Seven Failures on the Same Two Fields
A veterinary technology program admits 24 second-year students to a hematology course. The practical examination requires each student to prepare a Wright-stained smear from an EDTA canine blood sample, perform a 100-cell differential, and identify neutrophils, lymphocytes, monocytes, eosinophils, and platelets. The pass mark is 75%, and the differential contributes 40 of 100 points while smear preparation contributes 10. Seven students scored between 62% and 72%. All seven misidentified the same two fields at the feathered edge, counting smudge cells as lymphocytes; those two fields were the only ones with a readable monolayer because the students' smears were too thick. One of the seven has an approved 50% extra-time accommodation that the proctor never applied because the roster did not reach her. The course coordinator proposes adding 5 points to all seven scores, calling the smears 'bad.' The program director asks you to review the examination before grades post.
1. Is this a learner performance problem, an assessment design problem, or both? What evidence would you gather before deciding, and what does the rubric weighting tell you?
Read the worked answer
Both, and the distinction matters because the remedy differs. Start by separating the two constructs the exam claims to measure: smear preparation and differential accuracy. The rubric gives 40 points to the differential and 10 to preparation, but if thick smears left only the feathered edge readable, then a preparation failure contaminated the differential score. A student cannot demonstrate monolayer cell identification on a slide that has no monolayer; counting smudge cells as lymphocytes at the feathered edge is a predictable artifact, not necessarily a knowledge deficit.
Evidence to gather: (1) blinded re-score of all 24 smears by a second examiner using the same rubric; (2) frequency of the same error among passing students; (3) item-level difficulty and discrimination; (4) whether the 75% pass standard was set by a defensible method such as a modified Angoff panel or simply chosen; (5) the accommodation record and the proctor roster; (6) retained slides or photographs, if policy allows. If passing students also misread feathered-edge fields but had readable monolayers, the task is sound and the seven share a learning gap. If the error tracks with smear thickness, the assessment is confounded and the scores are not interpretable as intended. The coordinator's +5 is arbitrary and unfair to students near the boundary; decide on evidence, not sympathy.
2. Recommend a defensible outcome for the seven students, including the student with the accommodation and the grading decision. Justify with assessment principles.
Read the worked answer
Treat the accommodation case separately and first. That student was denied a legally approved accommodation through an administrative failure. Do not simply add time to the original score; offer a supervised retest of the same competencies with 50% extra time and an equivalent, validated smear, and document the accommodation and the remedy.
For the other six, the decision depends on the evidence from Task 1. If the assessment was confounded, do not curve. Either void the differential portion and re-assess the affected competency with instructor-prepared smears, or apply a pre-specified correction that is available to all students, not only the seven. A blanket +5 to seven students changes the standard after the fact and disadvantages students who passed. If the assessment was sound, provide structured remediation: review smudge cells versus lymphocytes, practice at the monolayer, a formative quiz, and a parallel practical after remediation. Keep the 75% standard; a retest is not a lower standard if it samples the same blueprint. Build a record: rubric, second examiner, remediation plan, retest result, and appeal route. Communicate in writing to all 24 students so the process is transparent and defensible.
3. Redesign the practical so that the same competencies are measured without this confounding. Give concrete structural changes.
Read the worked answer
Redesign around construct validity and constructive alignment. Split the exam into two stations: a preparation station scored with a checklist (wedge angle, monolayer present, feathered edge, staining quality, no holes) and a differential station using pre-validated, instructor-made smears that all students read under the same conditions. A thick smear can then no longer destroy the differential score.
Use an analytic rubric with explicit descriptors for each cell line, and score smudge cells separately. Set the pass standard before the exam using a modified Angoff panel of three instructors and document it. Increase sampling: two smears and two 100-cell counts, because a single 100-cell differential has substantial sampling error. Add a station where students explain a discrepant manual versus analyzer platelet estimate, which tests integration rather than recall. For accommodations, send an automated roster to proctors 48 hours ahead and designate a backup proctor. Pilot the exam on volunteers, run item analysis, and revise items with negative discrimination. Finally, teach the feathered-edge artifact explicitly before the exam so the error becomes a learning target rather than a hidden trap.