Did you know? Two classes can have averages of 80 and 65 marks, yet the combined average for all their pupils need not be 72.5. If one class has twice as many pupils as the other, a simple average of the two class averages gives the smaller class more influence than each of its pupils should have.
A combined mean is the total of all eligible individual scores divided by the total number of eligible scores. When you know only each group’s mean and group size, recover each group’s score total by multiplying mean × number of scores, add those totals, then divide by the combined number. For a pupil-level mean, group sizes are the weights. The immediate answer is: do not average group means equally unless the groups are equally large or your question genuinely gives equal importance to the groups.
This is the difference between a calculation that merely looks tidy and one that answers the real question. The core aim of education is to teach learners to define who or what is being averaged, justify the denominator, check the arithmetic and explain the limits of the result to a reader.
First ask: an average of pupils or an average of classes?
Suppose a school report says ‘What is the average quiz mark among all pupils who took the quiz?’ Each pupil is one observation, so each pupil contributes one score. A group with twenty-four pupils contributes twice as many observations as a group with twelve. That leads to a group-size-weighted mean.
A different question, ‘What is the average of these two class averages, treating each class as one unit?’ deliberately gives each class equal weight. That answer can be useful for a different reporting purpose, but it is not the mean score of all pupils when group sizes differ. Neither number should be labelled vaguely as ‘the school average’ without stating its basis.
The NIST Engineering Statistics Handbook defines an arithmetic mean as the sum of data values divided by their count. The practical question here is how to reconstruct that sum when you have grouped information instead of one long score list.
Worked example: two fictional classes and one common quiz
Imagine two fictional groups sat the same twenty-question quiz, reported as a mark out of 100 after a common marking scheme. Every pupil in the example completed it, and each class mean is exact—not a rounded summary. These are teaching numbers, not a claim about real children or schools.
- Class A: 12 pupils, average mark 80/100.
- Class B: 24 pupils, average mark 65/100.
- Question: What is the average mark for all 36 pupils who took this same quiz?
Step 1: reconstruct the total in each class
Class A’s combined marks are 12 × 80 = 960 marks. Class B’s combined marks are 24 × 65 = 1,560 marks. It does not matter that the class sizes are different; the totals preserve how many pupil scores each mean represents.
Step 2: add compatible totals and counts
Total marks = 960 + 1,560 = 2,520 marks. Total pupils tested = 12 + 24 = 36 pupils. Divide: 2,520 ÷ 36 = 70 marks per pupil. The correct combined mean is 70/100, not 72.5/100.
Step 3: check the answer without repeating the same calculation
The combined mean must lie between 65 and 80 because it is made from scores in those two groups. It should sit closer to 65 because more pupils are in Class B. The result, 70, lies five marks above 65 but ten below 80. That directional check supports the result; it does not replace the arithmetic.
A second independent calculation uses proportions: Class A makes up 12/36 = 1/3 of the tested pupils; Class B makes up 24/36 = 2/3. The combined mean is (1/3 × 80) + (2/3 × 65) = 80/3 + 130/3 = 210/3 = 70. Both routes agree.
What goes wrong when you average averages?
The tempting shortcut is (80 + 65) ÷ 2 = 72.5. This gives the mean of the two class means, counting each class once. If the principal explicitly wanted to treat each class as one equal unit, 72.5 would answer that different question. For the 36 individual pupils, it overweights the smaller class.
Imagine another school where one class has one pupil scoring 100 and another class has ninety-nine pupils averaging 50. The equal-class mean would be 75, but the student-level combined mean would be (1 × 100 + 99 × 50) ÷ 100 = (100 + 4,950) ÷ 100 = 50.5. The dramatic difference makes the denominator problem visible.
This extreme illustration is mathematical, not a realistic recommended class arrangement. It teaches a principle: an average is inseparable from the population and weighting rule that define it.
The missing-pupil trap: enrolment is not the same as completed scores
Suppose a report lists 24 pupils in Class B, but four were absent for the quiz. If the displayed Class B mean of 65 was calculated using only the 20 actual quiz-takers, multiplying 65 by the enrolment of 24 is wrong. The legitimate pupil-level weight for that observed mean is twenty, not twenty-four.
Keep Class A at twelve completed quizzes averaging 80, but take Class B as twenty completed quizzes averaging 65. Total recorded marks now are 12 × 80 + 20 × 65 = 960 + 1,300 = 2,260. There are 12 + 20 = 32 observed scores. The mean of those recorded scores is 2,260 ÷ 32 = 70.625/100, which can be reported as approximately 70.6/100 to one decimal place when that rounding is suitable.
This second scenario is a separate illustrative data set, not a correction to the first. The scores of four absent pupils are unknown and must not silently become zero. If the objective is to describe all enrolled pupils rather than those tested, the available data may not be sufficient.
The distinction echoes counting attendance with the correct eligible sessions: a meaningful numerator and denominator must describe the same population.
Before calculating, diagnose the data
Ask the student to inspect the question rather than race to a formula. What exactly is one observation—a pupil, a class, a paper or an assessment attempt? Were all scores out of the same maximum? Were they obtained under comparable assessment conditions? Are the reported means exact, or have they already been rounded? How are absences, retakes and missing results treated?
- Equal-size groups: averaging group means equally happens to produce the pupil-level mean, but the learner should still know why.
- Different-size groups: use the number of eligible scored observations as weights for a pupil-level mean.
- Different maximum marks: raw marks should not be pooled as if they represented the same scale. Converting to percentages may align numerical scales but does not make unequal tests equally difficult.
- Missing or excluded records: do not treat unknown marks as zero unless the published scoring rule explicitly assigns that value.
- Rounded group means: reconstructed totals are approximate. Do not claim an exact combined mean with more precision than the input allows.
- Multiple attempts by one pupil: decide whether the statistic is about attempts or individual pupils; they lead to different counts.
Progressive practice: from guided totals to a defensible report
Guided practice: make totals visible
Give the learner two boxes: ‘number of pupils’ and ‘mean mark’. Using the 12 and 24 pupil example, have them draw 12 and 24 small symbols or write a multiplication beside each group. Ask, ‘What total could have produced this class mean?’ Reconstruct the marks, then divide once by the combined count. Use a second route to check.
Supported practice: change the sizes
A fictional Class C has 10 quiz-takers averaging 76/100; Class D has 30 quiz-takers averaging 84/100. The learner works through (10 × 76 + 30 × 84) ÷ (10 + 30) = (760 + 2,520) ÷ 40 = 3,280 ÷ 40 = 82/100. The simple average of class means would be 80/100. Ask the learner to explain which answer addresses each question.
Independent practice: write the rule and defend the denominator
Try this without a model solution: eight learners have an average of 75/100, and twelve have an average of 85/100 on a common quiz. What is the combined mean? After calculating, justify why the denominator is 20 and why neither 8 nor 12 can be ignored.
Self-check: 8 × 75 = 600; 12 × 85 = 1,020; total 1,620; total learners 20; combined mean = 81/100. The result lies between the group means and nearer 85 because the larger group has the higher mean. If you obtained 80, you averaged two group means rather than all pupil scores.
Transfer: spot a misleading school-report headline
Suppose a newsletter says ‘Our average improved by ten marks’ but changes from an average across equal-weighted classes to an average across all pupils. Even if the underlying marks were unchanged, a change of weighting could alter the number. Ask the learner to request class sizes, scoring basis and both calculation methods before treating the headline as evidence of improvement.
A parent and teacher checklist for responsible reporting
- Students: label every group size and score unit; show total marks before computing the final mean.
- Parents: ask what the reported statistic includes, not just whether it rose or fell; avoid using it to judge an individual child.
- Teachers: state the scoring scale, eligible population, attendance treatment, and whether the published value is a mean of pupils or a mean of class means.
- School projects: put a one-sentence methods note below the table: ‘Combined mean weighted by the number of pupils who completed the common quiz.’
Observable progress is a learner who can choose the right denominator from words, detect when an average-of-averages is unsuitable, identify missing data honestly, and independently defend the result. A correct numerical answer without the right population is not yet reliable statistical literacy.
Limitations worth saying aloud
A combined mean compresses many scores to one number. It does not show the spread, whether some pupils improved, what topics were difficult, the effects of access to support or whether two quizzes were equally demanding. Small groups may differ in ways unrelated to teaching. Treat a combined class mean as one description, not proof of any intervention’s success.
If scores are rounded, based on different assessments, or collected using different marking criteria, first reconcile the underlying definitions. In a real school comparison, follow the institution’s authorised privacy and reporting practices; never publish identifiable individual scores.
Frequently asked questions
When is it correct to average the two class averages?
When both classes have equal numbers of relevant scored observations, an equal average of their means is also the pupil-level mean. It is also correct if the question deliberately treats each class as one equally weighted unit. State which question you are answering.
Can I just convert every result to a percentage?
You can convert compatible score scales, but percentages do not automatically control for different question difficulty, marking standards or missing work. Mathematical scale agreement is not assessment equivalence.
What if all I know is the two class means?
Without class sizes, you cannot uniquely calculate the combined pupil-level mean when the groups might differ in size. Ask for the number of eligible scores or state the limitation.
Is the combined mean a mark that a particular pupil received?
Not necessarily. It is a summary of the collection. No pupil need have earned exactly that score.
Where to learn the connected skills
For another denominator question, read how attendance denominators change when pupils join mid-term. For transparent marking criteria, see how to handle a rubric criterion that does not apply. The Secondary 2 G3 Mathematics lesson on means from frequency tables explores the mathematics further.
Visit the Core Aim learning hub and eduKate Education Hub for connected guidance. The immutable Secondary 1 Mathematics Clementi tutorial reference shows why learners benefit from diagnosing the question, explaining the method, practising progressively and checking without hints. The lasting aim is to choose a mathematically correct statistic and to say truthfully what it means.