Psychology IA Ideas Examiner-ranked experiments · 2026
Open the Psychology IA frame →

24 IB Psychology IA ideas that score highly

Experienced IB examiners's pick of Psychology Internal Assessment experiments for 2026 — each a simple, ethical partial replication of a published study, sorted by area, with the study to replicate, the IV and DV and why it scores. Choose one, then plan it in our examiner-written Psychology IA writing frame.

What makes a Psychology IA experiment score? A strong IA is a simple, ethical replication of an established published study with one clearly operationalised independent variable (exactly what you manipulate, in two conditions) and one clearly operationalised dependent variable (exactly the number you measure) — with no stress and no meaningful deception, a feasible sample of consenting participants, and results you can analyse with descriptive statistics plus one simple inferential test. Every idea below is built to tick all of these — start from a known effect, not a study you invent.

Found an experiment you like?

Drop it straight into the free Psychology IA frame. The planning sections are free; unlock the full step-by-step IA — operationalised IV and DV, hypotheses, the right inferential test, results and evaluation — to take it to the top band.

Start this IA in the Psychology frame →

MEMORY

Memory effects are the examiner's favourite for a reason: a clean two-condition manipulation, a countable recall score and a clear theory to test.

1 · Does depth of processing (shallow vs deep) affect the number of words correctly recalled?

Study replicated: Craik & Tulving (1975), levels of processing · IV: encoding task — shallow/structural (e.g. "in capitals?") vs deep/semantic (e.g. "fits this sentence?") · DV: number of words correctly recalled in a surprise free-recall test

The model Psychology IA: a clean manipulation, a strictly ethical surprise-recall task, a countable DV, and a textbook theory to test. An independent-measures design gives you a t-test (or Mann–Whitney U) out of the box.

🧠 cognitiveethical📊 statistics

2 · Does the serial position of a word affect the probability it is recalled?

Study replicated: Glanzer & Cunitz (1966), serial position effect · IV: position of the word in the list (primacy / middle / recency) · DV: proportion of words correctly recalled at each position

The classic U-shaped curve makes a striking graph and links straight to the multi-store model. Add an interference/delay condition and you have a clean two-group inferential test.

🧠 cognitiveethical📊 statistics

3 · Does a schema-providing title affect recall of an ambiguous passage?

Study replicated: Bransford & Johnson (1972), schema theory · IV: context given before reading — relevant title vs no title · DV: number of idea-units correctly recalled

A simple, ethical text task with a strong theoretical payoff (schema-driven encoding). The two conditions give a direct comparison of means and an easy inferential test.

🧠 cognitiveethical📊 statistics

4 · Does the verb in a leading question affect estimated speed (the misinformation effect)?

Study replicated: Loftus & Palmer (1974), reconstructive memory · IV: verb in the question ("smashed" vs "hit") about the same short clip · DV: estimated speed in mph

A famous, fully ethical demonstration of the misinformation effect. Two wordings, one number per participant — a textbook independent-samples comparison with a real-world hook.

🧠 cognitiveethical📊 statistics

ATTENTION & PERCEPTION

Attention tasks give you precise reaction-time or accuracy data — perfect for descriptive statistics and a clean inferential test.

5 · Does colour–word congruence affect naming reaction time? (the Stroop effect)

Study replicated: Stroop (1935) · IV: congruent vs incongruent colour words · DV: time taken (s) to name the ink colours of the list

Reliable, robust and completely ethical, with interval reaction-time data ideal for a repeated- or independent-measures t-test. A near-guaranteed significant result that still demands real interpretation.

🧠 cognitiveethical📊 statistics

6 · Does divided attention affect performance on a simple detection task?

Study replicated: dual-task / attention paradigms (e.g. Strayer & Johnston, 2001, simplified) · IV: single-task vs concurrent secondary task · DV: number of targets correctly detected

A safe, classroom-friendly test of attentional capacity with a countable DV. Keep the secondary task harmless (e.g. counting backwards) and you have a clean, ethical two-condition design.

🧠 cognitiveethical📊 statistics

7 · Does the Müller-Lyer illusion alter judged line length?

Study replicated: Müller-Lyer illusion (perception) · IV: fin direction (inward vs outward arrowheads) · DV: estimated/adjusted line length (mm)

A perfectly ethical perception study with a measurable, quantitative DV and a clear cognitive explanation. Two conditions, paired or independent — straightforward descriptive and inferential analysis.

🧠 cognitiveethical📊 statistics

DECISION-MAKING & COGNITIVE BIAS

Bias studies manipulate a single cue and measure an estimate or a choice — short, ethical and rich in theory.

8 · Does a high vs low numerical anchor affect estimates of an unknown quantity?

Study replicated: Tversky & Kahneman (1974), anchoring · IV: anchor shown first (high vs low) · DV: the numerical estimate given

A two-minute, fully ethical questionnaire study with a beautifully clean manipulation and an interval DV. A textbook independent-samples t-test, with dual-process theory to discuss in the evaluation.

🧠 cognitiveethical📊 statistics

9 · Does positive vs negative framing of identical information affect choice?

Study replicated: Tversky & Kahneman (1981), framing effect · IV: "gain" vs "loss" wording of the same scenario · DV: proportion choosing the risky option (or rating)

One scenario, two wordings, a categorical or rated choice — ethical and quick. A chi-square (or t-test on ratings) makes the inferential step simple and the prospect-theory link gives depth.

🧠 cognitiveethical📊 statistics

10 · Does priming with a concept affect responses on an unrelated task?

Study replicated: semantic priming (e.g. scrambled-sentence priming, simplified) · IV: prime category presented vs neutral prime · DV: reaction time or rating on the target task

Keep the prime neutral and harmless and this is a clean, ethical demonstration of automatic processing with quantitative data. Avoid stereotype or emotional primes to stay within ethics.

🧠 cognitiveethical📊 statistics

11 · Does the availability heuristic affect frequency estimates?

Study replicated: Tversky & Kahneman (1973), availability heuristic · IV: ease of retrieval (easy- vs hard-to-recall examples requested) · DV: estimated frequency of the event

A short, ethical estimation task with an interval DV and a clear bias to test. Two conditions give a direct mean comparison and a tidy inferential test.

🧠 cognitiveethical📊 statistics

12 · Does the decoy (attraction) effect shift preference between two options?

Study replicated: Huber, Payne & Puto (1982), attraction effect · IV: presence vs absence of a decoy option · DV: proportion choosing the target option

A neat, ethical consumer-choice manipulation with a categorical DV ideal for a chi-square. Strong real-world hook and a clear cognitive explanation for the evaluation.

🧠 cognitiveethical📊 statistics

Ready to write it up properly?

The Psychology IA frame walks you through every criterion — and the paid unlock builds your operationalised variables, hypotheses, the right statistical test and evaluation into one export-ready report.

Open the Psychology IA frame →

SOCIAL INFLUENCE & SOCIAL COGNITION

Social effects are powerful, but keep them gentle and ethical — manipulate information, not people's emotions.

13 · Does a unanimous majority's stated answer affect a person's judgement on a simple task?

Study replicated: Asch (1951), conformity — simplified and ethical · IV: confederate majority gives the wrong answer vs no majority · DV: number of trials the participant conforms

A famous effect that can be replicated ethically on a trivial line- or count-judgement task — no stress, full debrief. A clean count DV and a two-condition inferential test.

👥 socialethical📊 statistics

14 · Does a social-norm message affect a stated intention or simple choice?

Study replicated: social-norms approach (Cialdini-style, simplified) · IV: norm message present ("most students do X") vs absent · DV: rated likelihood / proportion choosing the behaviour

A harmless, ethical message manipulation with a rated or categorical DV. Two conditions, a t-test or chi-square, and a clear social-cognition theory to evaluate.

👥 socialethical📊 statistics

15 · Does a person's labelled "warm" vs "cold" trait change impression ratings?

Study replicated: Asch (1946) / Kelley (1950), central traits · IV: one trait word in a description ("warm" vs "cold") · DV: mean rating of the person on a set of trait scales

A classic impression-formation study, fully ethical, with one word changed and an interval rating DV. A direct mean comparison and a strong primacy/central-trait discussion.

👥 socialethical📊 statistics

16 · Does an actor's behaviour being framed as situational vs dispositional change attributions?

Study replicated: fundamental attribution error (Jones & Harris, 1967, simplified) · IV: behaviour described as freely chosen vs assigned · DV: rated dispositional attribution score

An ethical vignette study with an interval rating DV and a well-known social-cognition bias to test. Two conditions give a clean comparison and a t-test.

👥 socialethical📊 statistics

LANGUAGE & THINKING

Language manipulations are quick, ethical and produce countable data — a strong, often-overlooked area.

17 · Does the label given to a colour boundary affect recognition (a linguistic-relativity test)?

Study replicated: linguistic relativity / categorical perception (Winawer et al., 2007, simplified) · IV: colours from same vs different named categories · DV: reaction time or accuracy in a discrimination task

An ethical, quantitative test of whether language shapes perception, with reaction-time or accuracy data. Two within-subject conditions give a clean paired inferential test.

🧠 cognitiveethical📊 statistics

18 · Does counting in a verbal vs visuospatial concurrent task interfere more with mental arithmetic?

Study replicated: working-memory / articulatory-suppression studies (Baddeley, simplified) · IV: verbal suppression vs spatial tapping during the task · DV: number of arithmetic items correct

A safe, classroom-ready dual-task design that tests working-memory components, with a countable DV and a clean two-condition inferential test.

🧠 cognitiveethical📊 statistics

19 · Does the phrasing of a problem (concrete vs abstract) affect reasoning accuracy?

Study replicated: Wason selection task — thematic version (Johnson-Laird et al., 1972) · IV: abstract vs concrete/realistic problem content · DV: proportion solving the task correctly

An ethical reasoning study with a categorical accuracy DV ideal for a chi-square, and a rich dual-process discussion for the evaluation.

🧠 cognitiveethical📊 statistics

20 · Does presenting numbers as words vs digits affect estimation or recall?

Study replicated: numerical-format / dual-coding effects · IV: quantities shown as words ("thirty") vs digits ("30") · DV: recall accuracy or estimation error

A quick, ethical format manipulation with a countable DV and a clear cognitive-processing explanation. Two conditions, an easy descriptive and inferential analysis.

🧠 cognitiveethical📊 statistics

21 · Does a sentence's context bias interpretation of an ambiguous word?

Study replicated: lexical ambiguity resolution (Swinney, 1979, simplified) · IV: biasing vs neutral sentence context · DV: proportion choosing the context-consistent meaning

An ethical language-comprehension study with a categorical DV for a chi-square and a clear story about context-driven processing.

🧠 cognitiveethical📊 statistics

22 · Does chunking information change how much is recalled from short-term memory?

Study replicated: Miller (1956), chunking and capacity · IV: items presented as unchunked vs chunked strings · DV: number of items correctly recalled in order

A safe, ethical capacity study with a countable DV and a direct test of chunking theory. Two conditions give a clean mean comparison and inferential test.

🧠 cognitiveethical📊 statistics

23 · Does background music vs silence affect performance on a comprehension task?

Study replicated: irrelevant-sound effect (Salamé & Baddeley, 1982, simplified) · IV: instrumental music vs silence during the task · DV: number of comprehension questions correct

A genuinely ethical, easy-to-run study with a countable DV. Keep the audio neutral (no lyrics, no distressing content) for a clean two-condition comparison.

🧠 cognitiveethical📊 statistics

24 · Does encoding context match at recall (context-dependent memory) affect recall scores?

Study replicated: Godden & Baddeley (1975), context-dependent memory (room-based, ethical) · IV: recall in the same vs a different room/context · DV: number of words correctly recalled

A safe, school-friendly adaptation of a classic with a countable DV and a clear encoding-specificity theory. Two conditions give a direct, testable comparison.

🧠 cognitiveethical📊 statistics

From a topic to a top-band IA

An idea is the easy part — the marks are in how you build it. The Psychology IA is scored out of 22 across four criteria: A Introduction (/6), B Exploration (/4), C Analysis (/6) and D Evaluation (/6), in a report of about 1,800–2,200 words. Whichever experiment you pick, the same moves win: a focused aim drawn from the original study and its theory, one operationalised IV and DV with a research and a null hypothesis, a justified design with proper sampling and controls, data collected strictly ethically, descriptive statistics plus one appropriate inferential test reported and interpreted in full, and an evaluation that weighs the strengths and limitations of your design against the original with realistic improvements.

Build your chosen experiment into a full IA

The examiner-written Psychology IA writing frame takes you through every criterion with the rubric, worked examples and the traps that cost marks. The planning sections are free — unlock Exploration, Analysis & Evaluation to finish the whole report and export it to Word or PDF.

Open the Psychology IA frame →

Psychology IA ideas — FAQ

What makes a good IB Psychology IA topic?

A simple, ethical experiment that partially replicates an established published study, with one clearly operationalised independent variable (exactly what you manipulate, in two conditions) and one clearly operationalised dependent variable (the number you measure — recall scores, reaction times, ratings or counts), no stress and no meaningful deception, a feasible sample of consenting participants, and results you can analyse with descriptive statistics plus one simple inferential test. Start from a known effect, not a study you invent.

What sample size and statistics do I need?

A manageable sample of consenting participants — often around 15–30 per condition — is enough. Report descriptive statistics first (a measure of central tendency and one of dispersion, with a clear graph), then one appropriate inferential test: an independent-measures design with interval data points to an independent-samples t-test; ordinal data or a broken assumption points to a non-parametric test such as Mann–Whitney U. State the test statistic, the p-value and the decision on the null hypothesis in full.

What is NOT allowed on ethical grounds?

No undue stress, pain or anxiety; no meaningful deception; no clinical or sensitive topics (mental illness, trauma, illegal behaviour); no testing minors without proper consent. You must have informed consent, the right to withdraw, anonymity and a debrief. Studies like Milgram or Zimbardo cannot be replicated — choose a simple cognitive or social effect that is safe to reproduce in a classroom.

How do I turn the idea into a top-band IA?

Build it section by section in the free Psychology IA writing frame — aim and theory, operationalised IV and DV, research and null hypotheses, a justified ethical design, descriptive and inferential statistics, and an evaluation that compares your findings to the original study with realistic improvements.

📬 Free: the IA topic-picker checklist + examiner tips

Get the Topic-Picker & Top-Band Checklist (PDF) plus short, examiner-written tips for each stage of your IA — straight to your inbox. No spam, unsubscribe any time.

IA ideas for other subjects

Biology IA → Sports Science IA → Maths AI → All IA tools →