Designing assessment — a crash course

In constructive alignment, the intended learning outcomes, the teaching activities, and the assessment tasks are designed together so they reinforce one another. When the assessment rewards something other than the outcome, students learn that something else.

• Write the outcomes first, then design tasks that make students do exactly what those outcomes describe.
• Check each assessment against an outcome — drop anything that assesses nothing you intended.
• Make sure the teaching gives students genuine practice at what the assessment will ask of them.
• Watch the verbs: if an outcome says ‘evaluate’, don’t assess it with recall.

Validity is whether a task actually measures the learning you care about; reliability is whether it does so consistently across students and markers. A task can be one without the other — design for both.

• Ask of every task: would a strong answer here really demonstrate the outcome?
• Use clear criteria and a rubric so judgements are consistent, not idiosyncratic.
• Moderate or double-mark high-stakes work to check markers are applying standards the same way.
• Reduce construct-irrelevant barriers — confusing instructions, needless time pressure — that distort what you’re measuring.

Assessment of learning certifies; assessment for learning develops. Formative tasks, used during a course, let students and teachers see where understanding stands while there is still time to act.

• Build low-stakes formative checkpoints before the work that counts.
• Share criteria and exemplars so students can judge their own work against the standard.
• Use the results to adjust your teaching, not only to record a mark.
• Sequence tasks so feedback from one genuinely feeds the next.

Authentic assessment asks students to do something resembling real practice in the discipline, which raises engagement and is harder to fake. Programmatic thinking then spreads assessment sensibly so neither students nor staff are overwhelmed.

• Design tasks around real problems, audiences or products, not just an exam by default.
• Map assessment across the whole programme to catch over-assessment and bunched deadlines.
• Vary task types so you assess a range of outcomes and reduce easy academic-integrity shortcuts.
• Prune redundant tasks — fewer, better-aligned assessments often teach more than many small ones.