Denise watches a number appear on the screen. Ten seconds later, another number arrives. Then another. The machine seems wonderfully certain: every row has a time, every reading has a place, and nobody has forgotten to write anything down.
Emily is less convinced. “What is it actually measuring?” she asks.
Faith points to the column heading. It says “temperature.” But that is only a label. Before the group can explain the numbers, they need to know where the sensor is, what it responds to and what happened around it while the record was being made.
This fictional classroom scene captures the central lesson of data logging: automation can collect readings; it cannot decide by itself whether those readings answer the scientific question.
This guide belongs to the Primary 4 Science Learning Hub. It is an optional digital-enquiry extension, not a claim that every Primary 4 pupil must own a particular device, learn programming or complete a standard national data-logging test. A prepared table can teach much of the reasoning before any electronics are used.
All classroom conversations and numerical records below are original teaching examples unless explicitly identified otherwise. The example values are not measurements collected by eduKate, by a named school or by the manufacturers linked in this guide.
What a data logger does
A data logger records values over time. Depending on the equipment and programme, readings may be stored when someone presses a button, at regular intervals, or when a specified event occurs. A sensor supplies a reading; the recording system preserves it; a learner still has to interpret it.
For a concrete reference, the Micro:bit Educational Foundation’s data-logging guide describes using a micro:bit V2 to store sensor readings and examine them later. The MakeCode Datalogger documentation specifies that its flash-storage data-logging blocks require V2 hardware. These are device-specific facts, not a requirement to use that product for the lessons here.
| What you need to understand | Go to | Question to keep visible |
|---|---|---|
| What a sensor is reporting | The measurement chain | Is the reading the property we intended? |
| How often to record | Sampling intervals | Could an important change happen between readings? |
| How to interpret a run | The light-record case | What happened at each time? |
| How to handle a suspicious graph | The fault-finding clinic | Is this a scientific change or a recording problem? |
| How to teach the method | A complete guided lesson | What will the child decide, not merely click? |
| How to check understanding | Practice and worked answers | Can the learner explain the evidence independently? |
1. Follow the measurement chain
Begin with the question, then trace the route to the number. Suppose the question concerns how the surroundings become brighter or darker at a particular spot. A light-sensitive component responds to light reaching it. The device converts that response into a reported value. A programme stores the value beside a time. A chart may then display the record.
These steps are connected, but they are not identical. The physical surroundings are not the number. The number is not the graph. The graph is not yet an explanation of why the surroundings changed.
A useful classroom chain is: question → property → sensor → recorded value → time and context → interpretation. At each arrow, ask what could be lost or changed. A sensor facing away from a lamp may answer a different question from one facing towards it. A timestamp in milliseconds may be misread as seconds. A graph may hide a gap by joining points with a line.
The label does not define the instrument
Writing “water temperature” above a column does not make a sensor suitable for immersion. Writing “brightness in lux” does not convert an arbitrary device scale into a physical unit. Column names should describe the measurement honestly, after the instrument is understood.
With a micro:bit’s built-in temperature function, the official MakeCode temperature reference explains that the reading comes from the processor and is usually an approximation to the surrounding temperature. The board can also warm as it operates. It should not be described as a waterproof probe measuring the water in a cup.
The teaching consequence is straightforward: first decide what temperature the question asks about. Then select equipment designed to measure it safely. A dry-board classroom demonstration and a liquid-temperature investigation are different tasks.
A scale may not be a physical unit
The MakeCode light-level reference describes the micro:bit’s reported scale from 0 to 255. This is not a lux reading. The same documentation notes a start-up behaviour: the first call can return zero while light sensing is being initialised.
Therefore a pupil should not write “the room contained 160 lux” merely because this function returned 160. A suitable heading would identify the device’s light-level scale. Nor should an isolated initial zero automatically be interpreted as evidence that the classroom was dark. The instrument’s documented behaviour matters.
What the learner should be able to say
Before starting a run, ask for three sentences: “We are trying to find out…”, “Our sensor reports…”, and “This reading does not directly tell us…”. The third sentence is especially valuable. It makes the measurement boundary visible before the graph becomes persuasive.
For a light record, the child might say that the device reports changes in its light-level reading at its location. It does not directly report shadow width, the size of the blocking object or the precise physical cause of every fluctuation. Those questions require additional observation or measurement.
2. Plan a run before collecting rows
The easiest part of a logging activity can be pressing Start. The harder part is deciding what a useful run should contain. An endless column of numbers is not automatically an investigation.
Write a one-sentence question. Then choose the sensor position, starting condition, recording interval, stopping point and event notes. Decide which changes will be deliberate and which surrounding conditions should remain as comparable as practical.
For example, “What happens to this light reading when a card is held between the lamp and sensor?” is a manageable question. “Record everything in the room” is not. The first gives the learner an interpretable comparison. The second creates data without a clear scientific job.
| Planning field | Example decision | What it prevents |
|---|---|---|
| Question | Compare readings before, during and after shading | Collecting unrelated data |
| Sensor position | Fixed on a marked dry surface | Movement becoming an unrecorded change |
| Light source | Same safe lamp and position | Changing source and obstruction together |
| Interval | A teacher-selected interval suitable for the changes | Assuming one interval fits every investigation |
| Event record | Note when the card is introduced and removed | Guessing why the trace changes |
| Stop and save | End the run and preserve its identity | Mixing several sessions into one record |
Position is part of the method
Two groups can use the same sensor model and obtain different readings because their positions differ. One sensor may face the lamp; another may be turned sideways. One may be shaded by a pupil leaning forward; another may remain unobstructed.
That does not automatically prove one instrument is faulty. It may show that the groups are sampling different conditions. Mark the intended position and orientation, and record any accidental movement. A number without a location can be much less useful than it first appears.
The adult prepares the equipment; the child owns the enquiry
A Primary 4 pupil can make scientific decisions without writing the programme. The adult can provide safe, checked equipment and a working logging setup. The child can choose a question, predict a pattern, identify controls, explain the event record and judge the conclusion.
This keeps a Science lesson from becoming an unexpected coding assessment. Programming can be a separate learning objective, but it should not silently replace the scientific purpose.
3. Sampling: what happens between the readings?
A recording interval is the time between scheduled readings. A record every ten seconds looks more detailed than a record every minute. Whether the extra detail matters depends on the question and the change being observed.
Imagine a sensor taking readings at 0, 30 and 60 seconds. A card blocks the light from 12 to 18 seconds and is removed before the next reading. The recorded values might show no obvious shading event at all. That does not establish that nothing happened. The change occurred between the sampled moments.
Now imagine a week-long observation of a slowly changing classroom condition. Recording extremely frequently may create many nearly redundant rows while adding practical storage and review work. Faster is not automatically more useful. Choose an interval that is capable of answering the question.
A paper activity makes sampling visible
Draw a timeline from 0 to 60 seconds. Place reading cards at 0, 20, 40 and 60. Then place an event strip from 8 to 12 seconds. Ask whether the readings necessarily capture the event. Move the event strip so that it overlaps a reading. The learner can see why timing affects what enters the record.
No advanced sampling theory is needed. The simple lesson is that a record contains observations at particular times, not a guarantee that every event between them was detected.
Count readings carefully
If a teaching run deliberately records at 0, 10, 20, 30, 40, 50 and 60 seconds, there are seven readings and six ten-second gaps. The elapsed span from first to last reading is sixty seconds.
Six readings are not automatically sixty seconds of observation. The timestamp list decides. A programme that first records at ten seconds produces a different set of rows from one that records immediately at zero, even if both are described casually as “logging every ten seconds for a minute.”
Do not teach an isolated formula without the timeline. Let the child count gaps, inspect endpoints and explain what “start” means in the particular programme.
The interval does not prove the duration of an event
Suppose readings are high at 10 seconds and low at 20 seconds. Without an event note or finer record, the fall happened somewhere between those moments; the data do not identify the exact instant. A graph line joining the points can look like a continuous observation, but it is a representation of sampled values.
For a Primary 4 explanation, “The reading had fallen by the 20-second sample” is more defensible than “It fell exactly at 15 seconds” when no such time was measured. Precision should follow the evidence.
4. Three kinds of time can appear in one investigation
Clock time identifies a moment such as 10:05 a.m. Elapsed time identifies a duration such as thirty seconds after the run began. Playback time belongs to a later video or animation of the data. Mixing these can make a perfectly valid record difficult to interpret.
A log beginning at “0” does not necessarily mean midnight. It may mean the beginning of a session or the time since the device reset. The micro:bit data-logging guide explains that its timestamps restart from zero on a reset; changes of headings or timestamp settings can also divide the displayed data into sections.
The practical lesson is to preserve the session identity. If time suddenly returns to zero, inspect the recording process before claiming that the scientific phenomenon returned to its starting state.
Use a run name, not just a file full of numbers
A simple label such as “Light-run-A, same lamp, sensor facing forward” can distinguish one session from another. Add date, relevant time units and a short setup note. The label is not bureaucracy: it prevents one group’s second trial from being mistaken for another group’s first.
For longer activities, the adult can manage file storage while the child keeps a short handwritten run card. Both should refer to the same session. A good record lets another reader match the table to the setup without reconstructing the lesson from memory.
5. Worked case: the graph falls, but what changed?
The following is an invented teaching log using a relative light-level scale. It is not an actual micro:bit experiment. The purpose is to practise joining numbers with event notes.
| Elapsed time / s | Relative light reading | Event note |
|---|---|---|
| 0 | 160 | Stable setup; no card |
| 10 | 162 | No deliberate change |
| 20 | 159 | No deliberate change |
| 30 | 40 | Card had been placed between source and sensor just before this reading |
| 40 | 42 | Card remained in position |
| 50 | 42 | Card remained in position |
| 60 | 158 | Card had been removed just before this reading |
Denise says, “The numbers went down and then up.” That is a valid description, but it is incomplete. Emily links the fall to the card. Faith notices that the setup notes matter: if the lamp had also been dimmed, the group could not attribute the difference to the card alone.
A stronger account is: “The readings were around 159–162 before the card was introduced, around 40–42 while it was in place, and 158 after it was removed. The event record is consistent with the card reducing light reaching the sensor in this setup.”
The answer avoids several overclaims. It does not say the value is lux. It does not say every card blocks the same amount. It does not infer an exact physical percentage of light removed from the ratio of arbitrary reported values. It does not treat seven readings as seven independent experiments.
What is directly in the log?
The table directly contains reported values, times and human-written event notes. The proposed causal account connects those entries with the scientific idea of light being blocked. If the event notes were missing, the trace alone would not tell us whether someone moved the card, turned the sensor or changed the lamp.
This is why an automatic record still needs a human observation record. The sensor reports what reached it; it does not necessarily report the reason.
What should be repeated?
Repeat the complete before–during–after sequence with the same agreed method. Do not simply copy the three low readings and call them three new tests. Repeating the sequence checks whether the pattern reappears when the planned action is performed again.
If the second run behaves differently, compare setup and event notes. Was the card in the same position? Was the sensor facing the same way? Was the light source unchanged? Did another shadow cross it? A difference between runs can guide an improved method rather than justify deleting the inconvenient run.
What is the useful next question?
Several possibilities exist: Does the result repeat? Does a different card material alter the reading? Does the position of the same blocker matter? Each could be investigated, but they should not all change at once.
Choose the question that follows from the present uncertainty. If the setup was unstable, repeating it consistently is more useful than adding several new materials. If the first comparison is stable, a clearly defined next variable becomes reasonable.
6. Automatic does not mean accurate
A machine can repeat the same mistake more neatly than a person. A sensor placed in the wrong location may generate hundreds of consistent values that do not measure the intended condition. A unit conversion error can affect every row. A misleading column heading can make the entire record look more meaningful than it is.
Separate three questions: Are the readings recorded consistently? Are they sufficiently close to the intended quantity? Do they answer this investigation? These questions concern different parts of the measurement chain.
More digits do not prove greater accuracy
A display showing 23.764 does not automatically justify reporting a real-world temperature to three decimal places. The instrument’s capability, calibration and conditions matter. A learner should report what the equipment and task can support rather than reward the longest number.
In a teaching exercise, the adult may deliberately round a log to a suitable level for the pupil. Label that as a processed copy and preserve the original record. Rounding for communication is different from inventing precision or quietly changing observations.
Two sensors should be checked together before being separated
Suppose one group wants to compare two positions using two instruments. First place them side by side in the same suitable stable condition. If they report differently, that difference deserves attention before the sensors are separated.
This preliminary comparison does not calibrate the instruments fully. It reveals whether a device difference might contaminate the planned location comparison. Do not assume a single offset will correct all later readings under every condition. For a simple class investigation, the teacher may choose one instrument used consistently or explain the remaining limitation.
Many readings in one run are not many independent trials
A minute-long run with sixty readings gives detail about that run. It does not automatically provide sixty independent tests of a material or method. All readings may share the same sensor position, same cup, same starting error and same surroundings.
To repeat an experiment, the group needs a clearly defined new run or comparison according to the investigation plan. The existing Repeated Trials, Reliability and Consistency guide develops that distinction more fully.
7. The event notebook: what the machine did not record
Give one pupil the job of noting meaningful events. Not every movement in the room belongs in the log, but events capable of affecting the measurement do.
Useful notes include “sensor accidentally turned,” “card removed,” “door opened,” “power disconnected” or “recording restarted.” The note should identify approximately when the event occurred and whether the time came from the logger, a separate clock or an estimate.
An estimate should remain an estimate. “Around 40 seconds” is more honest than assigning a precise timestamp after the lesson simply because the graph changes near that point.
A surprising result can become understandable
Imagine an unexpected fall during an otherwise stable light record. The event notebook says a pupil leaned over the device at that moment. That provides a possible explanation worth checking. It is stronger than deciding the sensor “must be wrong” because the reading interrupted a neat line.
Do not treat the notebook as infallible either. The note may be incomplete or the observer may have missed another change. It is additional evidence about the session, not a licence to explain every anomaly with certainty.
Keep the original and an explained copy
The raw record should remain available. Make a copy for annotation, calculation or a simplified display. If a reading is excluded from a classroom comparison because the sensor was moved, explain the reason and retain the original value.
This gives the learner a small but important experience of scientific integrity: improving an analysis does not require erasing what happened.
8. Fault-finding clinic: twelve records that need a second look
The following situations are original diagnostic cases. The task is to choose the next check, not to invent a single guaranteed explanation.
Case 1: the flat line
A graph remains at exactly one value while the group expects change.
Check: did the surroundings actually change? Was the sensor exposed to the relevant condition? Is the programme logging fresh readings or repeating a stored value? Is the reported scale too coarse to reveal a small change?
A flat line can reflect a stable condition or a recording problem. The picture alone does not choose between them.
Case 2: the first zero
The first reading is zero and later values are nonzero under apparently unchanged light.
Check: consult the instrument’s documented start-up behaviour. Some systems require initialisation or settling. Preserve the reading and note the behaviour rather than inventing a period of darkness. This case is particularly relevant to the documented initial call of the micro:bit light function linked earlier.
Case 3: time returns to zero
The record reads 0, 10, 20, 30, then 0, 10, 20 again.
Check: whether the device or session restarted. Do not join the rows as though the physical process ran backwards. Identify separate runs or a documented interruption before interpreting the sequence.
Case 4: a missing minute
Readings expected each minute appear at 0, 1, 2 and 4 minutes.
Check: what happened to the three-minute record. It may not have been saved, or the logging process may have paused. A blank is not automatically zero. Plot the known timestamps accurately and mark the gap.
Case 5: the sensor was moved
The light reading changes after the device is turned sideways.
Check: whether the intended investigation concerned material, location or orientation. If orientation was supposed to remain fixed, the movement is a method change. Reset the agreed position for a new run and retain the note about the interrupted one.
Case 6: two groups use different units
One time column is in seconds and another in milliseconds, but both are labelled “time.”
Check: the programme settings and original headings. Convert only when the units are known. Do not compare the visible numbers directly. A larger timestamp need not represent a longer elapsed duration when the units differ.
Case 7: the board is warm
A built-in temperature value differs from a separate room thermometer.
Check: what each instrument measures and where it is placed. A processor-based approximation is not the same as a purpose-designed ambient-temperature probe. The difference is not automatically evidence that the room thermometer is wrong.
Case 8: the graph looks dramatic
A small change fills most of the height of an automatically scaled graph.
Check: the vertical-axis values. Visual steepness depends on display scales as well as data. Describe the numerical change before calling it large. Retain the real units or identify a relative scale.
Case 9: the neat average
A pupil averages together readings before shading, during shading and after shading, then uses the single result to explain the effect of the card.
Check: whether the summary has erased the comparison. Different phases answer different parts of the question. Keep the before–during–after structure visible instead of compressing everything into one number.
Case 10: a disconnected file
A saved table has no run name, date, sensor identity or setup note.
Check: whether the session can be identified from the original record. Do not fill gaps with guesses. The useful repair is to add a run card to the next investigation, not reconstruct certainty after the event.
Case 11: simulated values are mixed with measured values
A programme was tested in a simulator, then run on a physical device. The exported teaching sheet does not say which values came from which run.
Check: the origin of each record. Software testing can verify parts of the programme, but simulated inputs are not observations of the classroom. Separate the records and label them clearly.
Case 12: the log stops unexpectedly
The group planned a long observation, but the file ends early.
Check: power, storage limits, programme behaviour and the stopping condition. The micro:bit data-logging FAQ notes that recording duration depends on factors including interval, number of recorded values, storage and power. The general lesson is to pilot the intended session before depending on it.
9. Read a log before admiring its graph
A graph is useful when it makes the relationship easier to inspect. It is misleading when it substitutes visual confidence for the underlying record.
Read the column headings first. Identify the time units, the measured quantity and whether the values are raw readings or calculated differences. Then inspect the start, finish, gaps and event notes. Only afterward decide what display would help.
For a time series, the horizontal positions should reflect the actual times. Equal spacing between rows on a table does not prove equal elapsed time. If a timestamp jumps from 20 to 80 seconds, a graph that places that gap like an ordinary ten-second gap distorts the session.
Lines between readings are not additional measurements
A joining line helps the reader follow the record. It does not mean the sensor measured every value along the drawn segment. A smooth curve can make the gap look more informative than it is.
Ask the pupil to point to a recorded observation and then to a connecting line. “Which one is in the file?” is a simple question that prevents a common confusion between evidence and display.
Descriptions should carry numbers and context
“The line goes down” is a starting description. “The reported light value fell after the card was introduced, while the source and sensor positions were kept fixed” is more useful. Add relevant values when they help the comparison, but do not copy the entire table into the explanation.
For graph-construction mechanics, continue with Constructing Tables, Graphs and Data Displays. Here the central question is whether the graph remains faithful to the automatic record.
10. A complete guided data-logging lesson
The lesson can be taught in two forms: with an adult-prepared dry sensor setup, or entirely from printed teaching logs. Choose the form that fits the available equipment, school rules and learning goal. No purchase is needed to practise the reasoning.
Phase A: predict before the display
Present a safe lamp, sensor position and card arrangement, or a written description of them. Ask each pupil to predict how the reading might change when the card is introduced and removed. The pupil should identify the changed condition and say what must remain fixed.
Do not reveal the example log yet. The prediction is a record of the learner’s model, not a score to be protected from later evidence.
Phase B: design the record
The group chooses headings: elapsed time, reported sensor value and event note. The teacher supplies the correct unit or identifies the relative scale. Pupils choose a sensible short interval from two prepared options and explain why it suits the planned event duration.
This is a genuine scientific decision even when the teacher has already written the programme. The child is deciding what information needs to exist for the result to be interpretable.
Phase C: run a short pilot
Check that a few readings are recorded and can be matched to the setup. Confirm that the start and stop procedures work. If the equipment behaves differently from the plan, repair the plan before collecting the main record.
A pilot should be small enough that failure is informative rather than disruptive. It might reveal an unreadable heading, an unsuitable interval or a sensor orientation problem.
Phase D: collect and annotate
One pupil watches the agreed setup, another notes events, and another checks the record. Rotate roles between complete runs. Avoid crowding the sensor in ways that change the light conditions.
With a printed-log version, distribute an event card separately from the values. Ask the group to combine them. Then remove the event card and ask which conclusions become weaker. This makes the contribution of context visible.
Phase E: explain individually before discussion
Each pupil writes a short account: what was measured, what pattern occurred and what conclusion is justified. Only then compare answers. This prevents the most confident reader from becoming the group’s entire interpretation system.
Use one challenge question: “What else could have changed?” or “Which part of your answer came from the event note?” The purpose is to improve the explanation, not to overwhelm the child with every possible measurement limitation.
Phase F: transfer the method
Give a new log with a missing timestamp or a restart. Ask the pupil to identify the record problem before explaining the Science. The scientific topic may remain familiar while the evidence format changes.
Success means the learner inspects the measurement chain independently. It does not mean the learner can operate every data logger or memorise a particular software interface.
11. Practice: make the recording decision
These questions assess interpretation and method. Several accept more than one justified response. Answer before reading the worked notes.
- A device logs every thirty seconds. A five-second shading event occurs between two readings. Does an unchanged log prove that shading did not happen?
- A teacher’s programme records at 0, 10, 20, 30, 40, 50 and 60 seconds. How many readings are there, and what is the elapsed span from the first to the last?
- What is the difference between a sensor, a stored reading and an explanation?
- Why should the position and orientation of a light sensor be recorded?
- A device reports a relative light value of 180. May the learner label it 180 lux without evidence that the instrument measures lux?
- A processor-based temperature reading is collected beside a cup. Does changing the heading to “water temperature” make it a measurement of the water?
- Write one useful event note for a run whose sensor was accidentally moved.
- A log contains 0, 10, 20, 0, 10 seconds. What should be checked before plotting one uninterrupted line?
- Why is a missing value not automatically zero?
- What can a short pilot reveal before a longer recording session?
- A group has one hundred readings from one unchanged setup. Why are these not automatically one hundred independent experiments?
- Two sensors disagree when placed side by side. What should the group do before using them to compare two locations?
- A graph line between two readings crosses a particular value. Is that crossing necessarily an additional recorded measurement?
- Why can an average of before, during and after readings hide an investigation’s main comparison?
- What should be preserved before annotating or simplifying a data file?
- Which is the stronger description: “the line fell” or “the reading fell after the card was introduced while source and sensor positions remained fixed”? Explain.
- Why should simulated programme-test data be labelled separately from physical-sensor data?
- A log ends early. Name two recording-system factors to check before concluding that the physical process stopped.
- What additional record helps explain a sudden sensor change?
- What should a learner be able to explain without the group or software speaking for them?
Worked answers
1. No. The event may have occurred between sampled moments. The interval must suit the event the group wants to detect.
2. Seven readings span sixty seconds. There are six ten-second gaps between the first and last values. The explicit timestamp list establishes this.
3. A sensor responds to a condition; a stored reading preserves a reported value; an explanation connects relevant evidence with a scientific relationship. These stages should not be treated as interchangeable.
4. The light reaching the sensor depends on where and how it is placed. Unrecorded movement can become another cause of a change.
5. No. Use the instrument’s documented scale. Adding a physical unit that the instrument does not report misrepresents the measurement.
6. No. A heading cannot change what the sensor actually measures. Use equipment suitable for the intended property and environment.
7. “Around 40 seconds, the sensor was accidentally turned sideways; the exact moment was not recorded” is useful if that is what the observer knows. Do not manufacture a precise time.
8. Check whether the device or session restarted and whether the rows belong to separate runs. A return to zero may concern the recording clock, not the phenomenon.
9. A missing value means the record lacks a measurement there. Zero is a particular recorded value with its own meaning.
10. A pilot can reveal incorrect headings, unsuitable intervals, storage or power issues, inconsistent positioning, or a start/stop procedure that does not work as intended.
11. The readings share one setup and its conditions or errors. They provide detail within that run, not necessarily independent repetitions of the investigation.
12. Inspect instrument differences, settings and placement. The teacher should decide whether the comparison is adequate or whether the method needs changing. Do not silently treat unequal devices as identical.
13. No. The line is a display connecting sampled points unless additional measurements are present. It should not be read as a continuous observed record.
14. It combines distinct phases into one number. The question may require comparing those phases, so the summary can erase the useful relationship.
15. Keep the original file, headings, timestamps and run identity. Work on a labelled copy and explain exclusions or transformations.
16. The second is stronger because it names the reported property, links the change to an event and states relevant controls. It still should not claim more than the setup supports.
17. Simulated inputs test the programme or model; physical readings record the sensor’s response in an actual environment. Mixing them misstates where the evidence came from.
18. Power, storage capacity, programme stopping conditions or an interruption are possible checks. The file ending does not by itself show that the physical change ended.
19. A time-linked event notebook describing relevant changes in the setup or surroundings. It supplies context the sensor may not record.
20. The learner should be able to state the question, what the sensor reported, the relevant pattern, one limitation and a sensible next check.
12. Safety, access and responsible data handling
Use school-approved equipment with an adult’s setup and supervision. A bare electronic board is not a general-purpose liquid probe. Keep it away from water, condensation, hot surfaces and improvised mains connections. Use a sensor designed and approved for the environment rather than making an unsuitable device “fit” by relabelling it.
The Micro:bit Educational Foundation’s safety guidance identifies the micro:bit as an exposed circuit board and gives handling and power precautions. Consult the guidance for the actual device in use; the learning activities here do not override manufacturer instructions.
Data logging also does not automatically mean internet publishing. The official micro:bit FAQ explains that its logged data are stored locally unless the user chooses to copy or share them. Other systems may behave differently. The adult should understand where the chosen tool stores data before a class uses it.
For a P4 lesson, collect environmental readings rather than identifiable information about pupils. A light investigation does not need names, voices, precise home locations or a public upload. Keep the data no more personal than the scientific question requires.
A low-equipment route
Use the invented tables in this guide, with the provenance label intact. Pupils can plan intervals, interpret events, find gaps and critique conclusions on paper. They should not describe this as an experiment they personally performed. It is a data-interpretation lesson using prepared examples.
Later, one teacher-operated device can supply a shared live record. Individual prediction and explanation still matter. Access to one instrument should not mean access to only one pupil’s thinking.
13. What to assess
A learner who can press Start has demonstrated a useful operational step. A learner who can explain why a timestamp gap matters has demonstrated scientific judgement. Assess both where they are part of the lesson, but do not confuse them.
| Evidence of learning | What the learner does | Useful next prompt |
|---|---|---|
| Measurement meaning | Names what the sensor actually reports | What does this reading not directly tell us? |
| Sampling judgement | Chooses an interval for a stated event | Could the event fit between two readings? |
| Record integrity | Preserves units, timestamps and run identity | Can another group reconstruct this session? |
| Evidence interpretation | Uses data and event notes together | Which part of the conclusion depends on the notes? |
| Boundary control | Distinguishes a sampled record from continuous observation | What do we not know between these points? |
| Transfer | Recognises a restart or gap in a new log | What should be checked before explaining the Science? |
Use this as a teaching checklist, not an official marking scheme. One well-chosen new log can reveal more than a long sequence of identical screen operations.
Sources and connected learning
Device-specific references used here are the micro:bit data-logging user guide, data-logging FAQ, MakeCode Datalogger reference, temperature reference, light-level reference and safety guidance. Follow the current instructions for the actual device rather than assuming every logger shares these characteristics.
For the wider scientific method, use Apparatus Choice and Measurement Strategy, Repeated Trials, Reliability and Consistency and Science Journals and Observation Logs.
The companion guides explain other evidence routes: Researching with Books, Websites and Secondary Sources, Simulations and Virtual Investigations, and Photographs, Video and Time-Lapse Observation. Return to the Primary 4 Science Learning Hub for the topic-specific model.
At the end of the fictional session, Denise still appreciates the neatly filled table. Emily can now explain the sensor’s boundary. Faith has kept the event notes that make the graph interpretable.
The machine has done the repetitive recording. The learners have done the scientific work: choosing what to measure, preserving how it was measured and deciding what the record can honestly support.
