Did you know that the longest whisker in a box plot does not always reach the largest observation? A learner may see a point far beyond the whisker and think the chart has lost an item. In fact, some box plots deliberately put unusually distant observations outside the whiskers as separate dots. That changes how the picture should be read.
A box plot summarises a numerical distribution with quartiles and a median. In a simple five-number-summary box plot, the whiskers can reach the minimum and maximum. In a common modified box plot, whiskers instead reach the smallest and largest observations that are not flagged by a stated outlier rule. The excluded-from-whisker observations remain visible as separate points. Always check which convention is used.
Diagnose the misunderstanding before calculating
Show a pupil a box extending from 4 to 9 minutes, a median line at 6.5, a right whisker ending at 10 and a separate dot at 30. Ask, ‘Does the largest recorded value equal 10 or 30?’ If the learner answers 10, diagnose a misunderstanding about the key; if the learner chooses 30 but cannot explain the dot, the next step is to work through the fence rule.
Every number below is an invented teaching example, not a pupil-performance report. Our imaginary observations are the minutes ten pupils need to assemble a small kit. We use the median-of-halves quartile method consistently. Different legitimate software quartile conventions can produce slightly different boundaries for small datasets.
Hand-check ten observations, every step and unit
Start with the sorted values 2, 3, 4, 5, 6, 7, 8, 9, 10, 30 minutes. There are ten observations. The median is the average of the two middle values, 6 and 7, which is 6.5 minutes. The lower half is 2, 3, 4, 5, 6; its median is Q1 = 4 minutes. The upper half is 7, 8, 9, 10, 30; its median is Q3 = 9 minutes.
The interquartile range is Q3 − Q1 = 9 − 4 = 5 minutes. Using a 1.5 × IQR modified box plot, the lower fence is 4 − (1.5 × 5) = −3.5 minutes and the upper fence is 9 + (1.5 × 5) = 16.5 minutes. Those are calculation thresholds, not necessarily observed values.
Only the value 30 exceeds the upper fence. The actual smallest observation still inside the fences is 2, and the actual largest inside them is 10. So the modified whiskers end at 2 and 10 minutes. A separate dot records the 30-minute observation. The box still extends from 4 to 9, and its median is still 6.5. Check every endpoint against the original list before drawing.
- Sorted dataset: 2, 3, 4, 5, 6, 7, 8, 9, 10, 30 minutes.
- Median: (6 + 7) ÷ 2 = 6.5 minutes.
- Quartiles: Q1 = 4 minutes; Q3 = 9 minutes; IQR = 5 minutes.
- Fences: −3.5 minutes and 16.5 minutes.
- Modified whiskers: observed values 2 and 10 minutes; separate mark: 30 minutes.
Why a different-looking box plot can show the same data
A different graphing convention may draw whiskers directly to the minimum and maximum. Under that definition the right whisker could reach 30. The entire dataset remains unchanged; the visual rule differs. Comparing whisker lengths across charts without checking the respective conventions can produce a false conclusion. Do not call either graph wrong merely because they differ.
A flagged point is a reason to look more carefully, not an automatic measurement mistake. Perhaps a kit was missing a component, perhaps someone paused, or perhaps the longer time is entirely legitimate. The box plot alone cannot identify the cause. It certainly does not give permission to erase the observation. The US National Institute of Standards and Technology’s box-plot guide explains the modified-whisker and 1.5 × IQR approach used here.
A practical five-question chart-reading method
- What variable is shown, and in what units?
- What do the box edges and interior line signify?
- Do the whiskers show actual minimum and maximum, or values inside a fence?
- Are separate dots flagged observations, or does the figure plot every individual measurement?
- Are the same definitions and axis scales used when comparing two groups?
The sequence matters. It prevents the habit of announcing that one group has a higher average merely because a whisker sticks farther out. A box plot reports a distribution’s structure; the arithmetic mean is not necessarily displayed at all. A bigger IQR indicates a wider central half, but it does not prove that every individual outcome is farther apart.
Progressive practice: guided, supported and independent
Guided: Sort these fictional completion times: 5, 6, 7, 8, 9, 10, 11, 12, 13, 32 minutes. The median is (9 + 10) ÷ 2 = 9.5. Q1 is 7; Q3 is 12; IQR is 5. The fences are −0.5 and 19.5. Whiskers are 5 and 13; plot 32 as a separate observation. Ask the learner to explain each mark rather than simply copy the result.
Supported: Replace that last 32 with 14, then repeat the calculation using the same quartile convention. Q1 is still 7 and Q3 still 12, so the upper fence remains 19.5. This time the maximum 14 is within the fence. The right whisker reaches 14 and no outlier dot is needed. The median remains 9.5, despite the large change in the greatest value.
Independent: Given 1, 2, 3, 4, 5, 6, 7, 8, 9, 27 minutes, draw the modified box plot without seeing an answer key. The self-check is median 5.5, Q1 3, Q3 8, IQR 5, lower fence −4.5, upper fence 15.5, whiskers 1 and 9, and a separate dot at 27. The pupil must articulate why the maximum is not the right whisker endpoint.
Common misconceptions and what repairs them
- ‘The whisker always gives the maximum.’ This is true only under some drawing conventions; read the key.
- ‘An outlier must be deleted.’ No. Investigate its validity before deciding what the number means.
- ‘A longer whisker means a higher mean.’ Not necessarily. The whisker is an endpoint, not the arithmetic average.
- ‘Any separate dot is a mistake.’ It might be a valid observation, or a chart may deliberately show individual dots.
- ‘Different applications must calculate the same quartiles.’ Compare their quartile definitions, especially for small samples.
How families and teachers can observe progress
A teacher can write the ten values on index cards and ask pupils to sort, halve, calculate and draw the chart physically. A parent can present an unfamiliar box plot and ask what its lone dot means. The learner’s task is to describe the convention in a full sentence before computing anything. Keep the dataset small enough to hand-check, then move to a larger spreadsheet-generated plot.
Meaningful improvement is visible when a pupil identifies the correct maximum, computes an IQR and fence with units, keeps an unusual observation in the evidence unless a justified exclusion rule applies, and explains the limitations of the summary. Those are transferable habits for reports, research projects and daily data claims.
FAQs and limitations
Do all Singapore school box plots use the 1.5 × IQR convention?
Do not assume that. The article explains two mathematical conventions rather than asserting an examination requirement. Follow the precise definition supplied by a teacher, textbook, question or software.
Can the box plot reconstruct every original value?
No. Unless individual observations are also shown, many raw datasets can share the same quartiles and whisker endpoints.
Does a separate point mean the mean should exclude it?
Not automatically. A statistical outlier flag and a data-cleaning decision are different actions; any removal needs a defensible reason.
Why may two applications disagree over Q1?
Quartile algorithms differ. State the convention used, check the sample size and do not merge outputs from incompatible methods without explanation.
Continue the learning
For another distribution lesson, read why classes with the same mean can have different spreads. The Mathematics Learning Hub and eduKate Education Hub offer wider routes. The Clementi Secondary 1 Mathematics reference illustrates the diagnosis-first and independent-checking style behind effective teaching.
The core lesson is not to memorise a picture. It is to know what each mark measures, check it against actual observations and explain where the chart stops being evidence.