Find your next learning step
Choose the question closest to what you need today, or read the guide from the beginning.
If you want an equal-share amount, the mean is useful. If you want the middle of an ordered numerical dataset, use the median. If you want the most frequent value or category, use the mode. These summaries can disagree without any calculation being wrong, because they answer different questions.
For the fictional values 2, 3, 3, 4 and 18, the mean is six, the median is three and the mode is three. Calling six the average is mathematically legitimate if average means arithmetic mean. Calling it a typical observed value without further explanation may hide the fact that four of the five values are below six.
This shared G1/G2/G3 Mathematics guide develops the choice as well as the calculation. The official 2027 G1 Mathematics syllabus includes the purposes and use of mean, mode and median. Examples are invented teaching datasets, not reports about actual students or communities. Schools determine the appropriate sequence and level of extension.
Contents
Sections 1–5
Sections 6–10
Sections 11–15
Sections 16–20
Sections 21–25
Sections 26–28
The arithmetic mean is the total of the numerical observations divided by their number. The median is the central value after ordering, using the mean of the two middle numerical values when there is an even number of observations. The mode is the value or category with the highest frequency.
Choose according to the information needed. A total reconstructed from a group mean uses the mean and group size. A middle ranked observation uses the median. The most common selected category uses the mode. A summary cannot be chosen reliably from the word average alone when the context leaves its meaning unclear.
Also inspect the data type, weights, extremes and what the dataset actually represents. A mean of numerical category codes may be meaningless. A mean of two group means may need group-size weights. A median can hide extremes. No single summary preserves every feature of the original data, so describe its purpose and limitations honestly.
Before calculating, translate the task into a concrete request. Does it ask for a fair equal share of a total, the middle observed amount, the most common choice, or a comparison of whole distributions? Those are different information needs.
Suppose five fictional boxes contain 2, 3, 3, 4 and 18 counters. If all thirty counters are redistributed equally, each box receives six. That is the mean’s equal-share interpretation. It uses every counter in the total.
If the question asks for the middle box count after ordering, the answer is three. If it asks for the count occurring most often, the answer is also three. Their numerical agreement in this dataset does not make median and mode the same definition.
Write the target before choosing a formula. This prevents a familiar arithmetic routine from taking over a question that requires interpretation. The first decision is what information matters; the calculation then produces the chosen summary. A correct calculation of the wrong summary still does not answer the original question.
For numerical observations, add all values and divide by the number of observations. With values 4, 6, 8 and 10, the total is 28 and the mean is seven. The dataset does not contain a seven, but seven is the amount each would have under equal redistribution of that total.
The mean need not be an observed value. It can also be fractional when observations are whole-number counts. Counts 1, 2 and 4 have mean 7/3. No observation contains one third of a counted object; the fraction describes the total per observation when expressed as an average.
That is not a reason to round a mean automatically to a possible individual count. If the question asks for the exact mean, 7/3 is appropriate. If it asks for a specified decimal accuracy, round accordingly. An actual allocation requiring whole objects is a different task.
The mean’s connection to the total is especially useful in reverse problems. Mean × number of observations = total. Keep the group size visible so that a mean of a small group is not mistaken for a total or treated as interchangeable with the same mean of a much larger group.
To find the median, order the observations. For 9, 2, 5, 3 and 7, the ordered list is 2, 3, 5, 7, 9. The middle value is five. The original third entry was five here as well, but that coincidence should not replace sorting.
For 12, 1, 8, 3 and 6, the ordered list is 1, 3, 6, 8, 12, so the median is six rather than the original third entry eight. The median concerns rank after ordering, not the original recording order.
With an even number of numerical observations, use the mean of the two central values. For 2, 4, 7 and 11, the median is (4 + 7)/2 = 5.5. Like the mean, a median need not be one of the observed values.
The median does not identify the total. Different datasets can share the same median while having very different sums. It is useful when the central rank is the intended summary, but it cannot replace the mean in a calculation that relies on total = mean × count.
The mode is the value or category occurring most often. In 2, 4, 4, 5 and 9, the mode is four because it occurs twice while each other value occurs once. The modal value is four; its frequency is two. Do not confuse those two answers.
Mode can also describe categorical data. If fictional activity choices are art, music, art, sport and art, the modal category is art. There is no need to assign numerical values or calculate an arithmetic mean of the activity names.
Some datasets have more than one equally most frequent value. In 1, 1, 3, 3 and 8, both one and three are modes under the usual school convention. If all listed values occur once, school exercises commonly describe the data as having no mode; follow any convention explicitly specified by the task.
The mode does not have to be central or close to the mean. It answers a frequency question. A most-common shoe size, activity or response can be useful for a particular planning task, but it does not by itself describe the full distribution or determine how many of every category are needed.
Five fictional observations are 2, 3, 3, 4 and 18 minutes. Calculate the mean, median and mode, then explain what each reports. Begin by noting that the list is already ordered and contains five observations.
The total is thirty minutes, so the mean is 30/5 = 6 minutes. The middle, third observation is three minutes, so the median is three. The value three occurs twice, more often than any other value, so the mode is also three minutes.
If the task asks for average time per observation using total time divided by count, six is the required summary. If it asks for the central recorded time or most frequent recorded time, the relevant answers are three and three for different reasons.
The value eighteen pulls the mean upward because it contributes fully to the total. The median remains at the third ordered value. This does not make the mean wrong or the median universally better. It shows that the summaries respond differently to the same data. Choose the one that answers the intended question, and mention the unusual high value when it matters to the interpretation.
The table makes the different jobs explicit. None of these measures is a complete description of the dataset. Knowing one summary generally leaves many possible underlying lists.
| Summary | Main question answered | Important limitation |
|---|---|---|
| Mean | What is the numerical total per observation? | Can be strongly affected by extreme values |
| Median | What is central in the ordered numerical list? | Does not reveal the total or extremes |
| Mode | Which value or category occurs most often? | May be tied, absent under school convention, or unrepresentative of other values |
| Range | How far apart are the minimum and maximum? | Not an average and does not show the middle |
Use the table as a decision aid rather than a ranking. If equal sharing is the question, sensitivity to every value is part of the mean’s purpose. If central rank is the question, the median’s different response can be useful. If the data consists of unordered category names, the mode may be the natural relevant summary.
The limitation column does not mean the measures are unreliable calculations. It identifies information each calculation leaves out. A strong interpretation states both what was summarised and what would require returning to the full dataset or another representation.
In the five-time dataset, eighteen is much larger than the other values. Do not delete it merely because it changes the mean. First ask whether it is a genuine recorded observation, a transcription error or outside the defined dataset.
If it is genuine and the task asks for the mean of all five observations, it belongs in the total. Removing it would change the question and the denominator. A different summary can be reported alongside the mean if the purpose requires describing the centre more fully.
For the first four values 2, 3, 3 and 4, the mean is three. Adding eighteen raises the five-value mean to six. That contrast shows the effect of the added observation, but it does not establish that the observation should be removed.
In a school exercise, use the supplied data unless instructed otherwise. In a real investigation, unusual values may deserve checking against the original record. The important habit is to justify inclusion and exclusion through the data definition and evidence, not through a preference for whichever summary gives a more comfortable number.
If the ordered values are 2, 3, 3, 4 and 18, replacing eighteen with 180 leaves the median at three while increasing the mean substantially. This is a useful example of the median’s reduced sensitivity to an extreme magnitude change.
However, adding or removing observations can change the central position. The median of 2, 3, 3, 4 and 18 is three, while the median after adding 20 is (3 + 4)/2 = 3.5. An extreme added value can therefore change the median through the count and rank structure.
Do not turn the useful phrase resistant to extremes into the false rule that the median never changes when extreme values are involved. Specify whether an existing value is being replaced, a new observation added or an old observation removed.
The distinction matters when explaining a comparison. The median may be a suitable central summary for a skewed list, but that suitability comes from the question and distribution, not from an absolute immunity to change. Keep the original ordered positions visible when analysing how a revised dataset affects it.
A fictional shop records item quantities 1, 1, 1, 4 and 8 across five transactions. The mode is one, but the total quantity is fifteen and the mean quantity is three. The most common transaction size does not determine the total amount sold.
If the question asks what quantity occurs most frequently, mode one is the right answer. If it asks for quantity sold per transaction on average, the arithmetic mean three is relevant. If it asks how much stock was used, calculate the total fifteen.
All three values describe the same dataset in different ways. A modal transaction can be common without accounting for most of the total quantity, because less frequent large transactions may contribute substantially to the sum.
This is why context words need interpretation. Common, typical, middle and average can be used loosely in everyday speech. In Mathematics, identify the precise operation required or explain which meaning you are using. Do not let a vague word replace the information about what the decision actually needs.
Suppose a survey records preferred activities as art, music and sport. The mode can identify the most frequently selected activity. An arithmetic mean of those category names is not defined in the same sense as a mean of numerical measurements.
Assigning codes such as art = 1, music = 2 and sport = 3 does not automatically make the mean code meaningful. The numbers are labels. Renaming the codes could change their mean without changing any respondent’s preference.
For example, choices art, art, sport have code mean 5/3 under the first coding. If art is coded ten and sport thirty, the mean code becomes 50/3. Neither result identifies an average activity between art and sport. The arithmetic is possible, but the interpretation is not justified by the coding.
Counts and measured quantities are different because their numerical operations have defined meanings. Before calculating a mean, ask what adding and dividing the values would represent. A string of digits in a data table is not enough; numerical labels and numerical quantities can look similar while supporting different conclusions.
For values 0, 2, 4 and 6, the mean is 12/4 = 3. Zero is an observation and contributes to the count even though it adds nothing to the total. Omitting it from the denominator would give four, which describes a different three-value dataset.
A missing observation is not automatically zero. If a table says a value was not recorded, replacing it with zero adds information that was never supplied. The appropriate treatment depends on the question’s instructions or the data-collection rules.
Likewise, frequency zero means no observations in that category; it is not a recorded value of zero. A frequency table needs the number of observations found by summing frequencies, not by counting the number of rows.
This denominator check often matters more than the addition. Ask how many actual observations the total represents. Write that count beside the sum before dividing. A mean is a ratio between a total and its corresponding observation count; using a denominator from another group or from the table layout breaks that relationship.
A fictional frequency table records value two with frequency three, value five with frequency two and value nine with frequency one. There are six observations, not three. The total is 2 × 3 + 5 × 2 + 9 × 1 = 25, so the mean is 25/6.
The mode is two because it has the highest frequency, three. The ordered list is 2, 2, 2, 5, 5, 9, so the median is (2 + 5)/2 = 3.5. The table contains enough information to reconstruct these numerical summaries exactly.
Averaging the three row labels would give (2 + 5 + 9)/3 = 16/3. That treats each distinct value as occurring once and ignores the frequencies. It answers a different question about the distinct labels, not the recorded dataset.
Keep value and frequency in separate roles. The product value × frequency contributes to the total; the frequency alone contributes to the number of observations. This structure also explains the mode and median positions. Reading a frequency table is not a new set of unrelated rules; it is a compressed way of storing the underlying list.
Group A has two observations with mean ten. Group B has eight observations with mean twenty. Their totals are twenty and 160 respectively, giving a combined total of 180 across ten observations. The combined mean is eighteen, not fifteen.
The simple mean of the two group means, (10 + 20)/2 = 15, gives each group equal weight rather than giving each observation equal weight. It is appropriate only for a different question or under conditions such as equal group sizes.
For equal group sizes, the simple mean of group means does match the combined observation mean. For unequal sizes, reconstruct the totals using mean × count, add those totals and divide by the combined count. This route makes the weighting explicit.
Do not average medians in the same way and expect a combined median. Group medians and group counts generally do not reveal the full ordered combined list. A summary compresses information; once some information is lost, a later calculation may need the original data rather than another operation on the summaries alone.
The datasets 5, 5, 5 and 5 and 1, 3, 7 and 9 both have mean five. Their medians are also five. Yet the first has range zero and the second has range eight. The matching central summaries do not make the datasets identical.
If consistency matters, a centre measure alone may be insufficient. At this stage, comparing the ordered lists and ranges can reveal the difference. More advanced spread measures can be added when they are part of the learner’s programme.
The range is maximum minus minimum, not an average. It uses only the extremes and can miss how values are distributed between them. A range of eight does not tell you whether most observations cluster near five or near the endpoints.
For a comparison question, name the feature being compared. Same mean, different range is a precise statement. One group is better is not supported unless better has been defined in relation to a purpose and the relevant data. Mathematical summaries should support a stated comparison rather than quietly supply a value judgement the question never established.
For the values 1, 2, 2, 5 and 10, find the equal-share amount, middle ordered value and most frequent value. State which summary answers each request before doing the arithmetic.
The total is twenty across five observations, so the equal-share amount is the mean four. The middle ordered value is the median two. The most frequent value is the mode two. Their agreement in the last two answers is numerical, not definitional.
Now ask which statistic determines the total if the count is known. The mean does: 4 × 5 = 20. The median two or mode two alone would not determine that total. Different datasets with those same summaries could contain different large values.
For a final check, replace ten with fifteen. The mean becomes five because the new total is 25. The median and mode remain two. This controlled change shows which information each summary uses. Do not interpret the unchanged median as proof that nothing important about the data changed; the total and upper extreme both changed.
A fictional club records favourite activities: reading, art, reading, sport, music, reading, art. What is the modal activity, what is its frequency and would a mean of category codes be a sensible answer to the favourite-activity question?
Reading is the modal activity and occurs three times. Art occurs twice, while sport and music each occur once. A numerical mean of arbitrary activity labels would not identify a meaningful average preference. The category frequencies answer the question directly.
If the task instead asks what fraction chose reading, the answer is 3/7. That is a proportion of respondents under the stated one-choice setup, not a mean activity. Different data questions can require different numerical summaries even within one survey.
Keep the observational unit clear. If multiple selections were allowed, the total number of selections might differ from the number of people. The denominator for a share of selections would then need to match that whole. The mode could still identify the most frequent category, but it would not by itself reveal how many distinct people chose combinations of activities.
Values one, four and seven have frequencies two, three and one respectively. Find the mean, median and mode of the six observations. Begin by reconstructing the list if the frequency-table method is not yet secure.
The list is 1, 1, 4, 4, 4, 7. Its total is 2 + 12 + 7 = 21, so the mean is 21/6 = 3.5. The middle positions are the third and fourth, both four, so the median is four. The mode is four because its frequency is three.
The frequency three is not the mode’s value. The row label four tells you which observation is most common; three tells you how often it occurs. This distinction is easy to lose when the table is scanned quickly.
Check the mean by multiplying 3.5 by six, recovering the total 21. Check the median by locating positions rather than averaging the first and last values. Check the mode by comparing frequencies rather than choosing the largest numerical value. Each summary has a different verification route because each definition uses different information.
A fictional group of three learners has mean score twelve on a defined classroom task. Another group of seven has mean score eighteen on the same task and scale. Find the combined mean across all ten observations.
The group totals are 3 × 12 = 36 and 7 × 18 = 126. The combined total is 162, so the combined mean is 162/10 = 16.2. It lies between twelve and eighteen and is closer to eighteen because the larger group has that mean.
The simple mean of the group means is fifteen, which gives equal weight to the two groups. It does not give equal weight to the ten individual observations. The total-and-count method makes that distinction visible without needing a separate memorised weighted-average formula.
This combination assumes the scores describe comparable quantities on the same scale. If the two means concern different tasks with different possible totals, combining them without an explicit interpretation may not answer a meaningful question. Correct arithmetic does not establish comparability. Define what the combined average is intended to represent before using the calculation.
For 1, 2, 3, 4 and 100, the median is three, but only one observation equals three. Median means central ordered value, not the value held by most observations. The mode is a different frequency concept.
Even a mode need not be a majority. In 2, 2, 3, 4 and 5, the mode is two because it occurs more often than any other individual value. However, only two of the five observations equal two, which is less than half.
Use most frequent for the largest individual frequency and majority for more than half of the relevant whole. These phrases can describe different conditions. A modal category in a survey does not automatically represent most respondents in the majority sense.
This language check can prevent an overstatement after a correct calculation. Reporting the mode as two is sound; claiming most observations are two may be ambiguous or false depending on how most is meant. A clear answer can state the frequency alongside the mode, making the strength of the claim visible to the reader.
For a nonempty finite list of equally weighted real numerical observations, the arithmetic mean lies between the minimum and maximum. If every value is at least the minimum and at most the maximum, their total divided by the count cannot lie outside those bounds.
This provides a useful rejection check. If the values are 4, 7, 9 and 12, a calculated mean of 32 is impossible. The correct total is 32, but the mean is eight. Forgetting to divide by the count produced a value outside the range.
Passing the range check does not prove a mean is correct. A mistaken denominator might still produce a number between the minimum and maximum. Check the total and observation count as well as the plausible magnitude.
The median also lies within the numerical range, and a mode must be an observed value under the usual definition. These checks are simple but different. Use them after applying the definitions, not as substitutes for the definitions. A plausible summary can still be the wrong statistic or the result of an arithmetic slip.
Everyday language often asks for a typical amount without specifying what typical should mean. If the dataset is fairly balanced and the purpose is numerical total per observation, the mean may communicate well. If extreme values dominate the total while the question concerns a central ranked observation, the median may be useful.
For categorical choices, the mode answers which category is most common. But a mode with a small share of the whole should not be described as representing everyone. Include the frequency or proportion if that distinction matters.
There is no need to turn every short school question into a long debate. If it explicitly asks for the mean, calculate the mean. If it asks you to choose and justify a suitable average, explain the purpose and one relevant feature of the data.
A good justification is specific: I use the median to describe the middle recorded waiting time because one unusually long wait makes the mean much larger than most observations. A weak justification is median is always best when there is an outlier. The first connects purpose with evidence; the second overstates a rule.
A correctly calculated mean of one class’s recorded observations describes that dataset. It does not automatically establish the mean for every class or the whole school. Generalising requires information about how the observations were collected and what population they represent.
Likewise, a modal preference in a small voluntary survey does not prove that the same category is most popular among everyone who did not respond. The arithmetic may be exact while the broader conclusion remains unsupported.
At this level, the practical habit is to name the dataset in the answer. Write the mean of these ten recorded values is 16.2, rather than silently turning the calculation into a universal claim about students. This keeps the statement aligned with the evidence supplied.
For open-ended data work, ask who was included, what was measured, whether values are missing and whether the collection method matches the question. Those issues arise before choosing a summary. A sophisticated average cannot repair a dataset that answers a different question or excludes the people the conclusion is meant to describe.
Ask the learner to explain mean, median and mode without using only their formula names. Look for total per observation, central ordered value and most frequent value. Those descriptions reveal whether the definitions are separate before a larger dataset is introduced.
Then provide one list where all three agree, such as 2, 3, 3, 3 and 4. Its total is fifteen, so all three summaries are three. Follow with 2, 3, 3, 4 and 18, where the mean changes to six while median and mode remain three.
The contrast prevents accidental agreement from hiding confusion. If a learner gives three for every average in the second list, inspect whether the same operation is being reused for different definitions. If the calculations are right but the chosen summary is not, ask what the question needs to know.
Finish with a categorical dataset and a frequency table. These variations check data type and observation count, not just arithmetic. A small set with deliberate changes can identify whether the next useful repair concerns definitions, ordering, frequencies, weights or interpretation of the final sentence.
For the broad topic foundation, read The Core Aim of Mathematics Mastery: Mean, Median, Mode and Range. This guide’s specific job is choosing a summary that fits the question, not simply calculating several statistics from a list.
For the larger role of data and evidence, use How Mathematics Works: Statistics. For a cross-subject application lens, see Median and Mode in Science. The same definitions matter when data comes from an experiment rather than a Mathematics worksheet.
Choose the next page according to the gap. A learner who understands the definitions but miscounts observations needs frequency practice. A learner who calculates accurately but calls every result typical needs interpretation practice. A learner who combines groups incorrectly needs the link between mean, count and total.
The transferable habit is to ask what a summary preserves and what it discards. That question supports reading charts, comparing groups and evaluating numerical claims without assuming that a single average can stand in for the whole dataset.
First identify the observational unit and the question’s purpose. Check whether the values are numerical quantities or category labels. Choose the requested or justified summary, then apply its definition using the correct observations, order and frequencies.
For the mean, verify total and count and check that the result lies within the range. For the median, sort and locate the central position or pair. For the mode, compare frequencies and distinguish the modal value from its frequency. If combining groups, recover totals before calculating a combined mean.
Then interpret the result in a sentence. State which dataset it describes and what the summary answers. If an extreme value, tie, missing observation or unequal group size affects the interpretation, mention that feature rather than hiding it behind the word average.
Finally decide whether another view is needed. A central summary may need a range or the full list when spread matters. A categorical mode may need its proportion when support is divided. The aim is not to report every possible statistic, but to provide enough relevant information for the actual mathematical question.
If the definitions blur together, compare equal sharing, ordered middle and highest frequency using a small list. If sorting is the difficulty, practise median positions separately. If the denominator is wrong, reconstruct the actual observations or sum the frequencies before dividing.
If group means are being averaged without weights, return to mean × count = total. If a summary is calculated correctly but overinterpreted, ask what the dataset includes and what the statistic does not reveal. These are different repairs and should be chosen from the working shown.
Use the G1/G2/G3 Mathematics explainer directory for connected data and checking guides. The Clementi Secondary 1 Mathematics reference describes the wider teaching emphasis on precise diagnosis, clear explanation and manageable next steps.
Keep the final principle simple: an average is useful when its definition matches the information needed. Mean, median and mode are not rival answers competing to be universally best. They are different views of data, and choosing responsibly means knowing which view the question actually requires.
These answers address the interpretation choices that often remain after the formulas have been learned.
Is the median always better when there is a large value?
No. It may better describe a central ranked observation, but the mean remains appropriate for total per observation and equal-share questions. A genuine large value belongs in the dataset unless there is a justified reason to exclude it. State the purpose of the summary.
Can a mean be a number that nobody recorded?
Yes. Values four and eight have mean six even though six was not observed. A mean is the numerical total divided by count, not necessarily an individual observation. A fractional mean of whole-number counts can also be meaningful as an average.
Can I average two group averages?
For a combined observation mean, use group-size weights unless the groups have equal sizes. Multiply each mean by its count, add the totals and divide by the combined count. Group medians generally do not provide enough information to reconstruct a combined median.
Is the mode’s frequency the mode?
No. The mode is the value or category occurring most often. Its frequency tells you how many times it occurs. If value four occurs three times and no other value occurs as often, the mode is four and its frequency is three.
Continue the G1/G2/G3 Mathematics series
All G1/G2/G3 Mathematics explainers
What Is G1 Mathematics | Why a Negative Temperature Is Not a Temperature Fall
What Is G2 Mathematics | Why Cancellation Needs a Factor of the Whole Numerator
What Is G3 Mathematics | Why the Semicircle Right Angle Needs a Diameter
G1 Mathematics | Fraction of the Remainder and the Original
G2 Mathematics | Inequality Direction After Negative Division