Assessment of Learning and Evaluation
Assessment of Learning and Evaluation
This domain tests whether you can choose an instrument, judge its quality, and use results to help learners rather than merely rank them. Expect items on purposes, norm- versus criterion-referencing, item analysis, and the meaning of validity and reliability.
Core rules
- Purposes: diagnostic (before teaching, to find gaps), formative (during, feedback, assessment for learning), summative (end, judgement, assessment of learning), ipsative (learner compared with own past performance), placement.
- Referencing: norm-referenced compares a learner with peers (useful for selection); criterion-referenced compares against a standard (useful for mastery). Ghana's standards-based curriculum favours criterion-referencing.
- Technical qualities: validity (measures what it claims), reliability (consistency across time and markers), fairness, practicality, washback (the effect of the test on teaching).
- Table of specifications ensures content coverage and a deliberate spread of Bloom's levels.
- Item analysis: difficulty p = number correct ÷ number tested; discrimination D = p(upper group) − p(lower group). Ideal p is about 0.5–0.6; D of 0.3 and above is acceptable; a negative D means the item rewards the weaker group - revise or discard.
- Feedback: specific, prompt and forward-looking ('add the unit', 'support your point with evidence'), not just a score.
Worked example
An item is answered correctly by 30 of 50 learners, so p = 0.60. In the upper 10 scorers 8 were correct; in the lower 10 scorers 2 were correct, so D = 0.8 − 0.2 = 0.60. Reasoning: difficulty is moderate and discrimination exceeds 0.3, so the item separates strong from weak learners and should be retained in the bank.
Traps that cost marks
- Calling a test valid because it is reliable - reliability is necessary but not sufficient.
- Using formative assessment only to award grades.
- Treating 'assessment of learning' as if it covered all assessment.
- Computing difficulty from the top group only, or reversing the discrimination formula.