RENZO DIAZ — Reference library
IB Physics IA Ideas: 15 Research Questions by Theme, With How to Measure Each
14 min readBy Renzo DiazLast reviewed October 8, 2026
A workable IB Physics IA idea has one independent variable you can change over a wide range, a dependent variable you can measure with a known uncertainty, and a model that turns the result into a straight line whose gradient or intercept is a physical quantity. Below are 15 such research questions, three for each theme of the current course (first assessment 2025), each with what to vary, what to measure, the graph that answers it and the uncertainty that will dominate. The IB asks every student to formulate their own research question, and the examiners' instructions rule out a simple repeat of a classic experiment while allowing one to be adapted and extended, so each idea ends with a way to extend it: treat them as starting points, not titles to copy.
The ideas are grouped by the five themes of the course: space, time and motion; the particulate nature of matter; wave behaviour; fields; and nuclear and quantum physics. Each comes with a one-line summary in its theme's table and a short plan below it. How the four criteria are marked, and how much data they need, is in IB Physics IA criteria; this page is about choosing and shaping the question.
I plan an investigation the way an engineer plans a test: decide what the final graph will be and which quantity its gradient gives, then choose the measurement that limits the uncertainty, and only then pick up the equipment. Every idea below is written in that order.
Decide what the final graph will be before you take the first measurement.
Key takeaways
- Your question, your data: each student formulates, investigates and answers a unique research question, and may not present the same raw data as another student.
- Five ways to get data: hands-on lab work, fieldwork, a spreadsheet model, a database or a simulation, alone or combined.
- Plan the graph first: most ideas here linearize, so the answer is a gradient or an intercept with its uncertainty.
- Name the dominant uncertainty: each idea says which measurement limits the result and how to shrink it.
- Same task at SL and HL: one investigation, 20% of the grade, marked out of 24 on four criteria of 6 marks.
- Extend, never simply repeat: the examiners' instructions for 2026 rule out a simple repeat of a classic investigation, but allow adapting and extending one.
What makes a Physics IA idea workable
The IB Physics guide asks for an investigation with enough extent and depth for every criterion to be met, built on quantitative data and supported by qualitative observations where they help. It does not ask for a topic beyond the syllabus: the guide says the research question need not go beyond the understandings in the course. Five checks, done before any data is taken, decide most of the outcome.
Five checks before you start
One variable you control, over a wide range
A range that spans a factor of five or ten shows a trend clearly; a narrow range buries it in the uncertainty. The guide leaves the number of values to you, but asks you to justify the range, the interval and the repetitions.
A dependent variable with a stated uncertainty
Know how you will measure it and how precisely before you start: a video frame, a sensor reading, a balance resolution.
A model that predicts the relationship
The conclusion criterion asks for a comparison with the accepted scientific context. A model gives you something to compare with, and a way to linearize the data.
A dominant uncertainty you can name
One measurement usually limits the result. Find it in a short pilot run and design around it.
It fits the time and the equipment
The guide allocates about 10 hours of teaching time to the whole internal assessment, so the data collection has to fit into a few lab sessions with what your school has.
Allowed by the guide
Lab, field, spreadsheet, database or simulation
The guide lists five approaches to data, usable alone or together: hands-on practical work, fieldwork, a spreadsheet for analysis and modelling, data extracted from a database, and a simulation. In a database investigation, the methodology is how you filter and sample the data. One of the ideas below, in theme E, is a database investigation.
The examiners' rule
Adapt and extend, never simply repeat
The IB's Examiner Instructions for 2026 state that the investigation cannot be a simple repeat of a classic investigation or of one listed among the course's skills, though a practical can be adapted and extended. Several setups below are classics of school physics. What turns each into an investigation is its extension, which every plan names: a new variable, a new system, or an intercept that measures something the textbook version ignores.
Theme A: space, time and motion
Mechanics data is easy to collect with a phone camera, so repeats cost little. Each idea below has a model with a clear straight line.
| Research question | Vary | Measure | Straight-line graph |
|---|---|---|---|
| How does the mass of a falling paper cone affect its terminal speed? | Mass (nested cones) | Terminal speed, by video | v² against m |
| How does the string length affect the period of a whirling bung at constant tension? | String length L | Period of 20 revolutions | T² against L |
| How does the angle of an incline affect the acceleration of a rolling cylinder? (HL) | Angle θ | Acceleration over a fixed distance | a against sin θ |
Terminal speed of a falling paper cone
Nest identical paper cones (coffee filters work) to change the mass without changing the shape, and drop them from at least two metres. Film the last metre at the highest frame rate your phone offers, with a ruler in the frame, and read the speed from the frame positions.
At terminal speed the weight balances the drag, mg = ½ρC_DAv², so v² is proportional to m and a graph of v² against m is a straight line through the origin. Its gradient is 2g/(ρC_DA), which gives the drag coefficient C_D once you measure the cone's area.
The dominant uncertainty is position on the frame and whether the cone really has reached terminal speed: check that the speed is the same over two consecutive intervals. To extend it, keep the mass fixed and change the cone angle instead.
Period of a whirling bung at constant tension
A rubber bung on a string passes through a glass tube, and a hanging mass M below the tube sets the tension Mg. Whirl the bung so the mass stays still, change the length of string L between the tube and the bung, and time 20 revolutions.
The tension is mω²L whatever angle the string makes with the horizontal, so using the string length rather than the circle's radius removes the hardest measurement. With ω = 2π/T, T² = (4π²m/Mg)L: a graph of T² against L is a straight line whose gradient you can predict from the two masses.
The dominant uncertainties are reaction time and keeping the hanging mass steady: time 20 revolutions, or film from above. Friction at the tube shows up as a systematic difference between the measured and predicted gradient. To extend it, measure that friction directly and correct the predicted gradient for it.
Rolling cylinder on an incline
Release a cylinder from rest on a ramp at angles from about 5° to 25°, and measure the time to roll a fixed distance s with light gates or video; a = 2s/t².
For rolling without slipping, a = g sin θ/(1 + I/mr²). A graph of a against sin θ is a straight line with gradient g/(1 + I/mr²), so the gradient measures the shape factor I/mr²: 1/2 for a solid cylinder, 1 for a thin-walled tube. This uses rigid body mechanics, an HL topic.
The dominant uncertainty is the start of timing, so use light gates or start from a released catch. At steep angles the cylinder may slip: stop below the angle where the measured acceleration departs from the line. To extend it, compare cylinders of different wall thickness.
Theme B: the particulate nature of matter
Thermal physics, gases and circuits all have well-defined laws and equipment that logs data automatically, which helps with repetition.
| Research question | Vary | Measure | Straight-line graph |
|---|---|---|---|
| How does the exposed surface area of hot water affect how fast it cools? | Container diameter | Temperature every 10 s | ln(T − T_room) against t, then k against area |
| How does the temperature of a fixed volume of air affect its pressure? | Temperature of a water bath | Pressure | p against θ, extrapolated to p = 0 |
| How does the current drawn from a cell affect its terminal potential difference? | Load resistance | Current and terminal p.d. | V against I |
Cooling rate and surface area
Pour the same volume of hot water into containers of different diameters, or cover one container with lids that leave different areas open, and log the temperature every 10 seconds for about 15 minutes.
If the cooling follows Newton's law, the temperature excess decays exponentially, so a graph of ln(T − T_room) against time is a straight line whose gradient is −k. Each run gives one cooling constant k with its uncertainty from the steepest and shallowest lines; a second graph of k against open area tests whether the losses through the surface dominate.
The dominant uncertainties are a drifting room temperature and uneven water temperature: record the room temperature, stir gently before each reading, and keep the probe at the same depth. Evaporation is part of the effect you are measuring, which is worth saying in the evaluation. To extend it, compare runs with and without a lid, which separates the losses through the surface from those through the walls.
Pressure of air at fixed volume
Seal a flask of air to a pressure sensor and immerse it in a water bath heated from about 0 °C to 80 °C, waiting at each temperature until the pressure reading stops changing.
At fixed volume, pressure is proportional to absolute temperature, so a graph of p against temperature in degrees Celsius is a straight line that reaches p = 0 at absolute zero. The x-intercept and its uncertainty, from the steepest and shallowest lines, is the result to compare with −273.15 °C.
The dominant uncertainty is the air in the tubing outside the bath, which never reaches the bath's temperature: keep the tubing short and narrow, and take readings both heating and cooling to see any lag. To extend it, repeat the run with more tubing outside the bath and use the shift in the intercept to correct for it.
Electromotive force and internal resistance of a cell
Connect a cell to a variable resistor with an ammeter in series and a voltmeter across the cell, and record the terminal potential difference for a range of currents.
From ε = I(R + r), the terminal p.d. is V = ε − Ir: a graph of V against I is a straight line whose intercept is the emf and whose gradient is −r.
The dominant uncertainty is the cell itself changing during the run, as it warms and discharges: take readings quickly, switch off between them and measure the currents in a mixed order. To extend it, change the cell rather than the load: a fruit cell with different electrode separations or areas gives a new research question about what sets the internal resistance.
Theme C: wave behaviour
Oscillations and waves give clean periodic data, and the intercepts of these graphs often carry as much physics as the gradients.
| Research question | Vary | Measure | Straight-line graph |
|---|---|---|---|
| How does the mass on a spring affect its period of oscillation? | Hanging mass | Period of 20 oscillations | T² against m |
| How does the tension in a string affect the frequency of its first harmonic? | Hanging mass on the string | Resonant frequency | f² against tension |
| How does the concentration of a sugar solution affect its refractive index? | Concentration | Angles of incidence and refraction | sin i against sin r, then n against concentration |
Mass on a spring and the spring's own mass
Hang masses on a spring, set each oscillating with a small vertical amplitude, and time 20 oscillations, or record them with a phone's accelerometer.
The period follows T² = (4π²/k)(m + m_eff), where m_eff is the share of the spring's own mass that moves; for a uniform spring it is about a third of it. A graph of T² against m gives the spring constant from the gradient and m_eff from the x-intercept, which you can compare with a third of the spring's measured mass, and k with the value from a static extension test.
The dominant uncertainty is timing: count many oscillations and start counting at the lowest point. Keep the motion vertical, since a sideways swing adds a pendulum mode. To extend it, compare springs in series and in parallel.
Tension and the first harmonic on a string
Run a string over a pulley with masses hanging from its end, drive it with a vibration generator at a fixed length L, and raise the frequency until the first harmonic appears.
The wave speed on a string is √(T/μ), and the first harmonic has a wavelength of 2L, so f² = T/(4L²μ). A graph of f² against tension is a straight line whose gradient gives the mass per unit length μ, which you can check by weighing a measured length of the string.
The dominant uncertainty is judging the resonance: approach it from below and from above and take the mean. Heavy loads stretch the string and change μ, which belongs in the evaluation. To extend it, compare strings of different materials, each with μ from the gradient against μ from weighing.
Refractive index of a sugar solution
Fill a semicircular transparent container with a sugar solution, aim a low-power laser at the centre of its flat face, and measure the angles of incidence and refraction for several angles. Repeat for a series of concentrations.
For each concentration, a graph of sin i against sin r is a straight line whose gradient is the refractive index n. A second graph of n against concentration shows how the index changes, and published refractive-index tables for sucrose solutions give the context for the conclusion.
The dominant uncertainty is reading the angles: use large angles and mark the beam with pins over a long path. Because the beam enters at the centre of the flat face, the refracted ray meets the curved face along a radius and leaves without bending, so only one refraction is measured. The series of concentrations is what extends the textbook Snell's-law experiment. Use only a low-power laser, with your teacher's approval.
Theme D: fields
Magnetic fields are easier to measure than they look: a phone's magnetometer and a top-pan balance can both do it.
| Research question | Vary | Measure | Straight-line graph |
|---|---|---|---|
| How does the distance from a long straight wire affect the magnetic field around it? | Distance r | Field, by phone magnetometer | 1/B against r |
| How does the angle between a wire and a magnetic field affect the force on the wire? | Angle θ | Change in a balance reading | F against sin θ |
| How does the speed of a magnet falling through a coil affect the peak induced emf? (HL) | Drop height h | Peak emf, by data logger | Peak emf against √h |
Magnetic field of a straight wire
Pass a steady current of a few amps through a long straight wire and measure the field at a range of distances with a phone's magnetometer or a Hall probe, subtracting the reading with the current off at every position.
The field of a long straight wire is B = μ0I/(2πr). The magnetometer sits somewhere inside the phone, so the true distance is r + d0 with d0 unknown; a graph of 1/B against r is still a straight line, its gradient gives μ0 and its intercept gives d0. That turns the hidden offset into a result.
The dominant uncertainties are the Earth's field and nearby steel, which the background subtraction removes if nothing moves between readings. The offset d0 from the intercept is what extends the textbook version: it locates the sensor. A wire carrying several amps gets warm: use a low-voltage supply, keep runs short and work with your teacher.
Force on a wire at an angle to the field
Place a pair of magnets on a top-pan balance and hold a stiff wire through the gap between their poles on a stand, so the wire never touches the balance. With a constant current in the wire, the change in the balance reading gives the force, F = (change in reading) × g. Rotate the wire to change the angle θ between it and the field.
From F = BIL sin θ, a graph of F against sin θ is a straight line through the origin whose gradient BIL gives the field strength B, with L the length of wire inside the field.
The dominant uncertainty is the balance resolution against a small force, and the effective length L in a field that fades at the edges of the poles: use the largest safe current and take L as the pole width, then discuss that choice. To extend it, move the wire out from between the poles in steps and map how the force falls off, which measures the spread of the field.
Magnet falling through a coil
Drop a magnet through a vertical tube into a coil from a range of heights h above the coil, and record the induced emf with a data logger at a high sampling rate.
The speed at the coil is about √(2gh), and the emf is the rate of change of flux, which for the same magnet and coil is proportional to that speed. A graph of the peak emf against √h should be a straight line through the origin. The area under each half of the pulse equals the change in flux linkage, which should be the same at every speed: a built-in check on the data.
The dominant uncertainties are the magnet tumbling and a sampling rate too low to catch the peak: guide the magnet with the tube and sample fast. This uses induction, an HL topic. To extend it, compare coils with different numbers of turns, checking the flux-linkage areas for each.
Theme E: nuclear and quantum physics
Theme E has the most restricted equipment, so one idea here uses a database. Radioactive sources are only for schools that hold them, under the teacher's control.
| Research question | Vary | Measure | Straight-line graph |
|---|---|---|---|
| How does the distance from a gamma source affect the count rate? | Distance d | Counts per minute, minus background | 1/√(count rate) against d |
| How does the wavelength of an LED's light relate to the voltage at which it starts to conduct? | LED colour | Threshold voltage | Threshold voltage against 1/λ |
| How does the luminosity of a main-sequence star depend on its mass? | Star, from a database | Mass, radius and temperature | log L against log M |
Count rate and distance from a gamma source
With a school's sealed gamma source and a Geiger–Müller tube, measure the count rate at a range of distances, and the background count with the source away.
For a point source, the corrected count rate falls with the inverse square of the distance from the source to the effective point of detection inside the tube, at d + d0. A graph of 1/√(count rate − background) against d is a straight line whose x-intercept gives −d0.
The dominant uncertainty is the randomness of decay: a count of N has an uncertainty of √N, so count for long enough to reach several hundred counts at the farthest point. To extend it, compare d0 with the tube's dimensions, or repeat with a thin absorber in place. Handling of the source follows your school's rules and your teacher's supervision.
LED threshold voltage and the Planck constant
For LEDs of several colours, measure current against voltage and find each LED's threshold voltage by extending the straight part of its current–voltage curve back to zero current. Take each LED's wavelength from its datasheet, or better, measure it yourself with a diffraction grating.
A photon of wavelength λ carries energy hc/λ, and the threshold voltage is roughly the voltage at which each electron crossing the junction gains that energy, eV ≈ hc/λ. A graph of threshold voltage against 1/λ is a straight line whose gradient, hc/e, gives an estimate of the Planck constant to compare with the accepted value.
The dominant uncertainty is systematic: how the threshold is defined, and the LED's spread of wavelengths. Discuss both in the evaluation. To extend it, measure every wavelength yourself with a diffraction grating instead of taking it from the datasheet. The photon energy belongs to quantum physics, an HL topic.
Mass and luminosity of main-sequence stars
This is a database investigation. A catalogue of well-studied eclipsing binaries, such as DEBCat, gives the masses, radii and surface temperatures of stars measured to a few per cent. Select the main-sequence stars, compute each luminosity yourself from the Stefan–Boltzmann law, L = σ4πR²T⁴, and compare it with the catalogue's value.
If luminosity follows a power of mass, L ∝ Mⁿ, a graph of log L against log M is a straight line whose gradient is n. Whether one straight line fits all masses, or the gradient changes from low-mass to high-mass stars, is itself a result.
In a database investigation the methodology is the selection: state which stars you kept and why, and how you excluded giants. The dominant uncertainty comes from the temperatures, which enter to the fourth power. The relationship goes beyond the guide's content, which the guide allows. To extend it, fit low-mass and high-mass stars separately and compare the two gradients.
Ideas that tend to stall, and why
Most weak investigations fail on the plan, not the write-up. These are the common ways, each tied to the criterion it costs.
| Weakness | Example | Criterion it costs | Fix |
|---|---|---|---|
| A range too narrow to show a trend | A pendulum swung only between 5° and 15° | Data analysis | Widen the range, or choose a variable with a larger effect |
| A measurement dominated by reaction time | Timing one swing by hand | Data analysis and evaluation | Time many cycles, or use video or a sensor |
| No model to compare with | "Does the type of ball affect how high it bounces?" | Conclusion | Ask a question a law or model answers |
| Too many things change at once | Different surfaces, balls and heights together | Research design | One independent variable, the rest controlled |
| A simulation that returns its own input | Plotting the equation the simulation uses | Conclusion and evaluation | Use a simulation to explore a case the simple model does not cover |
Making an idea your own, within the rules
Each student must formulate, investigate and answer a unique research question, and must not present the same raw data as another student, and the examiners' instructions add that a simple repeat of a classic investigation does not qualify. The way to make one of these ideas yours is to extend it: change the system, the independent variable, the range or the measuring method, or make the intercept or the deviation from the simple model the result.
Working in a group is optional. A group has at most three students, and each investigates their own question: a different independent variable, or the same independent variable with a different dependent variable, or a different selection from a shared data set. Each student writes their own report; a group report is not allowed.
Before collecting data, the guide asks you to consider safety, ethical and environmental issues, and your teacher must confirm that the plan is feasible. Every source you use, including a database or a simulation, is referenced in the report.
Terms used in this article
- Independent variable
- The quantity you change on purpose, such as a mass, an angle or a distance.
- Dependent variable
- The quantity you measure to see how it responds, such as a period or a count rate.
- Control variable
- A quantity you keep constant so it cannot explain the result.
- Linearization
- Rewriting a relationship so that a graph of two derived quantities is a straight line, whose gradient and intercept carry the physics.
- Dominant uncertainty
- The measurement whose uncertainty limits the result most, and so the one worth improving first.
- Database investigation
- An investigation whose data comes from an existing data set, where the methodology is how the data is filtered and sampled.
Questions about IB Physics IA ideas
What makes a good IB Physics IA topic?
A question with one variable you can change widely, one you can measure precisely, and a model to compare with. The criteria reward how the question is investigated and answered, so a familiar system with a well-chosen variable and a careful treatment of uncertainty can do well.
Can I use a simulation or a database for the IB Physics IA?
Yes. The guide lists hands-on practical work, fieldwork, spreadsheet modelling, database extraction and simulations as approaches, alone or combined. In a database investigation, the methodology is how you filter and sample the data.
Does my IA topic have to be in the IB Physics syllabus?
It can go beyond it, but it does not have to. The guide says the research question need not encompass concepts beyond the course; several ideas here use relationships outside it, which is allowed.
Can I do the Physics IA with friends?
Yes, in a group of up to three, with your own question and your own report. Each member needs a different independent variable, a different dependent variable, or a different part of a shared data set.
Is a pendulum or spring IA too simple?
Not if you extend it. The examiners' instructions rule out a simple repeat of a classic investigation but allow adapting and extending one: a spring whose intercept measures its own effective mass, or a pendulum at large angles where the simple formula fails, is an extension with something real to compare.
How many data points does a Physics IA need?
The guide sets no number. It asks you to justify the range, interval and repetition of your measurements; the criteria article explains what that means in practice.
Written by
Teaching IB since 2016. Seven years of IB reviews you can check on Google — and named results with the documents behind them.
Last reviewed October 8, 2026
Sources
The requirements (approaches to data, unique research questions, group rules, the four criteria and their marks, the 10 hours and the 20% weighting) come from the IB Physics guide for the current course, first assessment 2025, and the rule on classic investigations from the IB's Examiner Instructions 2026, both paraphrased and checked on 8 October 2026. The 15 research questions, their models and their uncertainty notes are original to this page; the physics in them is standard textbook physics. Accepted values for the constants come from CODATA via NIST, and the stellar data suggested for the database idea from DEBCat.
- IB Organisation — Physics in the DP (course overview and subject brief)
- IB Organisation — Examiner Instructions 2026, Physics: internal assessment, scientific investigation
- NIST — Fundamental Physical Constants (CODATA 2022 values)
- DEBCat — catalogue of well-studied detached eclipsing binaries (J. Southworth)