Grace Marks or Item Analysis? Moderating Internal Assessment in a Postgraduate Course
You coordinate a four-credit postgraduate course with 52 students. Continuous internal assessment (CIA) carries 25 marks: Test I (10), Test II (10) and an assignment (5). Test I had a mean of 6.2 and SD of 2.1; Test II a mean of 5.4 and SD of 2.4. Assignment marks are compressed between 4 and 5. Nine students hold CIA below 12/25 and cannot comfortably reach the 50% pass mark even with a strong end-semester performance. One student, R, has 22.7/25 and asks you to round it to 23 because the difference is trivial. Two students missed Test II with dengue and produced certificates three weeks later. At the results meeting the head of department proposes adding three grace marks to every student's Test II score to normalise the cohort, and a parent has written alleging that the assignment was graded subjectively. The examination cell wants your decision in writing.
1. The head of department argues that adding three marks to every student's Test II score is a harmless normalisation. Evaluate this proposal against the principles of validity, reliability and fairness in classroom assessment, and state whether you would approve it.
Read the worked answer
Adding a constant to every score changes neither the rank order nor the shape of the distribution, so it cannot 'normalise' a cohort; it merely inflates all scores and compresses the CIA-to-end-semester ratio. It also fails validity: if Test II was too hard because of faulty items, the marks should be corrected at item level, not shifted wholesale. Reliability is not improved because measurement error is not reduced. Fairness cuts both ways: granting grace marks to students whose weakness is genuine advantage over students in other batches assessed without that concession, and it gives the examination cell an inflated signal about readiness.
I would reject the proposal. A uniform mark addition is defensible only for a documented institutional-level question-paper error affecting all candidates equally, approved before publication. Here the better route is item analysis: difficulty (p) and discrimination (D) for each Test II item, deletion or re-keying of defective items, and recomputation of marks for those items only, with the revised scheme recorded and approved. R's 22.7 should stay 22.7 unless a rounding rule applies to the whole class.
2. Design a defensible remedy for the two students who missed Test II and for the nine students with low CIA, and explain how you would document the decisions.
Read the worked answer
The two dengue cases need a compensatory assessment with an equivalent blueprint, not the same paper: same objectives, same weight, different items, conducted before the CIA is forwarded. Certificates should be verified and the date of intimation recorded, because fairness depends on the rule being available to all students, not on the illness alone. For the nine low scorers, the defensible remedy is instructional, not numerical: two remedial sessions targeting the weak units, followed by one retake of the lower of the two tests, with the retake capped at 70% of that test's weight and the option announced to the whole class in advance.
On the assignment allegation, publish the rubric retrospectively, have a second examiner blind-mark a random 20% sample, and re-mark the full set if discrepancies exceed two marks. Documentation should include the policy note, dates, eligibility criteria, approval signatures, a version history of the marks register, and student acknowledgement of the compensatory scheme. No individual rounding, no undocumented grace, and a written reply to the parent citing the rubric and the second-marking outcome.