file.

  • The pass-mark record travels with the exam rather than living in a meeting minute.

  • A mark that changes leaves a row of its own, carrying the time and the person behind it.

  • The board still makes the decision. The records exist, so they are based on evidence gathered in advance.

Frequently asked questions

How soon after an exam is the evidence ready?

Most of it is in place before the exam ends. The snapshot is written at publish. The conduct record is written while candidates work. The grading history grows as marks are entered. Item statistics and the reliability figure follow once the responses are in. Nothing in the list waits for somebody to ask for it.

Can an appeal panel see how the pass mark was set?

Yes. The exam carries its standard-setting record, which holds the method that set the line and the per-station decisions behind it. The regression report goes further. It states its own method on the page it prints, and cites the published reference for it. A panel reads the reasoning rather than accepting the number.

If an exam is corrected and republished, which version does an appeal look at?

The one the candidate sat. Each publication keeps the paper exactly as that group set it, and every statistic is attached to the version it was computed on. A correction made for a later sitting does not reach backward into an earlier one, so two cohorts can be defended separately.

What if the challenge is about one question rather than the result?

Two records answer that. The item statistics show how the whole cohort responded to it, which is evidence of whether it behaved as a question should. The item itself also keeps a numbered version trail, so the wording in front of the panel is the wording the candidate was shown.

Book a pilot

Bring us an appeal your board actually handled. We will run the equivalent in a sandbox and show you which record answers each point.

Book a pilot

Related