Distinguishing between marking errors and differences of opinion


July 27
12:00 2017

Ofqual has today (27 July 2017) published research that shows examiners were able to distinguish between marking errors and legitimate differences of opinion when conducting reviews of marking last summer. The analysis follows the introduction of new rules in 2016 to ensure reviewers only change marks when there is an error, and ensure fairness for all students.

Data published today shows that marks were unchanged following a review in more than half of cases requested by schools and colleges in 2016. Analysis of a sample of cases where marks were changed, including some of the most difficult cases to judge, reveals that examiners acted consistently with the new rules in a clear majority of cases, such that only marking errors were corrected. In a small number of cases a change was made even though there was no error in the original marking, and in a very small number of cases errors were not corrected. The reasons for this, including unusual responses, examiner error and mark schemes, are discussed further in the report.

Commenting on todays publications, Sally Collier, Chief Regulator said: It is pleasing to see that our new rules were used in many cases in the way we intended last summer. There will always be a period of adjustment following any change, and we are working with exam boards to identify what can be done this year to be even more confident that students are getting the results their performance deserves.

Of course, the first best solution remains for original marking to be as good as it can be. Our findings provide some useful information here too, and we believe that improvements are being made through the comprehensive redesign and introduction of new GCSEs and A levels.

New data have also been published today that show the extent of mark and grades changes in 2016 resulting from reviews of marking and moderation by subject and centre-type. These show that the incidence and extent of mark and grade changes varied by subject, explained by factors including the degree of non-exam assessment, structure of the qualification or nature of the assessment (objective vs subjective).

A further piece of research, also published today, provides an insight into the approaches of exam board moderators when considering centre-marked assessments. It identifies several areas for improvement, including factors that may influence their decisions. However, overall, the evidence is of consistency in their approaches to moderation.

The 4 documents published today are:

  1. Evaluation of Reviews of Marking and Moderation 2016: Study and survey

  2. Reviews of marking and moderation resulting in grade changes of 2 grades or more: Summer 2016 examination series

  3. Reviews of marking and moderation: subject level analyses

  4. An exploratory investigation into how moderators of non-examined assessments make their judgements


Ofqual is introducing changes to the systems schools and colleges use to challenge GCSE, AS and A level results in England to make them clearer, more consistent, and fairer for all students.

Among several reforms in summer 2016 we changed the rules so that mark changes would only be permitted where there was a marking error. This reflected earlier research that showed inconsistency in how marks were being reviewed, giving those who requested a review an unfair advantage.

In spring 2017, we announced that we would in future give schools and colleges a second opportunity to challenge GCSE, AS level, A level and project results if they continue to have concerns about marking or moderation errors. This adds to their existing right to appeal results on the grounds that an exam board hasnt followed its own procedures.

Related Articles


  1. We don't have any comments for this article yet. Why not join in and start a discussion.

Write a Comment

Your name:
Your email:

Post my comment

Recent Comments

Follow Us on Twitter

Share This

Enjoyed this? Why not share it with others if you've found it useful by using one of the tools below: