Extraneous variable: two named types
An extraneous variable is anything other than the IV that could affect the DV; uncontrolled, it becomes an alternative explanation, so the study loses validity. The specification names two types. Situational variables belong to the environment — noise, heat, lighting, time of day, equipment. Participant variables belong to the person — age, mood, intelligence, prior experience. The test: room or person? Being in a rush belongs to the person, so it is participant.
Five controls, each for a different threat
Statement 11.1.2 names five controls for different threats. Standardised procedures — identical instructions, materials, timings, setting. Counterbalancing — in repeated measures, half do A first, half B first, spreading order effects. Randomisation — chance sets order or allocation. Single-blind — the participant is unaware of their condition, controlling demand characteristics. Double-blind — neither participant nor experimenter knows, also controlling researcher bias.
A control must be concrete and specific
Examiners reward concrete controls, not the label. Standardised procedures earns little; what was held the same earns the mark — the same word list in the same order, the same recorded instructions, the same study and recall times. The same goes for the rest: for counterbalancing say which half did which first; for randomisation say what was randomised; for a blind say who was kept unaware of what. A control must also fit this study and lie within the researcher's control.
Drawn from real examiner reports.
Situational and participant are muddled
The definition trap runs both ways: asked for a situational variable, candidates give something about the person; asked for a participant one, a room feature. Situational means the environment; being in a rush fails that test — it depends on the person — so it is participant. Situational: noise, lighting, time of day. Participant: age, mood, tiredness.
June 2023 Paper 2 Q2c rules that being in a rush is not a situational variable because it depends on the person, not the environment. The reverse error is reported in November 2020 Paper 2 Q04d.
Standardisation is not controls in general
Asked what standardisation means, candidates drift onto controls, sampling or ethics — none of which answers it. It means keeping every part of how the study is run the same for everyone: same instructions, materials, timings, setting. Credited answers were concrete: the same trigrams, viewing time and recall time. It does not control participant variables.
June 2022 Paper 2 Q1b reports that standardisation was answered instead on controls, participant variables, sampling or ethics, with credited answers concrete: the same trigrams, the same viewing time and the same recall time.
A blind technique must be applied here
Candidates can define a single-blind technique — the participant does not know their condition — but fail to say what it means here and why it matters. A full answer names it, shows it in the study (identical drinks, untold which has caffeine), and says why: they cannot guess what is expected, so a difference more likely reflects the drink. A bare definition scores zero.
November 2020 Paper 2 Q03c records that a single-blind technique was understood but not applied to the study in the question.
Improvements must fit this study
A control or improvement must be chosen for this scenario, not pre-learned and dropped in — examiners warn against pre-learned improvements that do not fit. It must also lie within the researcher's control: you cannot control the weather or a participant's personality, only how they are allocated and what is held constant.
June 2024 Paper 1 Summary warns against pre-learned improvements that do not fit the scenario. That improvements must be within the researcher's control is set out in June 2019 Paper 2 Q3c.
A bigger sample is not an automatic fix
A common improvement offered is simply more participants. Examiners are clear a bigger sample is not automatically an improvement: it is the nature of the sample, not its size, that decides how far findings generalise. Adding more of a biased sample does not fix the bias — suggest a more representative sample instead.
That a bigger sample is not automatically an improvement, because it is the nature of the sample that matters, is set out in June 2019 Paper 2 Q3c and June 2022 Paper 1 Q5b.
The IV is not an extraneous variable
Do not offer the IV as the extraneous variable. The IV is what the researcher changes on purpose, stated in full with both conditions — naming the topic alone (telephone numbers) is incomplete. An extraneous variable is a different, uncontrolled variable that could also affect the DV. Confusing the two wastes the identification the answer depends on.
November 2021 Paper 2 Q05a states that an IV must be given in full with both conditions, and June 2019 Paper 2 Q2a records the IV and DV being swapped.
Counterbalancing is not two groups
Counterbalancing is often called putting half the participants in each condition — but that is independent measures. In counterbalancing every participant still does both conditions; only the order is split (half A then B, half B then A). It applies only where there is an order effect, and does not remove it — it spreads it equally so it cannot explain a difference.
More control is not always more valid
Control is a strength, but two beliefs are wrong. First, tighter control can reduce validity: a heavily standardised lab task may not resemble everyday behaviour, trading realism for certainty. Second, control is not correctness: a standardised, repeatable study can still measure the wrong thing. Reliability is consistency; validity is measuring what it claims.
Name it, apply it, say what it removes
Almost every mark here is won in one sentence. Name the control. Apply it using the scenario detail — a name or a reworded question is not application. Justify it — name the extraneous variable it removes and say why that makes the result more trustworthy.
Do the AO2 before any AO3
AO3 cannot be awarded without AO2 present, so never write a control in the abstract. Anchor it in the study — this task, these participants, this setting — before judging how far it strengthens the result. No scenario link means zero for evaluation too.
Match the control to the design
Match the control to the design. Counterbalancing exists only where each participant does both conditions (repeated measures); suggesting it for independent measures wastes the mark. For independent measures, randomise allocation so participant differences spread evenly.
Extraneous variable — any variable other than the independent variable (IV) that could affect the dependent variable (DV).
Situational variable — an extraneous variable that is a feature of the environment or situation: noise, lighting, temperature, time of day, location, equipment.
Participant variable — an extraneous variable that is a feature of the person taking part: age, mood, intelligence, prior experience, tiredness, being in a hurry.
Full notes, flashcards, Q&A and the topic quiz for every premium subject.
Premium plans are US$8.99/month or US$49.99/year — first month free.
Studying with a parent's blessing? Show them this.
