Did you know a blank cell can tell the truth more clearly than a perfectly continuous graph? In a school experiment, one reading may be missing because the class was away, the sensor battery failed, or a student forgot to note a value. The tempting response is to enter zero so the graphing software will draw something. Unfortunately that single shortcut can turn a missing measurement into a false scientific claim.
The immediate answer: a missing observation means no usable measured value is available for that time or category. It is not the same as a measured zero. On a line graph of regular observations, leave a gap at the missing time and label the reason where known. Do not create a zero dot, invent a measurement, or join across a gap without disclosing the assumptions. The UK Office for National Statistics line-chart guidance explicitly recommends gaps where otherwise regular data are missing.
- Write a genuine reading as a number with its unit; record missing data as missing, not as zero.
- Keep all planned observation times on the x-axis so the missing day remains visible.
- For regular observations, do not draw a data point at a day that was not measured.
- Distinguish a solid line connecting observations from a dashed estimate or interpolation.
- Calculate a mean only over the observations actually included, and state the denominator.
- Always keep an unedited measurement log and explain what the gap does and does not show.
The question hidden inside the chart
A line graph shows how a quantity relates to another ordered variable, often time. Each plotted point should correspond to a real recorded value or be clearly labelled as an estimate. A point at y = 0 means something specific: the measured quantity had a value of zero in the stated units. An empty observation instead says, ‘We do not know this value from the available evidence.’ These statements are not interchangeable.
The difference matters across subjects. Zero rainfall in a correctly observed period means no measured rain fell in that interval; a missed rain-gauge visit means the rainfall amount for that interval is unknown. A temperature of 0°C is a possible physical temperature; an unrecorded thermometer reading is not a temperature of 0°C. A pupil who simply puts zero in every blank changes the meaning of the dataset.
Worked example: a fictional five-day water study
Imagine a fictional classroom that checks the temperature of a water container at 8 am each school day. The planned observations are Monday to Friday. The logged values are Monday 22°C, Tuesday 24°C, Wednesday missing, Thursday 26°C and Friday 25°C. All recorded values are invented for explanation; this is not a real school experiment. Assume each valid observation was measured using the same thermometer and method.
Write the data table with columns ‘day’, ‘time’, ‘temperature (°C)’ and ‘status’. On Wednesday, leave the temperature field empty or enter an explicit missing-value code in a separate status column, depending on the software. Use a phrase such as ‘not observed; class away’ if that was truly the reason. Do not place the number 0 in Wednesday’s temperature column.
Draw the horizontal axis with all five days and a vertical axis with a labelled temperature range. Plot the four known values. Where Wednesday would lie, show a gap, a labelled missing marker without a y-value, or a clear annotation. The exact design should make it impossible for a reader to mistake Wednesday for a measured temperature. A continuous straight line from Tuesday’s 24°C to Thursday’s 26°C can visually imply an unobserved path; without a stated interpolation policy, keep the break visible.
An ordinary average becomes misleading if a blank is silently changed
The four observed values add up to 22 + 24 + 26 + 25 = 97°C when written as the numerical sum of four readings. The average of those four observed daily temperatures is 97 ÷ 4 = 24.25°C. This is not automatically the mean for all five days because Wednesday has no measurement. Say which days are included.
If someone replaces Wednesday with 0°C, the reported five-day average becomes 97 ÷ 5 = 19.4°C. That calculation is arithmetically correct for the newly entered five numbers, but it answers the wrong question because the zero is fabricated. The 4.85°C difference between those two means is not a discovered cooling event. It is an artefact of mislabelling missing data.
What about estimating Wednesday’s value?
A student might suggest halfway between Tuesday and Thursday: (24 + 26) ÷ 2 = 25°C. This is an interpolated estimate under a straight-line change assumption, not an observed Wednesday measurement. If a teacher asks for an estimated five-day mean and makes the assumption explicit, then (22 + 24 + 25 + 26 + 25) ÷ 5 = 24.4°C. Label that result estimated, not measured.
There is no guarantee the temperature rose steadily from Tuesday to Thursday. The container might have cooled, warmed or been moved. An interpolation is useful for some defined analytical tasks, but it is not permission to rewrite a real observation log. Report both the assumption and its limitation. When the missing interval is long, when sudden changes are plausible, or when a result is important, a simple straight-line interpolation may be especially poor.
Learn to spot the mistake before drawing
- Diagnostic one: If the log says ‘no usable reading’, is 0°C an observation? No.
- Diagnostic two: If a graph programme connects a missing Wednesday automatically, has it measured Wednesday? No.
- Diagnostic three: Does every drawn segment mean every intermediate temperature was measured? No; it shows a graphical relationship or assumption.
- Diagnostic four: Can you calculate the mean of four known values? Yes, but label its four-day denominator.
- Diagnostic five: Can you claim the temperature peaked on Thursday? Not definitively; an unobserved Wednesday could have been higher.
This five-question diagnosis often identifies the precise misconception: students believe a visually complete graph is automatically a scientifically complete record. A teacher can ask the learner to explain what evidence each point represents. If they cannot locate a raw observation for a dot, they should remove or relabel it.
Guided practice: choosing the right display
Now imagine a different fictional experiment recording the number of seeds that germinated each day: Monday 3, Tuesday 0, Wednesday not counted and Thursday 4. On Tuesday a careful count produced zero newly germinated seeds. Plot zero because it was measured. On Wednesday the count was not made. Leave it missing. Two visually similar low or empty spaces can have very different meanings. The presence or absence of an observation, not the convenience of the graph, determines which is which.
For time-series data, the ONS guidance recommends showing markers when intervals are irregular so readers can see when a value actually exists. For regular time series with missed readings, a gap is the safer default. In specialist analysis, interpolation, model-based reconstruction and imputation may be justified, but those processes require documented assumptions and a distinction from the original observations.
Independent challenge: a weather club’s notebook
A fictional weather club has daily midday readings: 29°C, 30°C, missing, 31°C and 30°C. Draw a five-day axis and plot the four available values. Compute the observed mean: (29 + 30 + 31 + 30) ÷ 4 = 30°C. Would a plot at 0°C on day three be honest? No. Could you claim the maximum temperature over the five days was 31°C? You can say that the highest observed reading was 31°C, but the missing day prevents a firm claim about the maximum for all five days.
Try a second version with the middle value explicitly recorded as 0°C. Now it is a real data point and should be plotted, subject to checking whether the instrument and conditions make sense. Explain the difference using the evidence log, not just the picture. This is the independent performance that shows the concept has transferred.
What families and teachers can do next
Parents can ask their child to distinguish three phrases aloud: ‘zero measured’, ‘not measured’ and ‘estimated’. Teachers can prepare identical-looking spreadsheet templates with different source logs and ask students to choose a display. A good explanation names the status of each value, keeps the raw log intact and records the effect of the missing day on any calculation.
That instructional sequence matches the Clementi Mathematics teaching reference: diagnose an incorrect assumption, explain the underlying rule, practise a concrete example, then check independently. The progress indicator is visible: the student produces a chart whose dots can each be traced to an observation, and explains one honest limitation. No particular test outcome is guaranteed.
Limitations, exceptions and honest uncertainty
A line chart may be used in contexts where segments are understood as a visual guide between actual observations, rather than proof that every intervening value was observed. Conventions differ across disciplines, and some continuous measurements can justify connected traces even when samples are intermittent. The author must decide how to represent missingness responsibly, tell readers the rule and not conceal important gaps. For a student learning data integrity, visible gaps and explicit status labels are a sound default.
A missing datum may not be random: a sensor may fail only in extreme heat, or a pupil may miss measurements when it rains heavily. In such cases, the observed mean can be biased and simply dividing by the number of known points does not solve that deeper problem. Don’t claim that the observed mean represents every day equally without checking why data went missing.
Frequently asked questions
Should an empty spreadsheet cell be entered as 0?
No, unless a real observation was zero. Check how the particular spreadsheet or graphing tool treats blanks; some tools may connect points or hide a series, so configure and explain the gap correctly.
Can I draw a dotted line across missing days?
Only if its meaning is clear. A dashed or dotted line can signify an estimate or an assumed connection; label it so nobody mistakes it for observed measurements.
Does a gap mean the experiment failed?
Not necessarily. A complete and truthful account of missing data can be stronger than a seemingly perfect but invented sequence. Explain the source of the gap and improve the collection plan.
Can I fill in Wednesday later using Thursday’s reading?
You cannot retrospectively observe Wednesday by measuring on Thursday. A later model can estimate the earlier state under assumptions; that estimate should remain separate from the original log.
Continue learning
For spreadsheet problems that hide observations, see How to Check a Chart When Spreadsheet Rows Are Hidden. For honest interpretation of what a chart supports, visit the eduKate Education Hub. The Office for National Statistics provides practical line-chart rules for irregular and missing data.
The lasting lesson: an unknown number is not zero, and a drawn line is not a replacement for an observation. Good education teaches us to leave the right space blank and explain why. That is how a student learns to trust evidence without manufacturing certainty.