How to Write a High-Scoring IB Physics IA

Dr Sri Sharma — Institute of Science & Medicine, Dubai

The Internal Assessment is worth 20% of your final IB Physics grade, and it's the one component almost entirely within your control — no exam pressure, no time limit on thinking, just you, a question, and enough time to get it right. Most students still underperform on it, and it's rarely because the physics is too hard. It's because the investigation isn't structured the way examiners are trained to mark.

The five criteria, and what they actually reward

The IA is marked against five criteria, each worth a slice of the total 24 marks: Personal Engagement, Exploration, Analysis, Evaluation, and Communication. Students tend to pour their effort into Analysis — the calculations, the graphs — and treat Evaluation as an afterthought. That's backwards. Evaluation is where most marks are actually lost, because it demands a level of critical honesty about your own methodology that doesn't come naturally.

Choosing a research question that survives analysis

The single biggest predictor of a strong IA is the research question — chosen before any data is collected. A good question needs three things: a genuinely variable independent variable you can control precisely, a dependent variable you can measure with real precision (not just "measure the time" — how, with what resolution, and how will you reduce random error), and a plausible physical model you can test the data against.

Avoid questions that are really just "verify a known law" with no interesting variable — examiners can tell immediately, and it caps how much genuine exploration is possible. Instead, look for a question where the relationship isn't a simple direct proportionality you already know the answer to — somewhere with real uncertainty in the outcome.

Uncertainty propagation is where marks quietly disappear

This is the single most common technical gap I see across student IAs: raw measurement uncertainties are stated correctly, but they're never properly propagated into transformed or derived variables. If your dependent variable involves a calculation — a gradient, a squared term, a ratio — the uncertainty in that final value needs to be calculated from the original measurement uncertainties, not just estimated from the spread of data points on a graph. Missing this is one of the fastest ways to lose marks in Analysis, and it's almost entirely avoidable with the right method taught early.

Literature comparison in Evaluation

A genuinely strong Evaluation section compares your result against an accepted literature value or theoretical prediction, discusses the size and direction of any discrepancy, and links that discrepancy back to specific, identifiable sources of systematic error in your method — not vague generic ones. If your conclusion doesn't reference what the data was expected to show based on established physics, it's missing a criterion component entirely.

A practical starting checklist

1. Choose a topic that genuinely interests you — Personal Engagement is assessed, and it shows when it's faked.
2. Draft your research question before touching any equipment, and pressure-test it: can you actually vary the independent variable with precision?
3. Plan your uncertainty propagation method before collecting data, not after.
4. Collect more data points than you think you need — a thin dataset limits what Analysis can credibly claim.
5. Find at least one literature or theoretical value to compare your result against before you start writing the Evaluation.
6. Write the Evaluation section like you're trying to find flaws in someone else's work, not defending your own.