Accountable by Design — Highveld's Assessment Redesign

A four-round, advanced higher-education management simulation set inside Highveld School of Business, an accredited MBA provider in Pretoria, South Africa (≈2,600 students, ≈ZAR 480M revenue, a ~280-student flagship core course). In February 2026 a CHE/internal quality-assurance review formally flags the flagship 'Managing for Results' course: ~70% of the grade comes from a single final group project, so individual learning can't be evidenced. Pass rate is a 'healthy' 88% and completion 92%, but moderators call the result 'an artefact of group work, not evidence of individual competence' — and 41% of students report free-riding while peer-evaluation data (explaining only ~8% of variance) is collected but never used in grading. You have one term and a hard envelope: an extra ZAR 1.6M and ~600 marking/support hours per cohort, an 8-week QA deadline, or the course's NQF-credit standing goes 'under review'. Playing the programme leadership team, you (1) diagnose where individual accountability disappears — map each learning outcome to whether it is assessed at the individual or group level and name the validity gap; (2) rebalance the assessment blueprint across individual exams/vivas, individual applied/reflective work, weighted peer evaluation and a reduced-weight group project, trading validity, accountability-clarity, marking cost and completion risk inside the 600-hour budget; (3) design accountability checkpoints — an early diagnostic, a mid-course individual checkpoint, contribution tracking and an early-alert path — so a struggling individual becomes visible in week 3, not week 12; and (4) allocate the ZAR 1.6M support budget to the at-risk bottom quartile and weak competencies, then defend the redesign to the QA/accreditation panel. The math rewards constructive alignment, validity-per-marking-hour, staged early checkpoints and targeted support — and punishes the five classic errors: a single high-stakes final exam that collapses completion (92%→74%), bolting peer scores onto heavy group weighting, a beautiful 1,400-hour blueprint the faculty can't deliver, an all-summative scheme with no early checkpoints, and spreading scarce support evenly instead of where the risk concentrates. Final KPIs track Validity, Accountability Clarity, Completion % and marking-hour load.