Research methods and investigation design (key science skills): VCE Psychology Units 3 and 4
“Characteristics of the selected scientific methodology and method, and appropriateness of the use of independent, dependent and controlled variables in the selected scientific investigation; identify independent, dependent and controlled variables in controlled experiments; formulate hypotheses to focus investigations; determine appropriate investigation methodology: case study; classification and identification; controlled experiment (within-subjects design, between-subjects design, mixed design); correlational study; fieldwork; literature review; modelling; product, process or system development; simulation; select and use methods appropriate to the investigation, including consideration of sampling technique (random and stratified) and size to achieve representativeness”
A VCE hypothesis names the population, both levels of the independent variable, the dependent variable and a direction. The IV is manipulated and the DV measured; controlled variables are held constant; extraneous variables might affect the DV; confounding variables have affected it, reducing internal validity. Only a controlled experiment supports cause and effect. Between-subjects designs put each participant in one condition (no order effects, but participant differences), within-subjects designs use the same participants in all conditions (controls participant variables, but order effects, managed by counterbalancing), and mixed designs combine both. Random sampling gives everyone an equal chance of selection; stratified sampling selects from subgroups in their population proportions. Random allocation is separate from random sampling.
What this dot point is asking
VCAA examines research methods in every paper. The examination specifications state that "all of the key knowledge that underpins the outcomes in Units 3 and 4, and the key science skills" are examinable, and questions often build a whole stimulus around an investigation. In Unit 4 Area of Study 3 you also design and run your own investigation and present it as a scientific poster, so the same skills are assessed in your SAC.
This page covers the planning half of the key science skills:
- Develop aims and questions, formulate hypotheses and make predictions: identifying independent, dependent and controlled variables, and writing a hypothesis.
- Plan and conduct investigations: choosing a methodology, the three controlled experiment designs, and sampling (random and stratified) and sample size.
Evaluating data (accuracy, precision, validity, errors, outliers) is on our data analysis and evaluation page, and ethics on our ethics page.
The answer
Aim, research question and hypothesis
- The aim states what the investigation sets out to find, usually as "to investigate whether [IV] affects [DV] in [population]".
- The hypothesis is a testable prediction about the relationship between the variables. A full VCE hypothesis names the population, both levels of the independent variable, the dependent variable and the direction of the expected result.
A reliable template: "It is hypothesised that [population] who [IV condition 1] will [direction] [DV] compared to [population] who [IV condition 2]."
The five variables in the study design
The study design defines five types of variable:
- Independent variable (IV): the variable the researcher manipulates (controls, selects or changes) and assumes has a direct effect on the DV. It is plotted on the horizontal axis.
- Dependent variable (DV): the variable the researcher measures. It is plotted on the vertical axis.
- Controlled variable: a variable the researcher holds constant, so that changes in the DV can be attributed to the IV. A controlled variable is not the same as a control group.
- Extraneous variable: any variable other than the IV that may affect the DV. It should be controlled, or at least monitored.
- Confounding variable: a variable other than the IV that has affected the DV, giving an alternative explanation for the results and reducing internal validity. An extraneous variable that was not controlled can become confounding.
Choosing a methodology
The study design lists nine scientific investigation methodologies: case study, classification and identification, controlled experiment, correlational study, fieldwork, literature review, modelling, product, process or system development, and simulation. Two are examined most often:
- A controlled experiment tests the relationship between one or more IVs and a DV while controlling all other variables, sometimes with a control group. It is the only methodology here that can support a cause-and-effect conclusion.
- A correlational study observes and records variables that are not manipulated, to find associations and make predictions. It cannot show cause and effect, because a third variable may explain the association.
Fieldwork (observations, interviews, questionnaires, focus groups and yarning circles beyond the classroom) usually identifies correlations rather than causes. The study design also asks you to consider the cultural appropriateness and cultural safety of each methodology.
Controlled experiment designs
| Design | What happens | Main strength | Main limitation |
|---|---|---|---|
| Between-subjects | Each participant is in only one condition (experimental or control), ideally by random allocation | No order effects; conditions can run at the same time; lower withdrawal | Differences between the people in each group can confound results; needs more participants |
| Within-subjects | The same participants complete every condition | Controls participant-related variables; needs fewer participants | Order effects (practice, fatigue, boredom), managed by counterbalancing; takes longer |
| Mixed | Combines both: separate groups are compared, and each group is also measured more than once (for example before and after) | Compares change from each person's own baseline across groups | More complex and time consuming; still exposed to some order effects |
Random allocation means each participant has an equal chance of being placed in any group. It spreads participant characteristics evenly across groups and so helps control extraneous variables. Do not confuse it with random sampling, which is about who is selected from the population in the first place.
Counterbalancing controls order effects in a within-subjects design: half the participants complete condition A then B, and the other half B then A. Some IVs cannot be counterbalanced, for example once a participant has learnt something it cannot be unlearnt, so the control condition must come first (the point VCAA rewarded in 2025).
Sampling
The population is the whole group the research is about; the sample is the subset who take part. The study design names two techniques:
- Random sampling: every member of the population has an equal chance of being selected. With a large enough sample this tends to produce a representative sample, but by chance a small random sample can over-represent some groups.
- Stratified sampling: the population is divided into strata (subgroups) on a characteristic relevant to the study, such as age or year level, and participants are selected from each stratum in the same proportions as in the population. This guarantees that the key characteristic is represented. (When the selection within each stratum is random, this is stratified random sampling.)
Sample size matters as well: larger samples reduce the effect of chance differences and random error, and make results more likely to generalise. A small, non-representative sample (for example, a group of friends) limits external validity.
The question
"Design a controlled experiment to test whether a short daytime nap improves memory for a word list in Year 12 students." (4 marks)
Step 1: aim and hypothesis
Aim: to investigate whether a 20-minute nap affects recall of a word list in Year 12 students. Hypothesis: It is hypothesised that Year 12 students who take a 20-minute nap after learning a word list will recall more words than Year 12 students who stay awake for 20 minutes.
Step 2: variables
IV: nap or no nap (two levels). DV: number of words correctly recalled. Controlled variables: the word list, learning time, time of day, room, and the delay before recall.
Step 3: design and sampling
Use stratified sampling so each Year 12 class is represented in proportion, then random allocation to the nap (experimental) and awake (control) groups: a between-subjects design, which avoids practice effects from learning a second list.
Step 4: link design to validity
Random allocation spreads individual differences such as memory ability across both groups, so any difference in recall is more likely to be caused by the nap.
Marker's note: design questions reward specific, operationalised variables ("number of words correctly recalled"), not vague ones ("memory").
- Confusing random sampling with random allocation
- Sampling is about who takes part; allocation is about which group they join.
- Calling the control group a controlled variable
- A control group is a group of participants who do not receive the experimental treatment; a controlled variable is a condition held constant.
- Writing a hypothesis without both IV levels or a direction
- "Music affects memory" is not a VCE hypothesis.
- Claiming a correlational study shows cause
- Only a controlled experiment can support a causal conclusion.
- Naming a design's feature instead of its benefit
- In 2025, "it has an experimental and control group" did not score; explain why the design suits that experiment.
Exam-style questions
Questions in the style of VCAA exam questions on this dot point, each with a worked answer. They are written by ExamExplained unless tagged "Past paper"; the year shows the paper a question is modelled on.
2025 VCAA-style3 marksFor a Psychology practical investigation, Arlo decided to examine whether an awareness of Lazarus and Folkman's Transactional Model of Stress and Coping had an impact on the severity of stress appraisals. Arlo used a sample of 10 Psychology students and 10 non-Psychology students in his experiment. The 10 Psychology students had spent one week learning about the Transactional Model of Stress and Coping. Each participant was presented with a written scenario about a social conflict and 10 questions that gathered data on how severely they would appraise this potential stressor. Write a possible hypothesis for Arlo's experiment.Show worked answer →
3 marks. The report says full marks needed both levels of the independent variable, a dependent variable referring to the severity of stress appraisals, and a clear direction (either direction was accepted). Lower-scoring answers were not phrased as a testable prediction.
Model answer (in the style of the report's high-scoring example): It is hypothesised that students who are aware of Lazarus and Folkman's Transactional Model of Stress and Coping will appraise the social-conflict scenario as less severely stressful than students who have not learnt about the model.
Check your own hypothesis against the three parts: IV level 1 (aware of the model) and level 2 (not aware), DV (severity of stress appraisal), direction (less severe).
Source: VCAA 2025 VCE Psychology examination, Section B, Question 7a (stimulus abridged), and the 2025 examination report.
2025 VCAA-style4 marksArlo's experiment (above) used a between-subjects design. (i) Outline one benefit of using a between-subjects design for this experiment. (1 mark) (ii) Summarise how a within-subjects design could be used in a new experiment with the same aim. (3 marks)Show worked answer →
(i) 1 mark. Only 19 per cent of students scored this mark. The benefit had to be outlined for this experiment, not just named. Accepted benefits included: time efficient because both groups can be tested at the same time; a lower rate of withdrawal because participants complete only one condition; no effect of prior participation because each participant is in only one group. Simply stating a feature ("it has an experimental and a control group") did not score.
(ii) 3 marks, one for each of: the same participants complete both conditions; the control condition must come first (participants complete the stress appraisal questions before learning the model, because learning cannot be undone); an application to this context, such as teaching the participants the model between the two sessions or using a different but equivalent scenario in each condition.
Source: VCAA 2025 VCE Psychology examination, Section B, Question 7d, and the 2025 examination report.
Practice questions
Original practice questions graded from foundation to exam level, each with a full worked solution. Try them before revealing the solution.
foundation3 marksA researcher tests whether listening to music while studying affects recall. Forty Year 12 students learn 30 words either in silence or with music, then write down as many as they can remember. Identify the independent variable, the dependent variable and one variable that should be controlled.Show worked solution →
1 mark each.
- IV: whether students study with music or in silence (the variable the researcher manipulates, with two levels).
- DV: the number of words correctly recalled (what is measured).
- Controlled variable (any one): the word list, the time allowed to learn, the time of day, the room, or the delay before recall. It is held constant so it cannot become a confounding variable.
core3 marksA study compares sleep quality in adolescents who use a phone in bed with those who do not. Explain the difference between random sampling and stratified sampling, and explain when stratified sampling would be the better choice for this study.Show worked solution →
1 mark random sampling: every member of the population has an equal chance of selection; 1 mark stratified sampling: the population is divided into subgroups (strata) on a characteristic relevant to the study, and participants are selected from each stratum in the same proportions as in the population; 1 mark application: stratified sampling is better when a characteristic likely to affect sleep (such as age band, gender or school year) must be represented in its true proportions, for example to make sure each year level is represented as it is in the population.
core3 marksExplain the difference between an extraneous variable and a confounding variable, using the music-and-recall study as an example.Show worked solution →
1 mark extraneous variable defined (any variable other than the IV that may affect the DV and should be controlled or monitored); 1 mark confounding variable defined (a variable other than the IV that has affected the DV, providing an alternative explanation for the results); 1 mark a correct example linked to the study.
Sample answer: Prior vocabulary knowledge is an extraneous variable: it might affect recall, so it should be controlled, for example by random allocation. If, by chance, the silence group had much stronger readers and recalled more words, prior knowledge would have become a confounding variable, because we could no longer tell whether the difference in recall was caused by the music or by reading ability.
exam4 marksA researcher wants to test whether a 10-minute mindfulness session reduces self-reported anxiety before a test. Propose a mixed design for this investigation and explain one advantage it has over a between-subjects design.Show worked solution →
Marks: 1 random allocation of participants to an experimental group (mindfulness) and a control group (for example, 10 minutes of quiet reading); 1 both groups measured before and after the session (the within-subjects part); 1 the DV clearly identified (self-reported anxiety on the same rating scale); 1 a valid advantage, such as allowing each participant's change from their own baseline to be compared between groups, which reduces the influence of individual differences in baseline anxiety.
exam3 marksA student plans to investigate whether caffeine affects reaction time. She will measure the reaction time of 12 friends twice: first after no caffeine, then after a caffeinated drink, on the same afternoon. Evaluate this method and suggest one improvement.Show worked solution →
1 mark a strength (within-subjects design controls participant-related variables, such as natural differences in reaction speed); 1 mark a limitation (order effects: practice on the first trial may improve the second, so improvement may not be due to caffeine; also a small convenience sample of friends reduces representativeness and external validity); 1 mark a matching improvement (counterbalancing, with half the participants doing each condition first on separate days, or a larger random sample from the target population).