Every QCAA Psychology student sits the same three kinds of internal assessment: a 10-mark data test, a 20-mark student experiment and a 20-mark research investigation. They use the same criteria as the other 2025 sciences, but Psychology changes what top-band work looks like. Your participants are people, so ethics is part of a mark. Your data are scores on memory tests, so the statistics have to match the design. And the behaviour your claim is about, such as "helping", is measured differently in almost every study you read.
This guide is for Year 12s starting Units 3 and 4 in Term 1 2027. It works through each IA under the current Psychology 2025 syllabus (v1.3, January 2026): the conditions, the marks and criteria, what the top band looks like in Psychology, and what QCAA's latest reports say students and schools get wrong. For the skills every science shares, such as justifying a modification or returning to the claim, it links to our student experiment guide and research investigation guide rather than repeating them.
The three Psychology IAs at a glance
Internal assessment is 50 of your 100 marks. The external exam is the other 50.
| Task | Marks and conditions |
|---|---|
| IA1 Data test, Unit 3 | 10 marks, one criterion. 60 minutes plus 5 minutes perusal, supervised, unseen data |
| IA2 Student experiment, Unit 3 | 20 marks, four criteria of 5. About 10 hours of class time, up to 2000 words or 11 minutes |
| IA3 Research investigation, Unit 4 | 20 marks, four criteria of 5. About 10 hours of class time, up to 2000 words or 11 minutes |
| External exam, Units 3 and 4 | 50 marks. Two papers, each 90 minutes plus 5 minutes perusal |
Unit 3, Individual thinking, covers brain function, sensation and perception, memory, and learning. Unit 4, The influence of others, covers social psychology, interpersonal processes, attitudes, and cross-cultural psychology. IA1 and IA2 draw on Unit 3. IA3 draws on Unit 4. To see how the four results combine, use the subject result calculator.
When each IA falls in 2027
Your school sets the exact dates, and your task sheet is the authority. QCAA's Senior Education Profile calendar sets the deadlines schools work back from.
- 27 January
Term 1 starts and Unit 3 begins. The practicals you run in these weeks are the source of the IA1 data and the base for IA2.
- Term 1
The data test. Schools must enter provisional marks for the first internal assessment by Tuesday 13 April 2027, the second day of Term 2, so expect IA1 in Term 1.
- Terms 1 and 2
The student experiment, usually with a draft checkpoint. Schools that opt in to QCAA's second confirmation event must enter IA2 provisional marks by Tuesday 8 June 2027. Other schools submit them with the remaining internal assessments on Tuesday 17 August 2027.
- Terms 2 and 3
The research investigation on Unit 4. Provisional marks for the remaining internal assessments, including IA3, are due on Tuesday 17 August 2027.
- 25 Oct to 16 Nov
The external assessment block. QCAA sets the day of the Psychology papers.
For how the whole cycle runs from task sheet to submission, see the internal assessment timeline.
IA1: the data test
The data test asks you to work with numbers and graphs from Unit 3 psychology experiments you may not have seen before. It is the only internal assessment sat under exam conditions.
About 30% of the marks go to working out unknown quantities or features of a dataset, about 30% to identifying trends, patterns, relationships, limitations or uncertainty, and about 40% to drawing conclusions based on analysis. In Psychology, that means calculating a mean, median, range or standard deviation from participants' scores; reading a scatterplot and its Pearson's r; and saying what a p-value, a standard error or a 95% confidence interval lets you conclude.
QCAA's own sample data test for this syllabus shows the range. It pairs hippocampal volume with years of experience in London taxi drivers, a class modification of Grant et al. (1998) on context and memory with a t-test result, and confidence intervals from a modification of Craik and Tulving (1975) on depth of processing.
How the data test is marked
There are no criteria to balance. Your raw score on the paper becomes a percentage, and the percentage becomes a mark out of 10 through fixed cut-offs. The top mark needs "consistent demonstration, across a range of scenarios", and more than 90% of the paper.
| Your raw score | IA1 mark |
|---|---|
| More than 90% | 10 |
| More than 80% | 9 |
| More than 70% | 8 |
| More than 60% | 7 |
| More than 50% | 6 |
| More than 40% | 5 |
| More than 30% | 4 |
| More than 20% | 3 |
| More than 10% | 2 |
| More than 1% | 1 |
| 1% or less | 0 |
Read "more than" literally. On a 30-mark paper, 27 out of 30 is exactly 90%, which is not more than 90%, so it earns 9. One raw mark can be one IA mark, which is why the habits below matter.
What full-marks answers do
- Justify with a number from the data. QCAA's 2025 report shows how a two-mark conclusion question is marked: one mark for a correct conclusion, and one for appropriate evidence from the data. A theory is not evidence from the data.
- Compare means a similarity and a difference. Under the 2025 syllabus, compare asks for similarities and differences without discussing significance. QCAA's endorsement advice gives a two-mark compare question one mark for each.
- Limitations means limitations in the data. A small sample, unequal standard deviations, outliers, or discrete scores. QCAA has told schools that questions about experimental methodology are outside the scope of the data test.
- Pick the measure that fits the data. The median suits data with obvious outliers. The mean suits interval data without them. QCAA's sample asks you to say why.
- Read what the error bars show. A standard deviation describes spread in the scores. A standard error or 95% confidence interval describes uncertainty about the mean. The graph's label tells you which.
Here is a two-mark item of the kind you will meet, with an original dataset. A class repeated Peterson and Peterson (1959): participants recalled consonant trigrams after counting backwards for 3, 9 or 18 seconds. Mean proportion recalled was 0.82 at 3 seconds (95% CI 0.74 to 0.90), 0.45 at 9 seconds (0.35 to 0.55) and 0.14 at 18 seconds (0.07 to 0.21). The question: Draw a conclusion about the effect of delay on recall. Justify your response.
1 of 2 marks
"Recall got worse as the delay got longer because information in short-term memory decays."
2 of 2 marks
"Recall decreased as the delay increased. Mean recall fell from 0.82 at 3 seconds to 0.14 at 18 seconds, and the 95% confidence intervals for those two delays do not overlap."
The first explains its conclusion with theory, which the question did not ask for. The second spends its second sentence on the data. Our guide to what QCAA's cognitive verbs ask for covers the other verbs on the paper.
How IA1 prepares you for the external exam
The data test is assessed on three of the external exam's four objectives. Two are worded almost exactly the same in both: analysing data to identify trends, patterns, relationships, limitations or uncertainty, and interpreting evidence to draw conclusions based on analysis. The third, applying understanding, is narrower in IA1, where it means working out unknown quantities from given data. The other difference is scope. The exam covers Units 3 and 4, across two 90-minute papers, and can ask you to calculate or to interpret unseen graphs, tables and diagrams.
The syllabus lists two class experiments whose data can be assessed in the external exam. In Unit 3, it is data about the context-dependency of memory from a modified experiment following Grant et al. (1998). In Unit 4, it is data about stereotype priming and behaviour from a modification of experiment 2 in Bargh, Chen and Burrows (1996). Treat both practicals as exam preparation, not just class activities.
QCAA's 2025 subject report shows the overlap. One 2025 exam question asked students to draw a conclusion from presented data about context-dependent cues and memory, and justify it. The excerpt QCAA chose justified the conclusion with the result of an inferential test of significance: the data test skill exactly. The report also asked for more practice with verbs such as deduce, infer, contrast and justify, and with using a question's marks to judge how much its answer needs.
IA2: the student experiment
You modify, which means refine, extend or redirect, an experiment relevant to Unit 3 to address your own related hypothesis or question. The base is a practical or simulation performed in class.
- Forming5 marks
- Finding5 marks
- Analysing5 marks
- Interpreting and Evaluating5 marks
Choose a practical that produces scores
The Unit 3 investigations the syllabus lists are all about memory: the context-dependency of memory (Grant et al. 1998), the duration of short-term memory (Peterson and Peterson 1959), its capacity (Miller 1956), encoding (Craik and Levy 1970), context-dependent cues (Tulving and Pearlstone 1966), and levels of processing (Elias and Perfetti 1973; Hyde and Jenkins 1973; Craik and Tulving 1975). Your school decides which of these you run in class.
Whichever you modify, it has to give you something to analyse. QCAA's June 2026 endorsement report told schools that practicals such as Bugelski and Alampay (1961), the rat-man ambiguous figure, yield categorical data and are not appropriate options, because they do not produce enough raw data for high-level analysis. The 2025 subject report pointed instead to experiments that produce ordinal, interval or ratio measurements and support descriptive and inferential statistics. A count of words recalled does. Which of two pictures someone reported seeing does not.
Forming: operationalise before you modify
The top band of Forming asks for "a considered rationale for the experiment", "justified modifications to the methodology", "a specific and relevant research question", "a methodology that enables the collection of sufficient and relevant data", and "appropriate use of genre and referencing conventions". In Psychology, "specific" starts with operationalised variables. QCAA's 2025 report singled out research questions and hypotheses that "clearly described the independent and dependent variables so the reader could understand what had been measured".
Relevant, not specific
"Does background noise affect memory?"
Specific and relevant
"Do students in Years 10 to 12 recall more words from a 20-word list when they are tested in the same recorded café noise they learnt in than when they are tested in silence?"
The second names the independent variable (whether the test context matches the learning context), the dependent variable (words recalled out of 20), and the participants. It also turns straight into the null and alternative hypotheses that the syllabus's inquiry skills include. Null: there will be no difference in the number of words recalled between the matching and mismatching conditions. Alternative: participants will recall more words in the matching condition than in the mismatching condition.
The modification is where most of Forming is decided. Our student experiment guide explains the general rule: a feasible change caps you at 2 to 3, and a justified one names the limitation, the change, and what the change makes possible. In Psychology, the limitations worth fixing are usually about people.
Feasible
"We repeated the class context experiment, but each participant did both conditions."
Justified
"In the class version, ten different students sat each condition, and scores within each group ranged from 3 to 12 words. A two-word difference between conditions could have come from who happened to be in which group. This modification uses a repeated-measures design: each participant learns two matched 20-word lists in café noise, and is tested on one in café noise and the other in silence. Each person acts as their own control, which removes differences in memory ability between groups, and counterbalancing the order of conditions and lists controls practice and fatigue effects."
Extraneous and confounding variables, the type of participant selection and allocation, counterbalancing and standardised instructions are all named in the syllabus's inquiry skills. Naming them, and saying what each controls, is what makes a Psychology modification read as justified.
Finding: ethics is part of the mark
Finding is worth a quarter of the task, and in Psychology its first line is about the people in your study.
-
4–5
considered
considered management of risks/ethical issues/environmental issues; collection of sufficient and relevant raw data; fluent and concise use of scientific language and representations
-
2–3
managed
management of risks/ethical issues/environmental issues; collection of relevant raw data; competent use of scientific language and representations
-
1
inadequate
inadequate management of risks/ethical issues/environmental issues; collection of insufficient and irrelevant raw data; simplistic use of language and representations.
The syllabus sets firm rules for research involving people. Any experiment involving humans needs their written permission, and students under 16 should also have written permission from a parent or carer. All participants should be over 12 and not vulnerable or at risk. They must understand the purpose of the study and know they can withdraw at any stage. Studies that create harm, distress or discomfort are not permitted, and that includes anything involving ingestion, such as food, drink, smoking or drugs, or deprivation, such as of sleep or food. All data must be kept confidential, and every participant must be anonymous.
Those rules close off some tempting modifications, such as an energy-drink condition or a "tired versus rested" condition. Check your idea against them before you plan it.
What separates "considered" management from plain management is the reason. QCAA's 2025 report described the strong responses as ones where management of ethical issues "was communicated using explanations of why particular strategies were used", with informed consent for minors as its example.
Two participants were 15, so a parent or carer also signed their consent form, as the syllabus requires for anyone under 16. Before consenting, every participant was told the study was about noise and memory, so their consent was informed, and that they could stop at any time without giving a reason, so participation stayed voluntary. The café recording was played at normal conversation volume so that no condition caused discomfort, and scores were recorded against participant codes so no result could be traced to a classmate. Each participant was debriefed with the class results after the final session, so no one left with a false impression of how they performed.
Why this scores: every strategy comes with its reason. A list of the same strategies without the reasons describes management, which is the 2–3 band.
The other two lines of Finding are lost by omission: show every participant's raw scores, not just the means, and name the design, variables and statistics correctly every time.
Analysing: match the statistics to the design
The top band of Analysing asks for "correct and relevant processing of data", "thorough identification of relevant trends/patterns/relationships", and "thorough and appropriate identification of the uncertainty and limitations of evidence". The 2025 syllabus gives you the tools: mean and median; range, interquartile range and standard deviation for spread; standard error and confidence intervals for uncertainty; Pearson's r for correlation; and the paired or unpaired two-sample t-test for differences.
QCAA's 2025 report flagged two points about Analysing under the 2025 syllabus. Your statistical test has to suit your methodology, and your graph has to match the analysis you ran. And you need to know the difference between variability and statistical uncertainty when you plot error bars. In practice:
- Repeated measures or matched participants: a paired t-test. Independent groups: an unpaired t-test. Two measured variables: Pearson's r and a scatterplot with a trendline.
- Say in the caption whether your error bars show standard deviation (variability) or standard error or a confidence interval (uncertainty about the mean).
- Report the p-value and what it lets you infer, allow for a type I or type II error, and say whether a small sample suits the test.
Mean recall was 13.1 words (SD 2.4) when the learning and test contexts matched and 11.2 words (SD 3.1) when they did not. Because every participant completed both conditions, a paired t-test was used, and the difference was statistically significant (p = 0.01). The error bars in Figure 2 show one standard deviation, so they describe the spread of individual scores, not uncertainty in the means. The larger spread in the mismatching condition came mainly from three participants who recalled fewer than 7 words. With 16 participants from one school, the sample is small and was not randomly selected, which makes it harder to judge whether each participant's difference between conditions is normally distributed and limits how far the result generalises.
Why this scores: the test is justified by the design, the error bars are named for what they show, the trend is quantified, and the uncertainty and limitations come from this data rather than from a generic paragraph about human error.
Its 2019 predecessor, Analysis of evidence, was the criterion QCAA most often marked down when it checked schools' marks in 2025. More on that below.
Interpreting and Evaluating: reliability and validity in Psychology
The top band needs "justified conclusion/s linked to the research question", a "justified discussion of the reliability and validity of the experimental process", and "suggested improvements and extensions to the experiment that are logically derived from the analysis of evidence". The middle band accepts reliability and/or validity, and improvements and/or extensions. The top band needs both halves of each.
The syllabus gives Psychology its own vocabulary for this discussion. Use it, and tie each point to your data.
Internal validity
Did the independent variable cause the change? Check the confounds your design could not rule out, such as one word list being harder or participants guessing the aim.
External validity
Does the result generalise? A convenience sample of 16 students from one school limits how far it reaches beyond them.
Reliability of the measure
Would the same responses get the same score? Decide in advance whether a misspelt word counts, and have a second person score a few sheets.
Validity of the measure
Does words recalled in two minutes measure what your question means by memory? Say what it captures and what it misses.
QCAA's 2025 report praised improvements and extensions that referred "directly to their effect on related issues with reliability and validity that were identified in the analysis of evidence". If your analysis found three low scorers in one condition who said a corridor conversation distracted them, the improvement is a closed room and a log of interruptions, not "more participants". The extension is the next question your data raises, such as whether the context effect survives a 24-hour delay.
IA3: the research investigation
You gather evidence related to a research question to evaluate a claim relevant to Unit 4. Your teacher provides a list of claims. QCAA's sample instrument uses three: "Gender is a social construct", "Social media is changing the nature of relationships", and "Violent media causes violent behaviour".
- Forming and Finding5 marks
- Analysing5 marks
- Interpreting5 marks
- Evaluating5 marks
There is no Communication criterion any more. Genre and referencing conventions are marked in Forming and Finding, and use of scientific language/representations in Interpreting, which has a descriptor at every performance level. For referencing, see our guide to referencing and using sources.
From claim to research question
A good claim is a single assertion with one broad independent variable and one broad dependent variable. QCAA's 2025 report gives an example of rewriting a claim that made three assertions into one: "the presence of others affects the way we behave". Your research question then needs variables more specific than the claim's. QCAA's advice on marking under the 2025 syllabus says research questions should "specifically address the variables used to provide a response to claim".
As broad as the claim
"Do people act differently when others are around?"
Specific and relevant
"Does the number of passive bystanders present reduce the likelihood that an adult helps a stranger in an emergency that poses no danger to the helper?"
"The presence of others" became the number of passive bystanders. "The way we behave" became the likelihood of helping. The scope, adults and non-dangerous emergencies, makes the question answerable with published studies inside 2000 words. Your rationale should show that path, step by step, from the claim to the question.
Analysing: check how each study measured the construct
The top band of Analysing asks for "the identification of sufficient and relevant evidence", "thorough identification of relevant trends/patterns/relationships in evidence", and "thorough and appropriate identification of limitations of evidence". In Psychology, the trap is the construct. "Helping" might mean reporting smoke filling a room in one study, responding to a seizure heard over an intercom in another, and picking up a stranger's dropped pens in a third.
Before you claim a trend across sources, make a short table for yourself: what each study measured as the dependent variable, who the participants were, the design, and the setting. A pattern that holds across different measures is strong evidence. A pattern that appears only with one kind of measure is a limitation you should name, and naming it is how you reach "thorough and appropriate".
Interpreting: say what the design can support
Interpreting needs "justified scientific argument/s" and a "justified conclusion linked to the research question". In Psychology, a justified argument is careful about what kind of evidence it rests on. Experiments can support a causal statement. Correlational and longitudinal studies support an association. If your claim uses the word "causes", say which of your sources could actually test cause, and keep your conclusion to what they show.
Evaluating: judge the evidence, then return to the claim
Its 2019 predecessor, Conclusion and evaluation, was the research investigation criterion QCAA most often marked down when it checked schools' marks in 2025. Here is the top band.
-
4–5
justified and extrapolated
justified discussion of the quality of evidence; extrapolation of credible findings of the research to the claim; suggested improvements and extensions to the investigation that are considered and relevant to the claim
The 2–3 band asks only for a "reasonable description of the quality of evidence", "application of relevant findings of the research to the claim", and improvements and/or extensions.
For the quality of evidence, Psychology gives you four questions that generic advice misses:
- Who was studied? The syllabus notes that most psychological research has been done in Western, educated, industrialised, rich and democratic societies and may not fully apply elsewhere. University students are not everyone.
- How was it studied? A laboratory experiment gives control but can feel artificial. A field study is more realistic but less controlled. A survey depends on honest self-report.
- When was it done? Classic studies from the 1960s and 1970s ran before current ethics guidelines, some cannot be repeated as designed, and the Stanford Prison Experiment has been critiqued for demand characteristics. Older does not mean wrong, but say what it means for your claim.
- Has it been replicated? One striking study is weaker than a consistent finding across many.
Then go back to the claim. Your findings about bystanders and helping support "the presence of others affects the way we behave" for one behaviour, under the conditions your sources tested. That is extrapolation, and saying it does not establish the claim for every behaviour is part of a justified evaluation. QCAA's report wants improvements and extensions discussed in terms of "improving the investigation's ability to evaluate the claim", so an extension that tests a second behaviour, such as conformity, does more than one that adds sources on the same question. Our research investigation guide explains why the conclusion to the research question and the return to the claim are separate jobs.
What QCAA's 2025 and 2026 reports flag
QCAA's Psychology subject report: 2025 cohort, published in January 2026, reviewed the last cohort assessed under the 2019 syllabus. Its criterion names are the old ones, but it also lists the key differences for the 2025 syllabus, which is the one you will be marked against. When QCAA checked schools' marks through confirmation, it agreed with the IA1 marks of all 237 schools. In IA2 and IA3, some criteria were marked down at more schools than others. The figures below are percentages of those 237 schools, not of individual students' responses.
| Criterion (2019 name) | Schools marked down |
|---|---|
| IA2 Analysis of evidence | 12.66% (30 of 237) |
| IA2 Interpretation and evaluation | 7.59% (18) |
| IA2 Research and planning | 7.17% (17) |
| IA3 Conclusion and evaluation | 9.28% (22) |
| IA3 Research and planning | 4.22% (10) |
| IA3 Analysis and interpretation | 3.80% (9) |
For the student experiment, the report notes that the criteria have been reorganised but the way teachers judge each characteristic has not changed. The closest 2025 home for IA2's most-lowered criterion is Analysing. For IA3's, it is Evaluating, with the conclusion to the research question now marked in Interpreting. The student-facing points the 2025 report and the June 2026 endorsement report raise for the current syllabus are built into the sections above: a practical that produces scores, both variables named, a reason for each ethical step, statistics and error bars that match the design, improvements derived from the analysis, and a return to the claim.
QCAA's report on the 2026 cohort, the first assessed in Units 3 and 4 under this syllabus, should follow the same January pattern. Read it in Term 1. Our guide to QCAA subject reports shows where to find it and how to use it.
Are there cut-offs for Psychology assessments?
Only in IA1. The data test turns a percentage into a mark through the cut-offs above. IA2 and IA3 have no percentage cut-offs: your teacher matches your response to the descriptors for each criterion, and our guide on how to read an ISMG explains how a mark is chosen within a band. For why an IA mark out of 20 has no letter grade, see what your ISMG mark means.
The A to E cut-offs apply only to your final result out of 100. For the 2025 Psychology cohort they were A for 87 to 100, B for 72 to 86, C for 50 to 71, D for 20 to 49 and E for 0 to 19. Of 5,888 students who completed Units 3 and 4, 27.34% received an A, 48.95% a B and 23.10% a C. The boundaries are set each year and can move, so treat them as a guide. For how the internal and external marks combine, see how your QCAA subject result is built.
A Term 1 plan for 2027
- Before Term 1: revise the statistics from Units 1 and 2: mean and median, standard deviation, standard error, confidence intervals, Pearson's r and the t-test. Check your calculator is on QCAA's approved list.
- After each Unit 3 practical: write a half-page summary of the design, the variables, the statistics used, and what the graph showed. The data test draws on these practicals.
- Before the data test: practise two-mark "draw a conclusion and justify" items until quoting a number from the data is automatic. When the test comes back, do the verb audit.
- When IA2 starts: choose a practical that produces scores, and send consent forms, including parent or carer forms for anyone under 16, in the first week.
Key takeaways
- IA1 is all data, and it is exam practice. Justify with a number, read the error bars, and keep limitations to the data.
- IA2 starts with a practical that produces scores. Operationalise both variables, justify the modification with controls, and give a reason for every ethical step.
- Match the statistics to the design. Paired or unpaired, variability or uncertainty, and a graph that shows what the test tested.
- IA3 turns on the construct and the claim. Compare how each study measured behaviour, keep causal language for causal evidence, and take your findings back to the claim.
Recommended
- the science student experiment, criterion by criterion
- the science research investigation, criterion by criterion
- what QCAA's cognitive verbs ask for in the data test
- how to use QCAA subject reports
- the Biology IAs, for comparison with another science
- the Health IAs, where you collect survey data about your own community
- finding the marks you are leaving behind on your ISMG
- what your ISMG mark means, and why it has no letter grade
- the QCAA sciences hub
- the QCAA subject result calculator



