IB Physics IA Topics: How to Choose One That Scores

What the investigation actually is

A single individual investigation, marked by your teacher and moderated by the IB, worth 20% at both SL and HL. It is not a lab report of a prescribed practical — you design it. The current syllabus allocates 10 hours to it, which is less than most students assume and a reason to start early.

The mark distribution is the important detail. Conclusion and evaluation carry around half the available marks. Data collection is necessary but it is not where the score is decided, which means the experiment itself can be simple as long as the analysis is serious.

What makes a research question work

Four tests, in order:

• Can you vary the independent variable continuously? You want at least five or six well-spaced values, ideally more. A question with only two or three settings cannot produce a graph worth analysing.
• Can you actually measure the dependent variable precisely enough? If your expected effect is smaller than the resolution of your instrument, the investigation is dead before it starts.
• Is there a relationship you can linearise? A theory that predicts a straight line when you plot the right quantities gives you a gradient to compare against a known value — which is exactly what a strong conclusion is built on.
• Can you control the obvious confounders? If three things change at once, the evaluation writes itself but the conclusion cannot.

A question that passes all four is worth more than a more exciting one that fails the second.

Topic directions by theme

Ideas that tend to work, grouped by where they sit in the syllabus:

• A — Space, time and motion. The period of a pendulum against length or amplitude; terminal velocity of falling objects against a measurable property; projectile range against launch angle; friction on an inclined plane against surface or load.
• B — The particulate nature of matter. Cooling curves against surface area or insulation thickness; how the pressure of a fixed gas volume varies with temperature; specific heat capacity by mixtures.
• C — Wave behaviour. Speed of sound in air against temperature; standing waves on a string against tension or linear density; refraction and critical angle in different liquids; single-slit diffraction against slit width.
• D — Fields. Magnetic field strength against distance from a coil; induced EMF against rate of change of flux; resistance of a wire against length or temperature; the field between charged plates.
• E — Nuclear and quantum physics. Attenuation of radiation with absorber thickness where equipment allows; the photoelectric effect with an LED-based method; inverse-square behaviour of intensity with distance.

Notice how many of these produce a linear graph after a sensible substitution. That is not a coincidence — it is the property you are selecting for.

Topics that quietly cap your score

• Anything with only two or three data points. Comparing three brands of something is a survey, not an investigation.
• Simulations with no real uncertainty. If the data comes from software with no measurement error, the evaluation has nothing honest to say and a large block of marks becomes unreachable.
• Questions where you already know the answer exactly. Verifying g = 9.81 with a stopwatch is fine as a practical and thin as an investigation, because there is nothing to interrogate.
• Equipment you do not have. Ambition is not rewarded if the data never materialises. Check availability before committing.

Where the marks are really won

Three habits separate a good investigation from an average one:

• Take more data than feels necessary. Repeat every measurement, keep the raw values, and record the instrument resolution as you go rather than reconstructing it later.
• Propagate uncertainties properly and plot them. Error bars, a line of best fit, and maximum and minimum gradients give you a numerical uncertainty on your result — the thing a strong conclusion quotes.
• Evaluate specifically. “Human error” scores nothing. Naming which measurement dominated your uncertainty, estimating its size, and proposing a concrete improvement is what the criterion rewards.

The conclusion should answer the research question with a number and an uncertainty, and say honestly whether the theory is supported within that range. If your gradient disagrees with the accepted value, saying so and explaining why scores better than pretending it does not.

If you want a second pair of eyes on a research question before you commit weeks to it, that is exactly what IB Physics tutoring is useful for — and unlike exam preparation, IA work happens over weeks, so there is time to fix a bad design early.

Frequently asked questions

How long should the IB Physics IA be?

Around 3,000 words. The word limit is not the constraint most students hit — the constraint is that roughly half the marks sit in conclusion and evaluation, so space spent on lengthy background theory is space taken from the sections that actually score.

What percentage is the IB Physics IA?

20% of the final grade, at both SL and HL. The remaining 80% is split between Paper 1 at 36% and Paper 2 at 44%.

What makes a good IB Physics IA research question?

One where you can vary the independent variable continuously across at least five or six values, measure the dependent variable precisely enough to see the effect, linearise the expected relationship so a gradient can be compared to theory, and control the obvious confounding variables.

Can I do my Physics IA as a simulation?

It is possible but it makes the evaluation much harder, because simulated data has no genuine measurement uncertainty. Since uncertainty analysis and evaluation carry a large share of the marks, a simple physical experiment usually scores better than an elaborate simulation.

Do I need to get the accepted value in my Physics IA?

No. You are marked on the quality of the investigation, not on agreeing with the textbook. A result that disagrees with the accepted value, with an honest account of why and which measurement dominated the uncertainty, scores better than a result quietly massaged to agree.

How many data points should an IB Physics IA have?

At least five or six distinct values of the independent variable, with repeats at each. Fewer than that and the graph cannot support a meaningful gradient or a serious uncertainty analysis.

When should I start the Physics IA?

Earlier than the timetable suggests. The syllabus allocates about 10 hours, but the parts that carry the marks — analysis, uncertainty propagation and evaluation — are the parts that cannot be rushed, and a flawed experimental design is only cheap to fix before the data is collected.