A table shows four temperature readings:
| Time / min | Temperature / °C |
|---|---|
| 0 | 70 |
| 5 | 65 |
| 10 | |
| 15 | 58 |
A pupil looks at the blank cell and writes 0°C.
The graph now plunges from 65°C to 0°C and rises to 58°C.
The graph is dramatic.
The evidence is false.
A blank is not zero. “Not measured” is not a measurement. Missing evidence must remain missing until a real observation fills the gap.
This guide belongs to the Primary 4 Science Learning Hub. It develops data-record integrity: how to label missing, unreadable, not-recorded and excluded values so that the table says what actually happened.
The existing Missing Information, Unknowns and What Cannot Be Concluded guide owns the reasoning question “What can I infer when information is absent?” This page owns the record-making question: how should the absence itself be represented in data?
Quick Answer: The Data-Gap Loop
EXPECTED OBSERVATION → DID WE ACTUALLY MEASURE? → IF YES, RECORD VALUE → IF NO, LABEL THE REASON → NEVER SUBSTITUTE ZERO → PRESERVE EXCLUSION REASON → GRAPH ONLY REAL DATA → BOUND THE CONCLUSION
This is an eduKate teaching routine, not an official MOE examination formula.
Wait, What? Four Empty-Looking Cells Can Mean Four Different Things
A table cell can be empty-looking because:
- the value was never measured;
- the instrument reading was unreadable;
- the observation was missed;
- the result was deliberately excluded from one analysis after a documented method failure.
Those are not the same as:
- a measured value of zero;
- zero change;
- no detected difference.
Science becomes clearer when those meanings stay separate.
1. What does zero mean?
Zero is a numerical result.
Examples:
- 0 cm change in measured height;
- 0 mL added;
- 0 s elapsed at the defined start;
- 0 on a stated relative sensor scale if the device actually reports zero.
Zero must come from a defined measurement or calculation.
It is not the default meaning of an empty cell.
2. What does blank mean?
A blank means only that no value is written in the cell.
It does not explain why.
A good scientific record should avoid ambiguous blanks when the missing observation matters.
Instead use a short code or note such as:
- NM = not measured;
- NR = not recorded;
- UR = unreadable;
- EX = excluded from this comparison, with reason in notes.
These codes are eduKate teaching examples, not official MOE symbols. A class may use different labels as long as the meanings are explicit.
3. Not measured
At 10 minutes, the pupil forgot to take the temperature.
The correct entry is not 0°C.
It is:
NM — not measured.
This preserves the truth of the record.
4. Not recorded
The pupil remembers looking at the thermometer but did not write the value.
Later, they cannot remember whether it was 61°C or 62°C.
Do not invent the most likely number.
Record:
NR — not recorded.
Memory is not a substitute for the original measurement.
5. Unreadable
A photograph contains a blurred measuring cylinder scale.
The liquid level is visible, but the exact value cannot be read.
Do not estimate a precise number merely because the table expects one.
Record:
UR — unreadable from the image.
The photograph can still provide other evidence.
6. Excluded result
A trial spills half the water before the final measurement.
The final temperature is still recorded.
Should it appear in the comparison of equal-volume cups?
Probably not, because the trial no longer matches the intended method.
But do not delete it.
Keep:
- the raw value;
- the spill note;
- the reason for excluding it from that specific comparison.
This is an exclusion with provenance, not a missing observation.
7. Zero change is derived, not missing
Start height = 15 cm.
End height = 15 cm.
Derived change:
15 − 15 = 0 cm.
This zero is supported by two measurements.
A blank end-height cell cannot produce the same conclusion.
8. No detected change is not the same as zero raw data
A ruler records 15 cm at both times.
The derived change is 0 cm at the measurement resolution.
The interpretation may be:
no detectable height change.
Do not write a raw height of “0 cm” unless the measured height itself was actually zero.
9. A missing reading should stay a gap in a graph
Suppose the values are:
- 0 min = 70°C;
- 5 min = 65°C;
- 10 min = missing;
- 15 min = 58°C.
Do not plot 0°C at 10 minutes.
Do not invent 61°C unless a justified method explicitly estimates it—and at Primary 4, the safest teaching approach is usually to leave the observation missing.
The graph can show the real measured points and the table can mark the gap.
10. Connecting points does not create a missing observation
A line drawn from 65°C at 5 minutes to 58°C at 15 minutes passes through many possible intermediate values.
Those line positions are not measurements taken at 10 minutes.
The display may help show a trend, but it must not be described as recorded data.
11. Missing data and time-lapse frames
Batch 20’s Photographs, Video and Time-Lapse Observation teaches that a missing frame should not be replaced by duplicating another image.
The same rule applies here:
presentation should not manufacture observations.
12. Missing data and data loggers
An automatic logger skips one timestamp.
Possible reasons include:
- logging pause;
- storage problem;
- power interruption;
- programme behaviour;
- communication failure.
Do not convert the missing record to zero.
Batch 20’s Data Loggers, Sensors and Automatic Measurements develops device-specific checks.
13. Missing data and raw/derived evidence
A missing raw value can prevent a derived value from being calculated.
Example:
- start temperature = 70°C;
- end temperature = missing.
Temperature decrease cannot be calculated.
Do not write 70°C decrease by treating the blank as zero.
Use Raw Data, Derived Values and the Evidence Trail.
14. Missing data and no-change results
Table:
- Day 1 = 15 cm;
- Day 2 = blank.
Can we say growth = 0 cm?
No.
We do not know Day 2 height.
Zero change requires two comparable measured values.
15. Missing data and small differences
A value is present but too imprecise to distinguish two cases.
That is not missing data.
It is limited-resolution data.
Different problem, different repair.
16. Missing data and excluded data
Missing:
observation does not exist in the record.
Excluded:
observation exists but is not used in one analysis for a stated scientific reason.
This distinction is essential.
17. An excluded trial should remain visible somewhere
Good record:
| Trial | Final temp / °C | Status | Reason |
|---|---|---|---|
| 1 | 58 | Use | Method followed |
| 2 | 42 | EX | Half water spilled before reading |
| 3 | 57 | Use | Method followed |
The reader can see the evidence and the decision.
18. Exclusion rules should exist before seeing which result is convenient
If the class says after the experiment:
“We will exclude every result that looks wrong,”
the rule invites cherry-picking.
Better:
exclude only trials with a documented method failure such as a spill, incorrect set-up or missed timing event.
Even then, preserve the raw record.
19. A zero can itself be suspicious
A light sensor suddenly reports zero while the room is clearly illuminated.
Zero is recorded—but should it be accepted immediately as the scientific condition?
No.
Check:
- instrument behaviour;
- initialisation;
- obstruction;
- range;
- connection;
- documentation.
Zero is a value, but values still need interpretation.
20. A zero can be meaningful
Elapsed time at the start of a trial = 0 s.
Added water in a control condition = 0 mL.
Derived change = 0 cm.
These zeros have defined meanings.
The key is not “zero is suspicious”.
The key is:
know what the zero represents.
21. Blank cells in worksheets
Sometimes a worksheet leaves a cell blank because the learner is expected to calculate it.
That is a document-design blank, not missing evidence.
Example:
| Start / °C | End / °C | Decrease / °C |
|---|---|---|
| 70 | 58 | ___ |
The raw data exist; the derived column awaits calculation.
Context matters.
22. Dashes, blanks and symbols need a key
A dash “—” can mean:
- not measured;
- not applicable;
- zero;
- missing;
- same as above.
Never assume.
If a table uses symbols, include a key.
In student-created tables, words are often clearer than unexplained punctuation.
23. “Not applicable” is not missing
A table compares several tests.
One test does not require a temperature reading at all.
Use:
N/A — not applicable.
This tells the reader the measurement was not expected for that condition.
That differs from “we forgot to measure it”.
24. “Below detection” is not automatically zero
Some instruments cannot detect very small amounts or changes.
If the instrument reports nothing detectable, the result may mean:
below the instrument’s detection capability
rather than:
exactly zero exists.
At Primary 4, do not overcomplicate this with formal detection-limit mathematics.
Teach the wording difference.
25. Missing data should influence the conclusion
Suppose a five-day plant record is missing Day 3.
The learner can still describe Day 1, 2, 4 and 5.
But they should not claim exactly what happened on Day 3.
The conclusion can say:
“Measured height increased across the recorded days; no Day 3 measurement is available.”
26. Missing data do not always destroy the whole investigation
One missing reading may still leave enough evidence for a bounded conclusion.
Whether the task remains useful depends on:
- which value is missing;
- whether it is essential to the comparison;
- how much other evidence exists;
- whether the missingness creates unfairness between conditions.
Example:
If starting temperature is missing for one cup, the cooling comparison may fail completely.
If one of ten repeated intermediate time points is missing, the overall trend may still be partially describable.
27. Missing data can make a comparison unfair
Condition A has five valid trials.
Condition B has only one because four readings were lost.
Comparing the two as though the evidence strength were equal is risky.
State the imbalance.
Consider repeating missing condition B trials if feasible and scientifically appropriate.
28. Do not backfill from expectation
Missing 10-minute temperature.
Values before and after suggest it was probably around 61°C.
Do not enter 61°C as though it was measured.
If a later advanced lesson teaches interpolation, that estimate must be labelled as estimated.
For this P4 evidence record, keep the gap.
29. Do not backfill from another group
Group A misses one shadow reading.
Group B has a value at the same distance.
Group A should not copy Group B’s value into its raw table.
The groups may compare records later.
Independent evidence must remain independently identified.
30. Do not backfill from memory
“I think it was 14 cm.”
If the value was not recorded and cannot be verified, keep it missing.
Memory can be noted separately if relevant, but it should not silently become raw data.
31. Corrections vs missingness
If a photograph clearly shows the original value and the table transcription is wrong, correct the transcription.
That is not missing data.
If no trustworthy source preserves the value, the gap remains.
32. Data-status column
For important investigations, add a status column:
| Time | Value | Status | Note |
|---|---|---|---|
| 0 min | 70°C | Valid | Start |
| 5 min | 65°C | Valid | |
| 10 min | NM | Timer alarm missed | |
| 15 min | 58°C | Valid |
This is an eduKate teaching structure, not an official lab standard.
33. Original Data-Gap Casebook
Case 1 | Blank becomes zero
Wrong: creates a false measurement.
Case 2 | Unreadable photograph
Correct: UR; preserve image, do not invent exact value.
Case 3 | Missed time point
Correct: NM; graph real observations only.
Case 4 | Spill trial
Correct: retain raw value, mark EX for the equal-volume comparison and state reason.
Case 5 | Calculated zero
15 cm start, 15 cm end.
Correct: derived 0 cm change.
Case 6 | N/A
Measurement not relevant to one condition.
Correct: mark not applicable.
Case 7 | Logger gap
Timestamp absent.
Correct: preserve gap and investigate device/session.
Case 8 | Copied group value
Wrong: destroys provenance.
Case 9 | Memory reconstruction
Wrong: uncertain recollection presented as raw data.
Case 10 | Table calculation blank
Raw start/end values present, derived cell blank.
Meaning: calculation not yet completed, not missing measurement.
Case 11 | Zero sensor output
Action: interpret with instrument context; zero is recorded but may indicate instrument state or true minimum.
Case 12 | Unequal missingness between groups
Action: state evidence imbalance; repeat if needed before strong comparison.
34. The Data-Status Card
| Status | Meaning | Can it be plotted as zero? |
|---|---|---|
| 0 | Measured or derived numerical zero | Yes, if zero is truly the value |
| NM | Not measured | No |
| NR | Not recorded | No |
| UR | Unreadable | No |
| EX | Existing result excluded from a stated analysis | Not in that analysis; keep elsewhere |
| N/A | Measurement not applicable | No |
These codes are examples. The class must define whatever symbols it uses.
35. The evidence rule: absence-looking values need a meaning
Whenever a table cell appears empty or zero-like, ask:
- Was it measured?
- Was it recorded?
- Was it readable?
- Was it applicable?
- Was it excluded?
- Was zero actually observed or calculated?
One question can prevent an entire false graph.
36. What this guide does not teach
Primary 4 pupils do not need:
- statistical imputation;
- multiple imputation;
- censored-data models;
- missing-at-random theory;
- formal detection-limit calculations.
The foundational habit is enough:
never turn “we do not have a value” into “the value is zero”.
37. Original Practice Set
- What does a numerical zero mean?
- Why is a blank not automatically zero?
- What is the difference between not measured and not recorded?
- What should happen when a value is unreadable?
- What is an excluded result?
- Why should excluded data still be preserved?
- How is zero change different from missing final data?
- Why should a missing point remain a gap in a graph?
- Does a line between two points create a measured middle value?
- What does N/A mean?
- Why can “below detection” differ from zero?
- Can one missing value destroy an entire investigation?
- When can missing data make a comparison unfair?
- Why should the learner not fill a missing value from memory?
- Why should Group A not copy Group B’s missing value?
- How is a corrected transcription different from missing data?
- What is the purpose of a data-status column?
- Why should exclusion rules be method-based?
- What should a learner do if a logger skips one timestamp?
- Write one sentence describing a trend with a missing observation.
38. Practice Answers
1. Zero is an actual numerical value produced by a measurement or calculation with a defined meaning.
2. The observation may be missing, unreadable, not applicable or excluded rather than zero.
3. Not measured means no observation was taken; not recorded means an observation may have been made but no trustworthy value was preserved.
4. Mark it unreadable and preserve the original evidence rather than inventing precision.
5. A result that exists but is left out of a particular analysis for a documented scientific reason.
6. So the decision can be reviewed and the evidence trail stays intact.
7. Zero change requires comparable start and end measurements; missing end data cannot support the calculation.
8. Plotting it as zero creates a false observation.
9. No. The joining line is a display, not a new measurement.
10. The measurement does not apply to that condition.
11. The instrument may be unable to detect a small value that is not exactly zero.
12. Not always. It depends on whether the missing value is essential to the scientific comparison.
13. One condition may have much weaker evidence than the other.
14. Memory is uncertain and should not silently become raw evidence.
15. That would destroy the independence and provenance of the two groups’ data.
16. A correction repairs an existing verifiable record; missing data lack a trustworthy original value.
17. It tells the reader whether each cell is valid, missing, unreadable, excluded or not applicable.
18. So inconvenient results are not removed merely because they challenge the prediction.
19. Preserve the gap, check the session/device and do not replace it with zero.
20. Example: “Temperature decreased across the recorded times from 70°C at 0 minutes to 58°C at 15 minutes; no 10-minute measurement was recorded.”
39. The Data-Gap Diagnostic
| If the learner… | Likely weak link | Repair |
|---|---|---|
| fills blanks with zero | data semantics | label missing status |
| deletes failed trials | evidence provenance | retain + mark exclusion reason |
| invents midpoint value | observation vs estimate | keep gap |
| copies another group | independent record | preserve own missingness |
| ignores missingness imbalance | comparison strength | state evidence asymmetry |
40. A 40-Minute Data-Gap Lesson
Minutes 1–5: sort zero, blank, missing and N/A cards.
Minutes 6–10: repair a false zero in a temperature table.
Minutes 11–15: plot a graph with one genuine gap.
Minutes 16–20: distinguish missing from excluded data.
Minutes 21–25: annotate a spill trial without deleting it.
Minutes 26–30: inspect logger/time-lapse gaps.
Minutes 31–35: write a bounded conclusion with missing evidence.
Minutes 36–40: transfer to plant, shadow or cooling data.
41. What Parents and Tutors Can Ask
- “Was zero actually measured?”
- “Why is this cell blank?”
- “Was the value missing, unreadable or not applicable?”
- “Why was this trial excluded?”
- “Where is the original excluded result?”
- “Can you calculate change if one endpoint is missing?”
- “Did the graph invent a point?”
- “What can you still conclude despite the gap?”
42. Complete Batch 22 | Primary 4 Science Learning Guide
- Primary 4 Science Learning Guide | Small Differences, Scale Intervals and Honest Precision
- Primary 4 Science Learning Guide | Raw Data, Derived Values and the Evidence Trail
- Primary 4 Science Learning Guide | No Change, Null Results and What You Can Conclude
- Primary 4 Science Learning Guide | Missing Data, Blanks, Zeros and Exclusions
Return to the Primary 4 Science Learning Hub.
The Quiet Return
The table has one empty cell.
The learner no longer rushes to complete it.
What does the absence mean?
That question protects the entire evidence record. A missing value can remain missing. A zero can remain zero. An excluded trial can remain visible. Science becomes trustworthy because the table refuses to pretend it knows what was never observed.
