Blueprinting a 70-Mark Paper in Your Discipline
You are a member of the Board of Studies for a four-credit core course in your discipline at a state open university. The course has five units with approved weightage 20:25:25:15:15. The terminal examination carries 70 marks and internal assessment 30. About 1,200 learners take the course each cycle across 40 study centres. Last cycle's paper had: Section A, ten 2-mark short answers, all recall, drawn only from Units 1 and 2; Section B, four 10-mark questions out of six, one of which had two defensible answers and had to be double-keyed; Section C, one compulsory 10-mark question that only 8% of learners answered correctly. Item analysis showed a Section A mean of 17.8 out of 20, and Units 4 and 5 contributed no marks at all. The Vice-Chancellor's office has asked the Board to fix the paper design before the next cycle, and the internal assessment is being criticised for grade inflation: mean 27.4 out of 30, with 71% of learners scoring above 26.
1. Prepare a blueprint for the 70-mark paper showing unit weightage and cognitive-level distribution, and explain the three design faults in the current paper that the blueprint corrects.
Read the worked answer
Blueprint (marks, unit × cognitive level). Unit marks approximate the approved 20:25:25:15:15 weightage after rounding to whole marks: 14, 18, 17, 10, 11.
| Unit (marks) | Remember | Understand | Apply | Analyse | Evaluate/Create |
|---|---|---|---|---|---|
| 1 (14) | 4 | 4 | 4 | 2 | 0 |
| 2 (18) | 2 | 4 | 6 | 4 | 2 |
| 3 (17) | 2 | 3 | 4 | 5 | 3 |
| 4 (10) | 0 | 2 | 3 | 3 | 2 |
| 5 (11) | 0 | 2 | 4 | 3 | 2 |
| Total (70) | 8 | 15 | 21 | 17 | 9 |
Lower-order (Remember + Understand) = 23 marks (33%); Apply 21 (30%); Analyse 17 (24%); Evaluate/Create 9 (13%).
Section structure: A - 10 × 1 = 10 (lower-order only); B - 5 × 4 = 20 (out of 7); C - 3 × 6 = 18 (out of 5); D - 2 × 11 = 22 (out of 4). Total 70.
The three faults it corrects.
Content validity. Units 4 and 5 carry 30% of the weightage but supplied zero marks, so the paper did not sample the syllabus and learners rationally skipped a third of it. Every unit now has a minimum floor.
Cognitive range and discrimination. Twenty marks of recall one-liners with a mean of 17.8/20 barely separate candidates, while the single compulsory high-level question was answered by 8% - the paper jumps from trivial to impossible with nothing in between. The blueprint builds a graded ladder.
Fairness and reliability. A compulsory, un-scaffolded long question plus a double-keyed item made the score depend on the paper's defects rather than on the learner's competence. Choice at the higher levels and a moderation gate remove that.
2. At moderation, one item has two defensible answers and another carries 12 marks on Unit 4, whose share is about 10 marks. What do you do, and who signs off?
Read the worked answer
Double-keyed item - treat it as two different situations.
Before the examination: it cannot go to print unresolved. The two subject moderators either agree a single key with a written justification, or the item is replaced with a blueprint-equivalent item (same unit, same level, same mark value) from the bank. The replacement is recorded against the blueprint line it serves.
After the examination: accept both keys in the marking scheme and re-scan every script for that item; do not silently drop the item, because that changes the total and re-weights the rest of the paper. If the error surfaces only after results, re-evaluate affected scripts and reissue. Any decision is minuted with the reasons.
The over-weighted Unit 4 item. The item is not invalid in itself - an individual question may legitimately draw several marks from one unit. The fault is at paper level: if Unit 4 takes 12 of 70 marks, another unit is starved, and the paper no longer matches the approved 20:25:25:15:15 weightage. Fix the rest of the paper, or reduce the item to 10 marks and redistribute 2 marks to the under-represented unit. Also check whether a 12-mark item sits at the right cognitive level for a unit of only 10 marks - a long high-level item on a light unit is usually a setters' convenience.
Sign-off. Two internal moderators (subject and language), one external subject expert, and the Board chair. The moderation minutes must map every item to its blueprint cell - unit, level and marks - so that a later audit can see the paper was checked against a design and not merely proofread.
3. Redesign the 30-mark internal assessment for 1,200 distance learners so that it is credible and less inflation-prone.
Read the worked answer
Design against the inflation, not against the learners. A mean of 27.4/30 with 71% above 26 is not evidence of a strong cohort; it is evidence that the components do not discriminate. Four components, each with a published analytic rubric:
- Assignment 1 (8 marks) - short structured answers, tutor-marked on accuracy, use of evidence and structure. One resubmission allowed after feedback, capped at 60% of the component, so feedback is used without rewarding endless retries.
- Assignment 2 (10 marks) - an applied task in the discipline: a case analysis or a field-based problem, 800–1,000 words, plagiarism-checked. The rubric weights application and justification, not length or presentation.
- Component 3 (7 marks) - a viva or practical at the study centre, or a timed online quiz with items drawn at random from a bank at least five times the size of the administered set.
- Component 4 (5 marks) - a moderated asynchronous discussion or reflective log; verifiable, low-bandwidth, and impossible to complete by copying.
Controls. Double-mark a random 10% sample across centres; compare centre means and re-mark any centre more than one standard deviation above the pooled mean; cap quiz-type marks at roughly a quarter of the internal total; and report internal and terminal means side by side to the Board each cycle. Rubrics go to learners in advance - transparency reduces disputes and makes a low mark defensible.