Common mistakes · Psychology · 8 min read
IB Psychology mistakes: 12 habits that cost marks, and the fix for each
What students write, why it loses marks, and what earns them instead, for Papers 1, 2 and 3 on the new course.
On this page
The mistakes that cost most marks in IB Psychology are rarely gaps in knowledge. They're describing a study without saying how it answers the question, evaluating with stock phrases like "small sample size", treating a correlation as a cause, and mixing up research-methods terms. Below are twelve, grouped by where they happen, each with a fix you can practise.
Everything here is written for the new course (first exams May 2027), and the paper references follow it. Most of the habits apply just as much if you're sitting the old course in November 2026. For the paper structure itself, see our IB Psychology exam guide.
Which mistakes happen when you use studies?
1. Describing a study without linking it to the question
The most common one. A paragraph retells a study accurately and stops. The examiner is left to work out why it's there, and won't.
Loses marks"Bandura, Ross and Ross (1961) had children watch an adult hit a Bobo doll. Later the children were left with the doll, and many copied the adult's aggressive actions."
Earns marks"…Children who had watched the aggressive model reproduced many of the adult's specific actions, which supports the claim that behaviour can be learned by observation without being reinforced directly. Because the copying was so precise, it is hard to explain as a general rise in arousal."
Fix: end every study with a sentence that starts from the question's wording: "This shows that…", "This supports the claim that… because…".
2. Writing out full procedures in Paper 1
Paper 1 doesn't assess detailed procedures. Six lines on participant numbers, ages and the room layout take time from the argument. Fix: one sentence of method, then the finding that matters for this question, then the link.
3. Memorising studies without the concept
Paper 1 Section C and Paper 2 Section B are built around concepts (bias, causality, change, measurement, perspective, responsibility). A student who learned "Study X for topic Y" has nothing to say when the question asks about measurement. Fix: for every study you know, write down which concepts it can illustrate and how. A correlational design is a causality example; a lab measure of a real-world behaviour is a measurement example; a sample from one culture is a bias or perspective example.
4. Misnaming researchers or blending studies
Under pressure, students merge studies: attributing the "lost in the mall" false-memory study to Loftus and Palmer, for example, when that was a different study by Loftus and Pickrell. A confident wrong citation suggests you don't know either study.
Loses marks"Loftus and Palmer (1974) implanted a false memory of being lost in a shopping mall."
Earns marks"In a classic eyewitness experiment, Loftus and Palmer found that changing one verb in a question (\"smashed\" rather than \"contacted\") raised participants' estimates of a car's speed."
Fix: if you're not sure of the name or year, describe the study accurately ("a 1970s eyewitness experiment in which…"). Accuracy about what happened matters more than the citation.
Which mistakes happen in evaluation?
5. Generic evaluation not tied to the claim
"Small sample size", "lacks ecological validity" and "culturally biased" can be written about almost any study. Without saying what the problem does to this conclusion, they earn little.
Loses marks"Loftus and Palmer's study lacks ecological validity because it was in a lab."
Earns marks"Participants watched short film clips and knew nothing depended on their answers. A real witness to a crash is emotionally involved and questioned later by police, so the study shows that wording can distort memory but not how much it does in court cases."
Fix: every evaluation point needs three parts: the feature of the study, the problem it causes, and the effect on the claim you're making.
6. Ethics points that don't apply
Students attach "it was unethical" to any study, including ones with no deception and no risk, or bring up ethics when the question is about measurement or causality. When ethics is relevant, be precise. In Milgram's (1963) obedience study, participants were told it was a study of learning and punishment, so deception was built into the design, and the experimenter's scripted prods to continue made withdrawing much harder than simply leaving. Those are specific, defensible ethics points. "Milgram was unethical because people were stressed" is not. Fix: only raise an ethical issue that the procedure actually created, and only when the question or concept calls for it (responsibility, for example).
7. Confusing correlation and causation
A correlation shows two variables move together. It doesn't show which causes which, or whether a third variable drives both.
Loses marks"The study found a positive correlation between screen time and anxiety, which proves that screens cause anxiety in teenagers."
Earns marks"The positive correlation means more screen time goes with higher anxiety, but anxious teenagers may use screens more to cope, and poor sleep could raise both. Only a design that manipulates screen time, or tracks both over time, could test the causal claim."
Fix: before claiming a cause, ask: was a variable manipulated, and were participants randomly allocated? If not, say "is associated with", not "causes".
Which research methods terms get mixed up?
8. Quasi-experiment, natural experiment and true experiment
| Design | Who sets the IV? | Random allocation? | Example |
|---|---|---|---|
| True (lab) experiment | Researcher manipulates it | Yes | Assigning students to a self-testing or a rereading revision condition |
| Field experiment | Researcher manipulates it, in a natural setting | Where possible | Changing the instructions on a real school's homework sheet |
| Quasi-experiment | A characteristic participants already have | No | Comparing bilingual and monolingual children's memory |
| Natural experiment | An event outside the researcher's control | No | Comparing behaviour before and after a new school phone ban |
A classic study can contain more than one: in Bandura's Bobo doll study the researchers decided which model each child saw (experimental), while the comparison between boys and girls is quasi-experimental, because sex can't be assigned. Some textbooks group natural experiments with quasi-experiments. Whatever label your teacher uses, the point that earns marks is the same: without random allocation, groups may differ in other ways, so causal claims are weaker.
9. Reliability and validity
Reliability is consistency: would the measure give the same result again, or would two observers code the same behaviour the same way? Validity is accuracy: does it measure what it claims to? Internal validity asks whether the IV really caused the change in the DV; ecological validity asks whether the findings hold in everyday settings. A measure can be reliable without being valid (a stopwatch started late by the same amount every time).
Loses marks"The study lacks reliability because it took place in a lab."
Earns marks"Being in a lab threatens ecological validity, not reliability. In fact, standardised lab procedures make the study easier to repeat, which supports its reliability."
For qualitative research (interviews, and HL Paper 3), use the matching terms: credibility (do the findings reflect participants' experiences?), transferability (could they apply to other settings?) and bias on the researcher's or participants' side.
Which mistakes happen in the exam room?
10. Ignoring the scenario in Paper 1 Section B
Section B gives you an unseen scenario and rewards applying psychology to it. Students write a textbook account of the theory and never mention the person in the question. Fix: underline two or three details in the scenario before you write, and use each one in your answer.
11. Running out of time on the 15-marker
Paper 1 and Paper 2 give you about 2.6 minutes per mark. Writing a full page for each 4-mark Section A answer leaves the 15-mark Section C rushed, and Section C is worth almost twice Section A. Fix: set checkpoints before you start. In Paper 1, aim to finish Section A by about 18 minutes and Section B by about 46, which leaves around 38 minutes for Section C plus a few to check. Our guide to IB exam time management has drills for holding the pace.
12. Leaving an approach or a context out of revision
On the new course all four contexts are compulsory, Paper 1 Section A asks about two of the three approaches without telling you which, and Section C offers a choice of only two questions from different contexts. A gap anywhere can leave you with no question you can answer well. Fix: build a grid of the four contexts against the three approaches and make sure every cell has at least one piece of content you can explain and evaluate.
In the 15-mark synthesis question, students either summarise each source in turn or ignore the sources and write from memory. The question asks you to judge a claim using at least three sources and your own knowledge. Group the sources by what they show, say how far each supports the claim, and finish with a judgement.
How do you stop repeating them?
Knowing the list isn't the same as not making the mistakes. Three habits help:
- Mark your practice answers against a mark scheme and write down the reason for each lost mark. If "no link" appears three times, that's your priority.
- Rewrite one paragraph, not the whole answer. Take the weakest evaluation point and rewrite it with feature, problem and effect.
- Learn the command terms. Most of these mistakes come from describing when the question says explain, evaluate or discuss; see IB command terms explained.
The same habits matter in the IA, where generic ethics and vague methods language cost marks too; see our IB Psychology IA guide.
Questions students ask
Is "small sample size" ever a good evaluation point?
Yes, when you say what it affects. A small sample matters if the effect is small or the participants are unusual, and you should say which conclusion it weakens. On its own, without that link, it is a stock phrase that earns little.
Do I lose marks for getting a study's date wrong?
A wrong year on its own is a small slip next to misdescribing what a study found. If you're unsure, leave the date out and describe the study accurately; a confident citation that blends two different studies does far more damage.
What is the difference between a quasi-experiment and a natural experiment?
In a quasi-experiment the groups differ on something participants already have, such as age or bilingualism. In a natural experiment, an event outside the researcher's control, such as a new policy, creates the conditions. Some textbooks treat natural experiments as a kind of quasi-experiment; either way, neither uses random allocation, so causal claims are weaker.
Try one: Psychology
Researchers gave 300 adolescents a questionnaire measuring their daily social media use and a loneliness scale. They found a positive correlation (r = 0.32) and concluded that social media use causes loneliness. Evaluate this conclusion with reference to causality.
[4 marks]Show the answer and why
- The study is correlational: neither variable was manipulated and participants were not allocated to conditions, so it cannot establish cause and effect.
- Direction is unclear: lonely adolescents may spend more time on social media because they are lonely (reverse causation).
- A third variable, such as low mood, poor sleep or having few offline friends, could increase both social media use and loneliness.
- Judgement: r = 0.32 shows only a modest positive relationship; testing causation would need, for example, a longitudinal design tracking both variables over time or an experiment randomly assigning some adolescents to reduce their use.
The tempting answer is "the sample is too small" or "self-report is unreliable". Neither addresses causality, which is the concept the question names. 300 is a reasonable sample for a correlation, and self-report is a measurement point.