IB Psychology HL common mistakes
IB Psychology HL · Understand the traps and how to avoid them.
In an extended-response answer on neurotransmitters and behaviour, the candidate spends most of the response narrating one key study's procedure (participants, apparatus, step-by-step method, results table) and then appends a single sentence such as "this shows that neurotransmitters affect behaviour."
Revision is organised around memorising study narratives because studies are easier to recall as stories than mechanisms. The candidate can retell the study but cannot state what the neurotransmitter actually does at the synapse, so the biological explanation is never present in the answer.
What to do instead
Rewrite every biological answer as repeating three-move cycles: mechanism (e.g. an agonist binds to receptors and increases neural firing), evidence (the specific finding that discriminates between competing explanations), and a link-back sentence naming the question. Impose a ratio: no more than one third of any paragraph may be procedural description, and underline the mechanism sentence in every cycle before submitting.
The candidate treats correlational findings as causal, writing that "stress shrinks the hippocampus" or "meditation increases cortical thickness" on the basis of a study that measured two variables at one time point, or generalises from a single case study (e.g. a patient with hippocampal damage) to the whole population without qualification.
Textbook summaries often phrase findings causally for brevity, and the candidate copies that phrasing without checking the original design. Small-n case and animal studies are also the most memorable evidence, so they get over-weighted.
What to do instead
Before writing, label each study with its design in the margin: experiment, quasi-experiment, correlational, case study, animal. Use hedged verbs that match the design ("is associated with", "co-occurs with") in correlational evidence, and convert the limitation into part of the evaluation: state what the design cannot establish and what would be needed to establish it.
Studies are attached to the wrong theory. A candidate evaluating working memory cites a misinformation-effect experiment, or answers a question on schema theory by describing the multi-store model, and the answer silently switches between models mid-paragraph.
Revision is organised by topic ("memory") rather than by theory, so studies are filed under a topic heading and retrieved by topic name. Multi-store and working memory are both "models of memory" and therefore merge in recall.
What to do instead
Build a matrix with one row per theory and columns for the study, the independent variable, the dependent variable, and the specific prediction the study confirms or disconfirms. If a study cannot be shown to discriminate between two theories, it does not belong in that row. Rehearse by covering the study column and recalling only studies whose prediction you can state.
Evaluation consists of generic methodological labels with no consequence stated: "the study was a lab experiment, so it had low ecological validity and was reductionist." The candidate never says what the artificiality or reductionism does to the conclusion being drawn about real cognitive processing.
Candidates are trained to produce lists of methodological strengths and limitations and are rewarded in class for naming terms. Naming a limitation feels like evaluating, so the extra step of connecting it to the conclusion is skipped.
What to do instead
Force every limitation into a three-part sentence: what the feature was, what it distorts, and which conclusion therefore becomes unsafe. For example: words were learned in a lab with no meaning attached, so elaborative encoding was suppressed, so the finding that acoustic similarity impairs recall may understate semantic effects in real study. If the third clause cannot be written, the limitation is not doing work and should be cut.
The candidate names a cultural variable and cites a cross-cultural comparison, but never explains the process through which culture produces the behaviour: for example, asserting that collectivist cultures are more conformist and citing a single comparison study, without invoking self-construal, social norms, social learning, or the meaning of the task for participants.
Cross-cultural studies are often taught as headline findings (country A vs country B) rather than as tests of a mechanism, so the candidate has the pattern but not the explanation. Culture also feels self-explanatory, so the mediating step seems unnecessary.
What to do instead
Write each sociocultural point as a chain: cultural context, mediating psychological process, measured behaviour. Practise by inserting "because" between the cultural label and the outcome and never allowing the sentence to end at the label. Also check whether the study used matched tasks and whether the measured behaviour meant the same thing in both samples.
On an "evaluate" or "to what extent" question, the candidate produces a sequence of supporting studies followed by contradicting studies and then stops, offering no overall judgement, no conditions under which the claim holds, and no weighing of which evidence is stronger.
The essay is treated as a container for content rather than as an argument. Time pressure and the fear of omitting a study push the candidate to keep adding evidence instead of adjudicating it.
What to do instead
Before writing, draft a one-sentence thesis with a condition in it ("social identity theory explains discrimination when group membership is salient, but not when participants interact repeatedly"). Reserve the final paragraph for the verdict and make it name the condition and the strongest single piece of evidence. Each body paragraph should end with a half-sentence stating what it does to the thesis, not just what the study found.
The candidate answers an options question using core-approach material or generic vocabulary instead of the option's own framework. For example, a question on classification or diagnosis is answered with a description of the biological approach, or a question on health behaviour is answered without naming any health-specific model, and the option's technical terms (validity of diagnosis, relapse, adherence, attachment classification) are absent.
Core material is revised most heavily and is rehearsed under exam pressure, so it is what surfaces first. Options are often studied late and in less depth, so the candidate substitutes a familiar answer for the requested one.
What to do instead
Build a one-page glossary and evidence sheet per option, listing its defining frameworks and the studies that test them, plus the terms that must appear. Before writing, decode the command term and the key noun, then check each paragraph against the glossary: if no option-specific term appears, the paragraph is off-target. Practise timed plans only, so the framework is selected before prose begins.
On a qualitative-research question, the candidate fails to identify or justify the qualitative approach: an interview study is described without naming thematic analysis, grounded theory, or case study, or the study is evaluated with the criteria for quantitative research (sample size, generalisability, statistical significance) rather than credibility, transferability, dependability, and confirmability or the equivalent criteria.
Qualitative methods are taught briefly and the candidate has few models of what a qualitative analysis section looks like. Quantitative evaluation vocabulary is the only evaluation vocabulary the candidate has automated.
What to do instead
Learn a decision rule: does the research question ask for meaning and experience (interview plus thematic analysis), theory generation (grounded theory), or an in-depth single instance (case study)? Memorise one justification sentence per approach, stating the question features that make the analysis appropriate. For evaluation, use a qualitative criteria checklist and ban the words "significant" and "generalisable" from those answers.
The candidate reports a calculated statistic without interpreting it: "the calculated value was 6, which is greater than the critical value of 5" and stops, or omits the significance level, or writes "p = 0.04 so there is no significant difference", or confuses Type I and Type II errors when asked what could have gone wrong.
Statistics is learned as a procedure (calculate, compare) rather than as a decision with an error rate. The alpha level, the direction of the test, and the meaning of rejecting the null hypothesis are never made explicit in practice.
What to do instead
Use a fixed four-sentence template every time: the test used and why it suits the data; the observed and critical values with the alpha level and one- or two-tailed decision; whether the null hypothesis is rejected; what that means in the terms of the study. Then rehearse the error link: rejecting a true null is a Type I error, failing to reject a false null is a Type II error, and smaller alpha reduces the first at the cost of the second.
In the IA the candidate states the independent and dependent variables in loose wording without saying how each was operationalised and measured, claims that extraneous variables were "controlled" without naming them or the procedure used, and then chooses an inferential test by habit rather than by matching it to the design (related or unrelated samples, nominal or ordinal data).
The IA is written last under deadline pressure, so the method and analysis sections are drafted from a template rather than from the actual procedure. The candidate has usually practised one test and reuses it regardless of design.
What to do instead
Complete a design-check table before drafting: IV with exact levels and operational definition, DV with the measurement scale and scoring rule, each named extraneous variable with the specific control, participants and allocation, and the resulting test choice. Paste the table into the method and justify the test in one sentence that cites the number of groups, the sample relationship, and the data level. Recheck that the raw data and calculations in the appendix let a reader reproduce every reported value.