Physics

QCAA Physics IA2 Student Experiment Guide: Aim for 20/20

A current QCE Physics IA2 guide to research questions, modifications, uncertainty, linearisation, evidence quality, conclusions and ISMG-aligned evaluation.

By Sylligence · Published 2026-07-17 · Updated 2026-07-17 · 10 min read

A top-band Physics IA2 derives a testable relationship from a physical model, modifies the original experiment for a reason, collects enough relevant data, uses uncertainty-aware processing, and turns the processed result into a justified answer to the research question. Linearisation is useful only when the governing physics supports it.

There is no guaranteed 20/20 structure. The QCAA ISMG judges the quality of the evidence across the response, not whether a report copies a fixed heading list, trial count or graph type.

Physics IA2 at a glance

Under the Physics 2025 v1.3 syllabus, IA2 is a 20-mark Student experiment based on Unit 3: Gravity and electromagnetism.

| Requirement | Current QCAA condition | | --- | --- | | Weighting | 20% | | Conditions | Approximately 10 hours of class time; individual response | | Response length | Written: up to 2000 words; multimodal: up to 11 minutes | | Starting point | A practical or simulation performed in class forms the basis of the methodology and research question | | Forming | 5 marks | | Finding | 5 marks | | Analysing | 5 marks | | Interpreting and Evaluating | 5 marks |

The sample task permits students to refine, extend or redirect a Unit 3 experiment. These words describe possible kinds of modification; they are not four boxes that every response must tick.

What does the Physics IA2 ISMG reward?

The four criteria form one reasoning chain:

1. Forming: develop a considered rationale, justify the modifications, pose a specific and relevant research question, and design a methodology capable of collecting sufficient and relevant data.
2. Finding: manage relevant safety, environmental and ethical issues, collect sufficient raw data, and communicate through accurate scientific language and representations.
3. Analysing: process data correctly, identify meaningful relationships, and identify uncertainty and limitations thoroughly and appropriately.
4. Interpreting and Evaluating: justify the conclusion, evaluate reliability and validity, and derive improvements or extensions from the evidence.

A polished conclusion cannot repair an experiment that never collected data capable of answering its question. Equally, a large raw-data table does not earn top-band analysis unless the processing and interpretation make its meaning clear.

How do you build the rationale into the research question?

Start from the physical model rather than from a variable you happen to be able to change. A strong rationale usually explains:

• the relevant physical quantities and governing relationship
• the assumptions under which that relationship should hold
• what the original practical measured and what it could not establish well
• why the proposed modification should produce more useful evidence
• the expected relationship between the independent and dependent variables

Your final research question should state the relationship that the collected data will test. It normally identifies the independent variable, dependent variable and enough system conditions to make the investigation reproducible and bounded.

Avoid filling the rationale with equations that never affect the method or analysis. Every major theoretical step should help justify a variable, transformation, prediction or comparison later in the report.

How should modifications improve reliability or validity?

Do more than list what changed. Explain the evidence problem that each important change addresses.

| Modification | Useful justification | | --- | --- | | Increase repeated trials | Averaging repeated measurements reduces the influence of random variation or one anomalous result on the mean, improving reliability when trial-to-trial variation is material. | | Increase or redistribute variable values | Improves coverage of the relationship when the original range or spacing could not distinguish the predicted model. | | Use a data logger, light gate or video analysis | Reduces a named timing or position-reading limitation when its resolution is appropriate to the measured effect. | | Control release method, angle, alignment or field geometry | Reduces a specific competing influence so changes in the dependent variable can more validly be attributed to the independent variable. |

More trials are not automatically useful if the dominant problem is systematic. A precise sensor can also produce consistently biased data if it is miscalibrated. Name the actual data effect instead of assuming that every modification improves all aspects of evidence quality.

How much data is sufficient for Physics IA2?

QCAA does not prescribe five variations, five trials or a minimum graph count. A design is sufficient when it can reveal the expected relationship, represent relevant uncertainty and support a defensible conclusion.

Roughly five distinct independent-variable values with repeated trials is often a useful planning baseline, but the right design depends on the experiment. A nonlinear relationship may need more values around a turning point. A simulation may allow dense sampling, while a long-period physical experiment may justify fewer carefully selected values.

Ask:

• Is the range wide enough to reveal the predicted behaviour?
• Is the spacing fine enough to distinguish competing relationships?
• Are repetitions sufficient to judge random variation?
• Are measurements independent and recorded at appropriate precision?
• Does the method control the factors the rationale says must remain constant?

The Finding criterion also includes considered management of safety, environmental and ethical issues. If an environmental or ethical issue is genuinely not applicable, a short justified statement can demonstrate that it was considered. It does not replace specific controls for hazards that are present.

When should you linearise Physics IA2 data?

Linearise only when a theoretical relationship predicts a useful transformation. The transformation should reveal or test physics, not merely make a graph look straight.

For a simple pendulum under the model's assumptions,

$T = 2\pi\sqrt{\frac{L}{g}}$

can be written as

$T^2 = \frac{4\pi^2}{g}L.$

Plotting $T^2$ against $L$ should therefore produce a linear relationship whose gradient can be used to estimate $g$. The analysis should explain the physical meaning of the gradient and any expected intercept, not stop at reporting an $R^2$ value.

The same principle applies to projectile data when theory predicts a square-root relationship: a suitable squared-variable plot may test the model and recover a physical parameter. However, an unrelated investigation does not need linearisation, and a curved relationship may be the correct representation.

When transforming data, also consider how uncertainty changes. Squaring a mean without processing its uncertainty consistently creates a precise-looking graph that does not fully represent the evidence.

How do gradients and accepted values strengthen the analysis?

A gradient is valuable when it answers a physical question. If the model connects the gradient to gravitational acceleration, field strength, resistance or another parameter, calculate that parameter with units and interpret it in context.

Where maximum and minimum plausible gradients are appropriate, they can produce an experimental range for the recovered quantity. Then ask whether the accepted value lies inside or outside that range.

This is more informative than saying the result is “close”. It separates two ideas:

• accuracy or validity: whether the accepted/model value is consistent with the experimental result and uncertainty
• precision: how wide the experimental uncertainty range is

A low percentage error can coexist with a wide uncertainty range. A high $R^2$ can coexist with a systematic offset. Correlation describes how consistently points follow a relationship; it does not prove that the apparatus measured the intended quantity without bias.

Not every Physics IA2 has a meaningful accepted constant. If theory supplies only a predicted pattern, evaluate that pattern instead of forcing an accepted-value comparison.

How should you handle outliers and limitations?

Do not remove an outlier only because it lowers $R^2$. First check the raw record, uncertainty, procedure and physical plausibility. If exclusion is justified, show what was excluded, explain why and state how the decision affects the analysis. Comparing results with and without the point can make the judgment transparent.

A strong limitation follows this chain:

specific source of limitation → direction or scale of its effect on data → consequence for reliability or validity → targeted improvement

For example, inconsistent identification of a pendulum bob's centre of mass changes the effective length. That can bias the recovered gradient, so measuring from the pivot to a clearly defined centre point addresses the mechanism more directly than “use better equipment”.

Improvements repair the current investigation. Extensions ask a related new question. A wider range is an improvement when the current range prevents a sound answer to the same research question; it becomes an extension when it deliberately investigates behaviour beyond that question's scope. See improvements versus extensions in QCE Science for more examples.

How do you write a justified Physics IA2 conclusion?

Begin with the relationship established by the processed data, including its direction, form and important quantitative result. Then connect that relationship to the physical model developed in the rationale and answer the research question directly.

A useful conclusion sequence is:

1. State the analysed trend or recovered parameter.
2. Account for relevant uncertainty.
3. Explain whether the result supports the predicted physical relationship.
4. Answer the research question with an appropriately qualified judgment.
5. Acknowledge the material limitation that controls how confidently the result can be used.

Do not introduce new evidence in the conclusion. Do not claim that “the hypothesis was proven” when the data only support the model within the tested conditions.

What two confirmed full-mark responses teach us

Sylligence reviewed two privately supplied Physics IA2 responses that the student confirmed received full marks. Their wording, raw tables and task-specific results are not reproduced, and no teacher-annotated ISMG was supplied.

One pendulum response derived the square-root relationship, transformed the variables to test a linear form and used the slope to estimate gravitational acceleration. It then considered whether the accepted value lay within an uncertainty-based experimental range. One projectile response displayed both the raw nonlinear relationship and a theoretically justified transformed relationship, disclosed an outlier decision and used the gradient to test the governing model.

The transferable pattern is not “always draw two graphs” or “always recover $g$”. It is to make theory, method, processing, uncertainty, interpretation and evaluation traceable to one another.

Physics IA2 submission checklist

Forming

• Does the rationale derive the expected relationship and assumptions?
• Does each important modification have a physics- and evidence-based reason?
• Does the research question explicitly connect the independent and dependent variables?
• Can the method collect sufficient data for that exact question?

Finding

• Are safety, environmental and ethical considerations addressed where relevant?
• Are raw data, units, precision and qualitative observations recorded clearly?
• Is there enough independent-variable coverage and repetition for this context?

Analysing

• Is every calculation or transformation theoretically relevant and correct?
• Are uncertainty and limitations represented, not merely named?
• Are trends described with quantitative evidence and physical meaning?
• Are outlier decisions visible and justified?

Interpreting and Evaluating

• Does the conclusion answer the research question through the analysed evidence and physics?
• Are reliability, validity, precision and accuracy distinguished appropriately?
• Does each improvement respond to a discussed limitation?
• Are extensions clearly separated from repairs to the existing investigation?

Frequently asked questions

Do I need five values and five trials?

No. That is a useful baseline for many experiments, not a QCAA rule. Justify a dataset that is sufficient for the relationship and uncertainty in your investigation.

Does every Physics IA2 need linearisation?

No. Linearise only when a governing model predicts a meaningful transformation. The correct analysis may be nonlinear.

Do I need maximum and minimum gradients?

Only when they are an appropriate way to represent the uncertainty in a fitted relationship. Do not add them as decoration.

Is a high R-squared value proof that my experiment is valid?

No. It can support the consistency of a fitted relationship, but systematic bias, uncontrolled variables or an unsuitable model can still weaken validity.

Can I state that environmental or ethical risks are not applicable?

Yes, when that judgment is genuine and briefly justified. You must still identify and manage any safety or other issues that actually apply.

Continue your Physics IA2 review

• Read the general QCAA Student experiment guide.
• Review improvements versus extensions in QCE Science.
• Build the underlying content through the QCE Physics study guide.
• Compare source-based work in the Physics IA3 Research Investigation guide.
• Use Sylligence assignment feedback and select Physics IA2 Student Experiment.

Sources and methodology

Official QCAA documents control the assessment requirements. This guide was checked against the current Physics 2025 v1.3 General senior syllabus, the Physics IA2 sample assessment instrument, the Physics 2025 subject report and the QCAA Physics subject page.

Sylligence also reviewed two privately supplied Physics IA2 responses that the student confirmed received full marks. The examples above are paraphrased and anonymised. They illustrate possible evidence patterns, not official QCAA exemplars, compulsory methods or a guaranteed result.