G2 Science K223, K224 and K225 experimental reasoning should not assume that an intended treatment actually changed the independent variable experienced by the system. A heater setting, lamp setting, labelled concentration, fan speed or treatment dose is a plan; the experiment needs evidence that the manipulation was real.
This one-hundred-and-seventy-first Learner’s Guide develops manipulation checks: verifying that the factor deliberately changed in the method actually differed in the samples, organisms or apparatus before the outcome is interpreted.
Mechanism: causal tests require a real treatment contrast
A causal comparison depends on treatment and control differing in the intended independent variable. If that contrast is weak, absent or misdelivered, the outcome cannot fairly test the proposed causal mechanism. Manipulation checks verify the upstream cause before interpreting the downstream response.
Diagnosis
When treatment groups show little or unexpected difference, separate two questions: did the manipulation occur, and did the system respond? Measure the independent variable itself, or a direct validated indicator of it, before concluding that the mechanism failed.
Smallest repair
Repair the treatment contrast at the point it failed: recalibrate the heater, remake the solution, align exposure time, reposition the sensor, mix the sample or relabel the groups. Do not change the outcome measurement until the manipulation itself is verified.
1. Temperature treatment
Manipulation risk: Do not assume the sample actually reached the intended temperature.
Check: Measure or verify sample temperature rather than only the heater setting.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
2. Light-intensity treatment
Manipulation risk: Lamp setting is not automatically the light intensity at the sample.
Check: Measure intensity or control distance/geometry.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
3. Distance treatment
Manipulation risk: Apparatus marks may not equal effective source-to-sample distance.
Check: Measure the relevant separation consistently.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
4. Concentration treatment
Manipulation risk: Prepared concentration can be wrong if volumes or stock solutions are mismeasured.
Check: Check preparation calculations and labels.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
5. pH treatment
Manipulation risk: Nominal solution label may not match actual pH after mixing.
Check: Measure pH if the conclusion depends on it.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
6. Humidity treatment
Manipulation risk: Room setting does not guarantee local humidity at the organism/sample.
Check: Verify local condition where possible.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
7. Airflow treatment
Manipulation risk: Fan setting is not the same as airflow at every sample position.
Check: Check position or measured flow.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
8. Water availability treatment
Manipulation risk: Added water does not guarantee equal soil moisture or uptake.
Check: Measure the actual state relevant to the mechanism.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
9. Nutrient treatment
Manipulation risk: Applied nutrient amount does not guarantee absorbed nutrient.
Check: Separate treatment delivery from biological uptake.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
10. Surface-area treatment
Manipulation risk: Cutting into pieces should change exposed area, not just sample mass.
Check: Check size/shape and total amount.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
11. Stirring treatment
Manipulation risk: Motor speed or number of stirs should correspond to actual mixing difference.
Check: Ensure the treatment changes mixing rather than another variable.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
12. Catalyst treatment
Manipulation risk: Adding catalyst should change catalyst amount while reactant conditions remain comparable.
Check: Check identity and quantity.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
13. Enzyme treatment
Manipulation risk: Enzyme volume may not equal active enzyme amount if preparations differ.
Check: Use same preparation/batch where relevant.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
14. Substrate treatment
Manipulation risk: Substrate concentration should be verified relative to final mixture volume.
Check: Dilution matters.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
15. Force treatment
Manipulation risk: Applied force label should correspond to actual force on the object.
Check: Account for apparatus geometry/friction if relevant.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
16. Load treatment
Manipulation risk: Nominal mass should translate to expected load consistently.
Check: Check units and suspended mass.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
17. Voltage treatment
Manipulation risk: Power-supply setting should match voltage across the component under measurement.
Check: Measure at the relevant points.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
18. Current treatment
Manipulation risk: Changing circuit configuration may alter more than current.
Check: Check voltage/resistance conditions if interpretation depends on them.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
19. Resistance treatment
Manipulation risk: Component label is not always the same as operating resistance.
Check: Temperature and tolerance can matter.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
20. Magnetic treatment
Manipulation risk: Using a stronger magnet by label does not quantify field at the sample.
Check: Control distance and orientation.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
21. Treatment duration
Manipulation risk: Starting clocks at different moments can change exposure.
Check: Align time from actual treatment onset.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
22. Treatment order
Manipulation risk: Earlier conditions can alter later response.
Check: Counterbalance or reset if carryover is plausible.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
23. Treatment volume
Manipulation risk: Equal concentration with unequal total volume can change amount delivered.
Check: Decide whether concentration or amount is the intended variable.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
24. Treatment mass
Manipulation risk: Equal mass does not guarantee equal number/size of particles or organisms.
Check: Match the variable actually intended.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
25. Treatment area
Manipulation risk: Applying same amount to different areas changes dose per area.
Check: Normalise where relevant.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
26. Treatment per organism
Manipulation risk: Same total treatment across different group sizes changes per-organism exposure.
Check: Use common base if needed.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
27. Control manipulation
Manipulation risk: The control should differ only in the intended treatment absence/reference.
Check: Verify other handling is matched.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
28. Sham treatment
Manipulation risk: A procedural control can test whether handling itself causes the response.
Check: Useful when treatment delivery has extra steps.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
29. Vehicle control
Manipulation risk: If treatment is carried in a solvent/medium, the carrier can need its own control.
Check: Do not attribute carrier effects to active treatment.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
30. Blank condition
Manipulation risk: A blank checks background signal from reagents/apparatus without target/sample.
Check: Use it to separate measurement background.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
31. Positive control
Manipulation risk: A known-responsive condition tests whether the system can show the expected response.
Check: Failure makes null sample results hard to interpret.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
32. Negative control
Manipulation risk: A known-nonresponsive condition tests background/false-positive response.
Check: Unexpected signal challenges specificity.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
33. Treatment label swap
Manipulation risk: Mislabelled groups can invert interpretation.
Check: Use clear identifiers and trace handling.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
34. Treatment contamination
Manipulation risk: Cross-contamination can reduce contrast between groups.
Check: Separate tools/containers where necessary.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
35. Treatment drift
Manipulation risk: Applied condition can change over time.
Check: Recheck temperature, concentration, voltage or other treatment state during long experiments.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
36. Unequal exposure
Manipulation risk: Groups may receive different effective durations despite same planned time.
Check: Record actual start/stop.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
37. Incomplete mixing
Manipulation risk: Treatment concentration can vary within the container.
Check: Mix consistently or sample multiple locations.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
38. Unequal starting state
Manipulation risk: Treatment groups begin with different baseline values.
Check: Measure baseline before claiming treatment effect.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
39. Ceiling effect
Manipulation risk: Treatment may have been changed, but response is already near maximum.
Check: A manipulation check can succeed while outcome still shows little change.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
40. Floor effect
Manipulation risk: Likewise near minimum response.
Check: No outcome change does not mean manipulation failed.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
41. Insufficient treatment contrast
Manipulation risk: Low and high conditions may be too similar to produce measurable separation.
Check: Measure the actual difference between treatment levels.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
42. Treatment outside valid range
Manipulation risk: Very high treatment can trigger new mechanisms or damage.
Check: Manipulation can be real but model may no longer apply.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
43. Treatment threshold not crossed
Manipulation risk: Treatment changes but remains below effective threshold.
Check: Manipulation succeeded technically but not mechanistically.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
44. Treatment saturation
Manipulation risk: Further increases do not alter the relevant internal state.
Check: Outcome plateau can coexist with larger nominal input.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
45. Treatment-response lag
Manipulation risk: Manipulation succeeds now, response appears later.
Check: Check treatment state and sampling time separately.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
46. Path dependence
Manipulation risk: Same manipulation can have different effect after different histories.
Check: Verify reset/starting state.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
47. Interaction
Manipulation risk: The manipulation effect can depend on a second factor.
Check: Check treatment contrast at relevant levels of that factor.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
48. Spatial gradient
Manipulation risk: The intended manipulation may differ across locations.
Check: Measure where the sample actually experiences it.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
49. Compartment delivery
Manipulation risk: Treatment delivered to one compartment may not reach another.
Check: Check transport/transfer.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
50. Manipulation and sampling
Manipulation risk: Sampling itself can disturb the treatment condition.
Check: Design measurements that do not erase the manipulation.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
51. Manipulation and calibration
Manipulation risk: A sensor used to verify treatment must itself be calibrated.
Check: A bad manipulation check can misdiagnose a good treatment.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
52. Manipulation and representativeness
Manipulation risk: One verified sample does not guarantee all treatment units received the same condition.
Check: Check enough units/locations.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
53. Manipulation and null result
Manipulation risk: A null outcome is interpretable only if the independent variable genuinely differed as intended.
Check: Otherwise the experiment tested weak/no contrast.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
54. Manipulation and positive result
Manipulation risk: A treatment difference plus outcome difference still needs controls for alternative causes.
Check: Successful manipulation is necessary, not sufficient, for causality.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
55. Manipulation check final rule
Manipulation risk: Verify the independent variable at the level the system actually experiences, not merely at the setting you intended to change.
Check: Interpret the outcome only after confirming the treatment contrast existed.
For practice, write the intended independent variable, the apparatus setting used to create it, and the actual system state that should be verified. Then predict what a failed manipulation would make the outcome data look like.
For transfer, change the apparatus or biological context while preserving the same causal structure. The learner should propose a manipulation check from the mechanism rather than reuse one instrument automatically.
Competing explanations
A null result can mean the mechanism is wrong, the treatment contrast was too small, the system hit a floor/ceiling, the response was delayed, or the detector was insensitive. A manipulation check removes one major branch of that uncertainty before deeper interpretation.
Observation versus inference
The measured treatment state is an observation. The statement that the intended manipulation succeeded is an inference from that measurement. The statement that treatment caused the outcome is a further inference requiring controls and appropriate design.
Support versus proof
A successful manipulation check supports the claim that groups differed in the intended factor. It does not prove that this factor caused the outcome; confounders, interactions and measurement problems can remain.
Model limits
Some independent variables are difficult to measure directly in school experiments. In those cases, use the closest valid indicator provided by the task and state the limitation. Do not pretend an apparatus setting is identical to the system state if the distinction matters.
Delayed transfer
Return later with unfamiliar experiments in which treatment delivery is imperfect or ambiguous. Ask the learner to identify the intended independent variable, propose a direct/validated manipulation check, diagnose a failed null result and state what causal claim remains unsafe.
Internal learning links
Use the Science Hub, Vol 0099 Control-Condition Logic, Vol 0167 Calibration Curves, Vol 0163 Temporal Sampling Cadence, Examination Craft and the PSLE Learner’s Guide.
MOE and SEAB current framework
MOE states that Full Subject-Based Banding is fully implemented and that, from 2027, the Singapore-Cambridge Secondary Education Certificate (SEC) replaces the former N- and O-Level examinations, with graduating students sitting subjects at their respective G1, G2 or G3 levels. See the official MOE Full SBB / SEC announcement. For the current 2027 G2 Science combinations K223, K224 and K225 and linked syllabuses, use the official SEAB G2 syllabus directory.
Final rule
Before interpreting whether a treatment worked, verify that the independent variable actually changed at the level the system experienced. A causal experiment without a real treatment contrast has not tested the causal question it claims to test.
G2 SEC Learner’s Guide: Open the Vol 0132–0175 hub and subject index.
