Overview
The best IB Psychology IA idea is almost always the boring-sounding one: a simple published experiment with two conditions, a clear thing you change, and a clear number you measure. Your IA is not a chance to invent original research — it is a controlled replication of an existing study, written up as a report. A "good idea" means picking an experiment that is (1) genuinely experimental, with one independent variable manipulated across two conditions and one measurable dependent variable, (2) ethical enough to run on classmates with informed consent and no risk of harm, and (3) based on a real, citable study you can read. Get those three right and the exploration, analysis and evaluation marks become far easier to reach. This post is a shortlist of experiments that reliably work, grouped by area, each with a sample aim, hypothesis and what to measure.
How the IB Psychology IA works
For the IB Diploma Programme, the IA is a single experimental study. You take a well-known effect, simplify it into something you can run in a classroom, manipulate one variable, and measure the outcome. You then write it up in the standard sections: introduction (the theory and study you are replicating, your aim and hypotheses), exploration (design, variables, controls, sampling, ethics), analysis (descriptive statistics, a graph, and an inferential test), and evaluation (what your results mean, limitations, and improvements).
This article is only about choosing the study. For how each section is marked, the statistics, and the report structure, read the IB Psychology IA guide. Come back here once you need an actual idea.
What makes a good IA study
For the IB Diploma Programme, before you fall in love with a topic, check it against four filters. A strong IA idea has all four.
- A simple IV with two conditions. One thing you change, split into exactly two levels — deep vs shallow, colour matches word vs it does not, positive framing vs negative framing. Two conditions keep your statistics and design manageable.
- An objective, measurable DV. Something you can count or time: number of words recalled, reaction time in seconds, a rating on a fixed scale, the value a participant estimates. Avoid vague outcomes like "how engaged they seemed."
- Ethical from the start. You must be able to run it on peers with informed consent and no distress. If the design needs deception that could upset people, or targets a vulnerable group, drop it.
- Based on a real, published study. You cite the original in your introduction and model your method on it. No source study means no theory to test and a weak introduction.
Ethics you must respect
For the IB Diploma Programme, ethics is not a box you tick at the end — it shapes which ideas are even allowed. The IA must follow standard research ethics, and moderators look for evidence you understood them.
- Informed consent in writing before anyone takes part, explaining broadly what they will do.
- No harm — nothing physically or psychologically distressing, embarrassing, or anxiety-inducing.
- No vulnerable participants — no children, no one who cannot freely consent. Use fellow students, aged 16+ where your school requires it.
- Right to withdraw at any point, including withdrawing their data afterwards, with no penalty.
- No unethical deception — mild, harmless withholding of the exact hypothesis is acceptable; deception that could cause upset is not. Debrief everyone at the end.
If an idea can only work by breaking one of these, it is the wrong idea.
Example experiments to replicate by area
For the IB Diploma Programme, each of these is a classic effect that reduces cleanly to two conditions and a measurable outcome. Pick one, find the original study, and simplify.
Memory
- Serial position effect — do people recall words better from the start and end of a list than the middle? Aim: to investigate the effect of a word's position in a list on recall. Hypothesis: words at the start and end of a list are recalled more often than words in the middle. Measure: the proportion of participants recalling each word by its list position.
- Levels of processing — does deep (meaning-based) encoding beat shallow (appearance-based) encoding? Aim: to investigate the effect of depth of processing on recall. Hypothesis: participants who process words for meaning recall more than those who process their appearance. Measure: number of words correctly recalled per condition.
- Chunking and digit span — does grouping digits improve short-term recall? Aim: to investigate the effect of chunking on recall of digit strings. Hypothesis: participants recall more digits when the string is chunked than when it is unbroken. Measure: number of digits recalled correctly.
- Reconstructive memory / leading questions — does the wording of a question change estimates? Aim: to investigate the effect of a leading verb on estimated values (e.g. speed). Hypothesis: a stronger verb produces higher estimates than a milder verb. Measure: the numerical estimate each participant gives.
Cognition and attention
- The Stroop effect — does a mismatched colour word slow down naming the ink colour? Aim: to investigate the effect of word–colour interference on naming time. Hypothesis: participants take longer to name ink colours when the word conflicts with the colour than when it matches. Measure: time in seconds to complete each list.
- Divided attention — does doing two things at once hurt performance? Aim: to investigate the effect of a secondary task on performance of a primary task. Hypothesis: accuracy or speed is lower when attention is divided than when it is not. Measure: errors made or time taken on the primary task.
- Schema and recall of a scene — does prior expectation shape what people remember? Aim: to investigate the effect of schema-consistent expectations on recall of objects in a scene. Hypothesis: participants recall more schema-consistent items and falsely recall expected-but-absent items. Measure: number of correctly and falsely recalled items.
Decision-making
- Framing effect — does positive vs negative wording change the choice people make? Aim: to investigate the effect of framing on decision-making. Hypothesis: participants choose the "safe" option more often when it is framed as a gain than as a loss. Measure: the proportion choosing each option per framing condition.
- Anchoring — does an initial number pull later estimates toward it? Aim: to investigate the effect of a numerical anchor on estimates. Hypothesis: participants given a high anchor produce higher estimates than those given a low anchor. Measure: the numerical estimate each participant gives.
Social
- Presence of others / social context — does having others around change performance on a simple task? Aim: to investigate the effect of the presence of others on performance of a simple task. Hypothesis: participants perform a simple task faster or more accurately when observed than when alone. Measure: time or accuracy on the task. Handle this one carefully: keep the task neutral and low-stakes so no one feels judged or embarrassed, and make observation clearly consented to.
Writing an aim and hypothesis
For the IB Diploma Programme, once you have chosen an effect, tighten it into one aim and two hypotheses. The aim states what you are investigating in a single sentence. The experimental (alternative) hypothesis predicts the difference between your two conditions and names both the IV and the DV operationally — for example, "Participants in the deep-processing condition will recall significantly more words than participants in the shallow-processing condition." The null hypothesis states there will be no significant difference. Make it directional only if your source study justifies a direction; otherwise keep it non-directional. Every idea above is written so its hypothesis follows in one line — that is the sign of a well-chosen study.
Common mistakes when choosing
This section covers Common mistakes when choosing — what IB examiners reward most often in past papers and coursework.
- Unethical designs. Anything that could stress, embarrass or deceive participants harmfully, or that uses vulnerable groups. If you cannot get clean consent, choose again.
- Too complex to run. Beautiful three-way designs, physiological measures, or studies needing specialist equipment collapse in a classroom. Simplify to two conditions and one measure.
- Correlational, not experimental. "Is screen time linked to sleep?" measures two things without manipulating anything. The IA must have an IV you actively change. If nothing is manipulated, it does not qualify.
- No real study base. Inventing your own effect leaves your introduction with no theory to explain and no method to model. Anchor to a published study you can cite.
How to refine your idea
For the IB Diploma Programme, start with the effect, then strip it down. Name the one variable you will manipulate and its two levels. Name the single number you will collect. Write the aim and hypothesis in one sentence each — if you cannot, the design is still too fuzzy. Check the ethics filter. Then pilot on two or three friends to catch confusing instructions and timing problems before the real run, and standardise everything you are not manipulating: same instructions, environment and timing, counterbalanced order if needed. A narrow, well-controlled replication scores far higher than an ambitious idea that falls apart in analysis.
How MarkScheme helps
Once you have a draft, you can [get an answer marked](/mark) to check your writing against IB assessment language and tighten weak sections before your supervisor deadline. The free [IB Psychology SL course](/ib/courses/psychology-sl) and [HL course](/ib/courses/psychology-hl) cover the research methods, ethics and statistics your IA leans on, and the broader [IB guides hub](/guides/ib) collects the rest of the Diploma support. If you are also chasing top exam marks, pair this with [how to get a 7 in IB Psychology](/blog/ib-psychology-how-to-get-a-7).
Frequently asked questions
For the IB Diploma Programme, any simple, ethical, published experiment with one manipulated variable and one measurable outcome. Popular choices include the Stroop effect, levels of processing, the serial position effect, chunking and digit span, the framing effect, anchoring, and reconstructive memory with leading questions. All reduce to two conditions and a countable or timed result.
What experiments can I do for the Psychology IA?
Does the Psychology IA have to be a true experiment?
Yes. The IA must be experimental — you manipulate an independent variable across conditions and measure a dependent variable. Correlational, observational or purely survey-based designs do not meet the requirement. If nothing is being actively changed by you, it is not an IA experiment.
How many conditions should my IV have?
Two is ideal. Two conditions keep the design, statistics and write-up manageable and map cleanly onto a single inferential test. More conditions add complexity that rarely earns extra credit and often causes analysis errors.
Do I have to base it on a published study?
Effectively yes. Your introduction must present the theory and the study you are replicating, and your method should be modelled on it. Without a real source study, your introduction is thin and your evaluation has nothing to compare against.
Can I use my classmates as participants?
Yes — a convenience sample of fellow students is standard, as long as they give informed consent, can withdraw, and are not a vulnerable group. Just discuss the sampling limitation in your evaluation.