Skip to main content
Student PlusFree month--D --H --M --S

IA foundations

How to write an IB IA research question

Written and cross-checked by IB Diploma graduates: Alexander (41), Alejandro (40), Martina (40), Victoria (38), Martin (35), Maria (34)

· Checked against official IB guidance · Editorial policy

A research question sets the target for the method, evidence and evaluation that follow. A vague question can produce a vague method, thin evidence and an evaluation with nothing concrete to test. A focused question makes it easier to design an investigation that can actually answer what it asks.

The four-part test

A usable IA question passes all four:

  1. 1. Focused. One sentence, one relationship. You can name the exact thing you will change and the exact thing you will measure.
  2. 2. Investigable with what you have. A school lab, a survey you can actually run, sources you can actually read. If it needs equipment or access you do not have, it is not your question yet.
  3. 3. Right scope. Small enough to go deep in the word limit, big enough to produce a real trend or a real argument. Five levels of the independent variable with repeats is a good target for the sciences.
  4. 4. Answerable. When the data is in, the question has a clear answer — even if that answer is “no measurable effect”. A question you cannot conclude on is the wrong question.

Weak vs strong — 21 examples by subject group

Group 1 — Language & literature

English Literature

Too broad · Themes in Hamlet.

Works · How does Shakespeare use soliloquy in Act III to position the audience's sympathy for Hamlet?

Close reading of one technique in one scene.

IV:
Soliloquy passages selected
DV:
Audience sympathy / positioning (argued)
Evidence:
Annotated extracts + secondary criticism
Trap:
Whole-play summary

Language & Literature

Too broad · How is advertising persuasive?

Works · How does multimodal layout in one charity campaign construct its target reader?

One campaign, one reader construction — fits Lang & Lit analysis.

IV:
Layout / mode choices
DV:
Reader positioning (argued)
Evidence:
Annotated ads + course concepts
Trap:
Many campaigns with no depth

Group 2 — Language acquisition

Spanish B

Too broad · Is Spanish useful?

Works · How does the candidate use past-tense narration vs present commentary to structure the individual oral photo description?

Oral IA — question targets performance structure, not a written RQ.

IV:
Tense / structure choices in oral
DV:
Coherence of description (rubric-aligned)
Evidence:
Recorded oral against rubric criteria
Trap:
Treating the oral like a written IA

Group 3 — Individuals & societies

Economics

Too broad · Are taxes good or bad?

Works · Commentary on a recent article about a national sugar tax, analysed with an indirect-tax diagram, incidence, and welfare loss.

One article, one diagram — the actual Economics IA format.

IV:
Policy change in the article
DV:
Incidence / DWL in your diagram
Evidence:
Named article + labelled diagram
Trap:
General debate with no article

Business Management

Too broad · How can a café improve sales?

Works · Should [local café] introduce a loyalty app to improve customer retention, given current repeat-purchase data?

Real business, one decision, tools you can apply.

IV:
Loyalty intervention (proposed)
DV:
Retention / repeat purchase rate
Evidence:
Sales logs, customer survey, financial estimates
Trap:
No access to real business data

History

Too broad · Was World War One inevitable?

Works · To what extent was the Schlieffen Plan responsible for Germany's failure to win a quick victory in 1914?

Bounded in time with evaluable sources.

IV:
Schlieffen Plan decisions (as factor)
DV:
Speed of victory / stalemate outcome
Evidence:
Two evaluated primary/secondary sources
Trap:
Whole-war question in 2,200 words

Geography

Too broad · Is urbanisation bad?

Works · How does distance from the city centre affect building height along a transect in [your city]?

Field data collection with one clear gradient.

IV:
Distance from CBD
DV:
Building height (m)
Evidence:
Transect survey points
Trap:
Mixed land uses on one transect line

Global Politics

Too broad · Why is politics important?

Works · How effectively does a local refugee-support NGO influence municipal housing policy?

Engagement project with a political issue and stakeholders.

IV:
NGO engagement activities
DV:
Policy influence indicators
Evidence:
Interviews, documents, meeting records
Trap:
Issue too broad for 2,000-word report

Psychology (2027 proposal)

Too broad · Does memory get worse with age?

Works · How could a survey study test whether social-media multitasking predicts self-reported attention difficulties in students aged 16–18?

Proposal format — aim, population, method — not a finished experiment write-up.

IV:
Multitasking score (proposed measure)
DV:
Attention difficulty score
Evidence:
Proposed questionnaire + ethics plan
Trap:
Writing up results you never collected

Group 4 — Sciences

Biology

Too broad · How does temperature affect enzymes?

Works · How does temperature (10–50 °C) affect the rate of catalase activity in potato extract, measured as volume of O₂ produced in 60 s?

Names enzyme, range, and measure — one experiment you could start tomorrow.

IV:
Water-bath temperature (10–50 °C)
DV:
Volume of O₂ in 60 s
Evidence:
Gas syringe readings, 5 temps × 5 trials
Trap:
No enzyme named; effect confounded by H₂O₂ age

Chemistry

Too broad · What affects reaction rate?

Works · How does the concentration of sodium thiosulfate (0.05–0.25 mol dm⁻³) affect the rate of its reaction with dilute HCl, measured by the disappearing-cross method?

One variable, one method, a range for order-of-reaction analysis.

IV:
Na₂S₂O₃ concentration
DV:
Time for cross to disappear
Evidence:
Timed runs at 5 concentrations
Trap:
Multiple variables changed at once

Physics

Too broad · How do pendulums work?

Works · How does the length of a simple pendulum (0.10–1.00 m) affect its period, and what value of g does the gradient of T² against L give?

States the graph and the constant you will extract.

IV:
Pendulum length
DV:
Period T
Evidence:
T² vs L plot, uncertainty on gradient
Trap:
Angle too large — small-angle approximation breaks

ESS

Too broad · How does pollution affect ecosystems?

Works · How does distance from a road (10–100 m) affect lichen coverage on tree trunks as a bioindicator of air quality?

Field transect with one gradient and one measurable indicator.

IV:
Distance from road
DV:
Lichen cover % on trunks
Evidence:
Quadrat/transect data along one site
Trap:
Multiple pollution sources along the transect

SEHS

Too broad · Does exercise help performance?

Works · How does warm-up duration (0, 5, 10, 15 min) affect vertical jump height in school athletes aged 16–18?

Controlled manipulation with a sport-relevant DV.

IV:
Warm-up duration
DV:
Vertical jump height (cm)
Evidence:
Repeated jumps per condition, paired design
Trap:
Fatigue order effects without counterbalancing

Computer Science

Too broad · Can AI help students?

Works · How effectively does a revision-quiz web app reduce average time-to-complete for a fixed question set compared with paper revision?

Computational solution with measurable success criteria.

IV:
Revision mode (app vs paper)
DV:
Time to complete quiz
Evidence:
Logged timings from prototype testing
Trap:
Product built but success criteria never tested

Group 5 — Mathematics

Maths AA

Too broad · How is calculus used in real life?

Works · What release angle maximises the range of a basketball free throw, modelling the ball as a projectile?

Specific optimisation with a quantity to compare against video/data.

IV:
Release angle
DV:
Horizontal range
Evidence:
Model output vs measured throws
Trap:
No comparison to real data

Maths AI

Too broad · Is there a link between screen time and sleep?

Works · Is the relationship between daily screen time and sleep duration in a sampled cohort better modelled as linear or logarithmic?

Compares models on the same dataset — clear analysis path.

IV:
Screen time (hours)
DV:
Sleep duration (hours)
Evidence:
Survey data + regression comparison
Trap:
Correlation without controlling confounds

Group 6 — The arts

Visual Arts

Too broad · How do artists use colour?

Works · How does your use of restricted palette in three resolved artworks communicate a consistent conceptual focus?

Tied to your own body of work, not art history in general.

IV:
Palette restriction (your choice)
DV:
Conceptual focus communicated
Evidence:
Photos of your works + process notes
Trap:
Generic art history with no own work

Film

Too broad · What makes a good film?

Works · How does sound design in one selected sequence establish tension before the narrative reveal?

One film, one sequence, one craft element.

IV:
Sound design choices in sequence
DV:
Tension (audience response argued)
Evidence:
Shot-by-shot analysis + audio cues
Trap:
Plot summary of entire film

Music

Too broad · How does music affect mood?

Works · How do changes in tempo and dynamics in your original composition reflect the stated programmatic narrative?

Your composition, linked technique to intent.

IV:
Tempo / dynamics sections
DV:
Programmatic narrative alignment
Evidence:
Score excerpts + recording
Trap:
General psychology of music with no composition

Dance

Too broad · Why is dance emotional?

Works · How does your use of stillness versus repetition in one choreographed phrase communicate the chosen stimulus?

Choreographic choice tied to one phrase.

IV:
Stillness vs repetition
DV:
Stimulus communication (argued)
Evidence:
Video of phrase + choreographic notes
Trap:
General essay on dance history

Subject-specific lists of ready-to-use questions live on each IA guide, e.g. Biology IA ideas, Chemistry IA ideas, History IA ideas.

Five mistakes that sink a question

  • Two questions in one. “How does X affect Y and Z?” — split it and drop one.
  • No measurable dependent variable. “How does music affect studying?” — studying how? Recall score? Time to complete? Name it.
  • A yes/no question with an obvious answer. “Does exercise raise heart rate?” — you already know. Ask by how much, or which type.
  • A confound you cannot control. If the thing you are changing also changes three other things, the data will not mean anything.
  • A topic, not a question. “Ocean acidification.” is a heading. Add a verb and two variables.

Have a draft with a research question in it?

iBacalao reviews your IA criterion by criterion using the configured criteria for your subject and assessment session — the first 3 lifetime checks are free.

Choose your subject →

Common questions

How long should an IB IA research question be?

One sentence. If it needs two, it is probably two questions — pick one. Aim for a form like "How does [independent variable] affect [dependent variable] in [context]?" for the sciences, or "To what extent..." / "How significant..." for the humanities.

Can I change my research question after I start?

Yes, and most students do. It is normal to run a pilot, discover the effect is too small or the method does not work, and narrow or swap the question. Change it before you write up, not after.

Does the research question need to be original?

It does not need to be new to science. It needs to be your own investigation — your variables, your data, your context. Replicating a known effect in a specific new setting is completely acceptable and often scores better because the method is proven.

What makes a research question too broad?

If you cannot name the exact thing you will measure and the exact thing you will change, it is too broad. "How does diet affect health?" is an essay topic. "How does daily caffeine intake affect resting heart rate in adults aged 16–18?" is an IA question.

Sources & verification

Primary-source references used to verify key assessment facts on this page. Last source check: September 22, 2026.

There is no single universal IA research-question formula across all Diploma Programme subjects. The examples on this page are practical patterns; the current guide for your subject and assessment session controls the actual task and criteria.

Stop reading about it

Upload your draft and get criterion-by-criterion feedback in minutes.

Check my draft free