How to write an IB IA research question (with before-and-after examples)

Most IA problems I fix late in the process started with the research question. A vague question leads to a vague method, data that doesn't answer anything, and an evaluation with nothing to evaluate. A precise question makes almost every later step easier.

Start from what you actually care about

When a student brings me a list of "IA ideas" taken from a website, I ask one question first: what do you actually enjoy about this? With one sports-loving student, we threw away the technique list and started from what made the game fun for her. That led to one specific, measurable relationship she actually wanted to know the answer to.

This matters more than it sounds. You'll spend weeks on this. Another student of mine lost interest in her own topic halfway through planning ("it doesn't really interest me"), and that cost her weeks. Interest also shows in the work. In the current maths exploration, personal engagement is its own criterion; the science IA dropped it as a criterion in 2025, but a question you actually care about still shows up in the quality of the design and the evaluation. If you're still at the idea stage, how to choose an IA topic comes before this guide.

Then narrow it to one relationship

A good science or maths IA question usually has one thing you change or compare (the independent variable) and one thing you measure (the dependent variable), in one specific context. Everything else is held constant or recorded.

A pattern that works in the sciences:

How does [independent variable, with range and units] affect [dependent variable, with how it's measured] in [specific organism, material or system, in a specific place]?

"To what extent…" works well when you're weighing evidence, as in ESS, Psychology or Global Politics, or when the answer is a matter of degree.

Before and after

These are modelled on research questions I've worked through with students. Topics and details are changed.

Ecology (Biology or ESS)

• Before: "Plant growth at different locations around the river."
• After: "How does distance from the river bank (0–50 m, in 10 m intervals) affect soil moisture content (%) and the height of [named plant species, binomial name] at [named nature reserve]?"
• What changed: the organism is named, the independent variable has a range and intervals, both measured quantities are defined, and the site is specified.

Air quality and lichens (Biology or ESS)

• Before: "Does pollution affect lichens?"
• After: "How does distance from a motorway (0–300 m) affect lichen percentage cover on the trunks of [one named tree species] at 1.5 m height?"
• What changed: pollution became something measurable (distance from the source), and the tree species and trunk height are fixed, so they can't distort the result.

Maths (sport and motion)

• Before: "The maths of a tennis serve."
• After: "How does the racket-face angle at contact affect the landing distance of a topspin forehand from a target point, and which angle minimises that distance?"
• What changed: one input, one continuous output, and a clear mathematical goal (finding an optimum).

Comparison question (humanities)

• Before: "How did the First World War influence the Second World War?"
• After: "To what extent were the long-term causes of the First and Second World Wars similar?"
• What changed: the task asked for a comparison, so the question became a comparison with a clear criterion.

The precision checklist

Run your question through this before you show it to your teacher:

• ☐ Is there one independent variable (or one clear comparison)?
• ☐ Is the dependent variable measurable, with units and a method?
• ☐ Is the range of the independent variable stated?
• ☐ Is the organism, material or case named precisely (binomial name in Biology)?
• ☐ Is the site or context specified?
• ☐ Could you actually collect this data yourself, safely, with the equipment you have, in the time you have?
• ☐ Is the dependent variable continuous where possible? (See below.)

Prefer a continuous measurement over yes/no

A student designing a sports IA wanted "in or out" as the outcome. The problem is that binary data throws information away: a ball that lands 1 cm out and one that lands 3 m out both count as "out". Measuring the distance from a target gives you far more to analyse. You can still report in/out as a secondary result.

The same applies in the sciences: "germinated or not" is weaker than germination time or seedling length; "survived or not" is weaker than a growth or activity measure.

Get it approved early

Show your teacher the question and the outline of your method before you collect data. It's a five-minute conversation that can save you a redo. If your plan depends on one specific day (a field trip, a borrowed piece of equipment), have a backup plan.

Two rules for later

• If your results contradict your hypothesis, that's a result. Report it and discuss it in your evaluation. Never change the data, invent extra runs or rewrite the question to fit. That's academic misconduct, and it can cost you the diploma, not just marks.
• Write the method first and the introduction last. You'll only know exactly what the introduction needs to set up once the investigation is done.

Keep going

• Want to see the whole thing? Read what a full science IA looks like and what a full maths exploration looks like, both annotated section by section.
• Doing fieldwork? Read the ecology field IA guide.
• Filming motion for Maths or Physics? Read video analysis for the maths IA.
• Using AI during the process? Read how to use AI in the IB without crossing the line.
• The free IA planner template has this checklist built in.