The Test That Did Not Match the Class
A primary school has decided to replace its continuous classroom assessment with a single written end-of-term test for children in their second year of English (age 7). The test lasts 35 minutes and contains 40 items: gap-fill sentences, a short reading passage with comprehension questions, and a writing task of about 60 words. There are no pictures. Results were low across the class; two children cried during the test and one refused to start. The teacher's own records - observation notes, a portfolio of children's drawing-and-labelling work, and notes from oral pair activities - show that most of these children understand and use the term's target language when speaking and listening in class. The school reports the test to parents as a percentage and uses it to decide progression to the next level.
1. Critique this test as an assessment instrument for this class, commenting on validity, reliability and appropriacy.
Read the worked answer
Validity - it does not measure the learning it is supposed to measure. The term's work, judging by the teacher's records, has been largely oral and receptive; the test is almost entirely a literacy test. A child can know the language and score near zero, which is exactly what the teacher's records suggest happened. A 60-word writing task at age 7 with two years of English is measuring handwriting stamina and control of written conventions as much as language.
Reliability - one occasion, one format. A single 35-minute sitting gives a single sample of performance from children whose attention, mood and confidence vary enormously across a day. Two children cried and one refused to start: their scores tell us about the testing situation, not their English. Forty items marked as a percentage also implies a precision that a 7-year-old's single performance cannot support.
Appropriacy - the format is wrong for the age. No pictures, dense print, no oral component, no familiar task types: the test asks children to decode an unfamiliar format under time pressure while also testing language. That produces anxiety and, for the weakest, refusal.
Consequences. The score is reported to parents as a percentage and used to decide progression, so an invalid measure is doing a high-stakes job. The washback risk is that teachers start preparing 7-year-olds for gap-fill instead of teaching them English.
2. The head teacher argues: 'A test is objective; teacher assessment is just opinion.' How would you respond, using this case?
Read the worked answer
I would separate two claims: that a test is standardised, and that it is therefore more valid. This test is standardised in format but not valid for what was taught, and an invalid measure is not made objective by being written down.
'Teacher assessment is just opinion' overstates the case. The teacher here is not offering an impression: she has observation notes, a portfolio of children's work collected over time, and records from oral pair activities. That is evidence, gathered on many occasions, in the conditions in which the language is actually used - which is precisely where a one-off written test is weakest.
What teacher assessment genuinely needs is moderation, not replacement: agreed criteria, a shared checklist across classes, and a colleague sampling the portfolio against them. That addresses the legitimate worry about bias without discarding the richest evidence available.
The case demonstrates the point: the test says these children have failed; the teacher's evidence says they can use the language. Where two sources disagree that sharply, the sensible conclusion is that one of them is measuring the wrong thing - and it is the 40-item gap-fill.
3. Design a more appropriate end-of-term assessment for this class and explain how the results should be used and reported.
Read the worked answer
Collect several sources over time, not one event:
- Observation checklist completed during normal lessons over the last three weeks, with a small number of 'can-do' descriptors (understands classroom instructions; names six animals; asks 'Have you got...?').
- A short oral pair task with picture prompts - two children, five minutes, teacher as observer: this assesses the speaking and listening the term actually targeted, and takes the same total time as a written test when run across a week.
- Portfolio of drawing-and-labelling and one or two short pieces of writing, dated, so progress is visible rather than a single snapshot.
- A child-friendly self-assessment - a smiley-face 'can-do' sheet read aloud by the teacher, kept alongside the teacher's records but not treated as the judgement.
- If a written component is required, keep it short, pictured, and in a format the children have met in class - not gap-fill under time pressure.
Moderation: two colleagues sample the portfolios against the shared descriptors so results are comparable across classes.
Reporting: to parents as can-do statements with examples of the child's work, plus a next-step comment - not a percentage. For progression decisions, use the full set of evidence, with the oral task and portfolio weighted at least as heavily as any written sample. Use the results diagnostically: if a descriptor is unmet across the class, that is information about the teaching, not only about the children.