Did you know that a student can score 70% on one practice test after scoring 80% on another and still be making useful progress? The smaller number gets all the attention, but the two papers may be asking very different things. Before declaring that learning has gone backwards, look at the questions behind the percentages.
Comparing practice-test scores fairly means checking whether the papers assess comparable skills, difficulty, conditions and marking standards. When these differ, raw percentages alone cannot establish improvement or decline. Use question-level evidence first, then a small set of matched follow-up questions to see what the learner can now do independently.
Why the same percentage can represent different work
Suppose two ten-question mathematics quizzes are both marked out of ten. Quiz A contains eight routine calculations and two unfamiliar applications. Quiz B contains four routine calculations and six applications that require choosing a method. Both have ten questions, but they do not sample the same demands. The number of marks is a denominator; it is not a guarantee that the tasks are equivalent.
Different scores may also reflect a change in topics, time limits, level of teacher hints, access to notes, or partial-credit rules. Even the same question can function differently after the student has seen a near-identical worked solution. A parent who asks only, “Did your percentage rise?” misses this context.
Worked example: the apparent drop from 80% to 70%
Here is a fictional record, not a real student’s results. On Quiz A, a learner answers all eight routine questions correctly but misses both applications: 8 correct out of 10, or 80%. On Quiz B, they answer all four routine questions correctly and three of the six applications: 7 correct out of 10, or 70%.
The overall score drops by 10 percentage points. Yet the application record changes from 0/2 to 3/6. That is a useful diagnostic clue, not proof of growth: the samples are small and the application questions may differ in complexity. The correct interpretation is “we need a closer comparison”, not “the student has become weaker” or “the student has definitely improved”.
For a follow-up, the teacher prepares two short sets of new questions, each deliberately balanced across the same three target skills and similar demand. On the first set, suppose the learner solves three of six independently; on a later comparable set, five of six. This supports a cautious claim of better performance on those sampled skills. It still does not prove future examination results, and a few questions cannot establish a stable trend.
A five-part comparison that parents can actually use
- Match the skill first. Put fraction equivalence with fraction equivalence, inference with inference, and explanation with explanation; do not compare a calculation quiz with a reasoning paper as though they were the same test.
- Look at demand. Note which tasks only require recall, which need a choice of method and which combine several ideas. A rough teacher judgement of difficulty is helpful, but it is not a scientifically calibrated test scale.
- Record conditions. Was the task timed? Were hints, notes, calculators or worked examples available? Mark any material differences.
- Inspect individual errors. Separate misunderstanding from copying slips, missing information, poor reading, or an unfinished question. Each calls for a different repair.
- Set a modest independent recheck. Use unseen questions at similar demand, mark with the same criteria, and check explanations as well as correct answers.
This approach protects both students who are discouraged by harder work and students whose scores rise because practice has become easier. A higher percentage on easier questions is still work completed, but it cannot by itself show the next level has been mastered.
What the wrong answers can reveal
A learner may succeed on familiar fraction addition but fail when the denominator is not supplied. Another may know a science definition yet struggle to connect it to an unfamiliar observation. Those are not simply “a 70% problem”. Ask the child to mark where they stopped, show their first attempted step and explain what choice they were trying to make.
For practical follow-up, choose one missed skill. Teach or revisit the underlying idea, show one carefully explained example, practise with help, then remove the hint. A later question with different surface details shows whether the child can select the method rather than merely replay the original answer. This mirrors the diagnosis-first, progressively independent approach in eduKateSG’s Clementi mathematics teaching guide.
A useful home conversation, without ranking the child
Instead of “Why did you lose ten marks?”, try “Which questions were harder this time, and what did you manage without help?” Let the student select one old error and one new independent attempt. Parents do not need to create a score-equating system or become examiners: their useful role is to preserve the question papers, notice changes in conditions and ask whether the learner understands the correction.
Teachers can use a short skill-by-skill tracker: topic, demand, first attempt, type of error, help received, fresh attempt and date. If several recent attempts under comparable conditions show the same weakness, instruction should change. If a one-off score drops on a much more demanding paper, treat it as information about a new challenge rather than a verdict on the student.
Mistakes to avoid when interpreting practice marks
- Calling two papers equivalent merely because they are both worth 20 marks.
- Treating a single difficult item as a precise measure of a whole skill.
- Repeating the identical questions until the percentage looks better, then labelling that independent transfer.
- Combining timed and untimed results into one trend without noting the different conditions.
- Promising a particular grade or exam outcome from a short run of practice scores.
How do we know the next month is going better?
Look for a pattern: the student identifies the correct method without a chapter hint, makes fewer of the same conceptual errors, can explain a fresh answer, and retains the skill on a later return. Do not chase daily percentage changes. A small set of comparable tasks over time is more informative than one dramatic high or low mark.
For related guidance, see reading test scores as skill evidence and the Study & Learning Methods Hub.
Frequently asked questions
Should we ignore percentages altogether? No. They are useful summaries when the assessment context is known. Keep the score, but read it alongside the task and the learner’s reasoning.
Can my child compare two school examinations? You can compare topics and recurring errors, but different papers may not be directly equivalent. Use the school’s marking and feedback as given; do not invent a conversion between tests.
What if there are no comparable questions? Record what is known, avoid a firm progress claim and ask the teacher for a small sample of new questions targeting the same learning goal.
The real aim: stronger independent work
A practice score should start a useful conversation, not end it. The goal is to discover what the learner can explain, apply and check without a cue. Once the comparison is fair, the next teaching decision becomes much clearer. For wider parent and student reading, return to the eduKate Education Hub.