4.2.3.1 - Experimental designs

What an experimental design decides

An experiment compares conditions of an independent variable to see whether they affect a dependent variable. The experimental design decides how participants are organised across those conditions. This matters because the design can make a result clearer, or it can introduce alternative explanations for the result.

Imagine a researcher testing whether listening to music affects memory for word lists. The independent variable is the listening condition, such as music or silence. The dependent variable might be the number of words recalled. The researcher must decide whether the same people do both conditions, different people do each condition, or carefully matched people are split between conditions.

There are three AQA designs in this lesson.

In repeated measures, every participant provides a score in every condition.

In independent groups, each participant provides a score in one condition only.

In matched pairs, the participants are different people, but each condition receives one person from each carefully matched pair.

Two design problems come up again and again in exam scenarios.

Participant variables are a threat when different people are compared. If one condition happens to contain more high-ability participants, the result may be due to the people rather than the independent variable.

Order effects are a threat when the same people complete more than one condition. The second score may be affected by doing the first task, not by the independent variable itself.

AO2 scenario: Finn studies whether background noise reduces proofreading accuracy. If the same participants proofread in silence and then with noise, repeated measures controls their individual reading ability but risks practice or fatigue. If different participants are used for silence and noise, independent groups avoids order effects but risks one group being better at proofreading before the experiment begins.

Experimental type and experimental design answer different questions. “Laboratory” describes the testing situation and manipulation; “repeated measures” describes who supplies the scores. A laboratory experiment could use any of the three designs. A comparison of existing age groups might use different people in each group, but that arrangement does not make age a manipulated IV or establish a causal effect of age.

Repeated measures

In a repeated measures design, every participant produces a score in each condition. In a memory experiment, all participants might recall one word list while listening to music and another comparable word list in silence. The researcher compares each person's music score with their silence score.

The main strength is control over participant variables. Because the same participants appear in every condition, stable differences in memory ability or personality cannot explain a difference between separate groups, because the same people are compared. Temporary states such as motivation or tiredness can still change between conditions. This can make the design sensitive: smaller differences between conditions are easier to detect because the researcher is not comparing two unrelated sets of people.

Repeated measures can also be efficient. Fewer participants are needed because each person contributes data to every condition. This is useful when participants are hard to recruit, such as patients with a rare diagnosis or students available for a short practical session.

The main limitation is order effects. Participants may improve because of practice, perform worse because of fatigue, or work out the aim because they have seen more than one version of the task. Counterbalancing distributes order effects across conditions: for 20 participants, randomly assign 10 to music then silence and 10 to silence then music (AB/BA). Use equivalent word lists and balance which list appears with each condition. This reduces a systematic order confound; it does not erase fatigue or solve irreversible learning. A washout interval may be needed if an effect, such as caffeine, persists into the next condition.

Repeated measures can increase demand characteristics because participants experience several conditions and may spot what is being compared. For example, if they realise a study is testing whether noise harms concentration, they may try extra hard in the noise condition or give the response they think the researcher expects.

AO3 judgement: repeated measures is strongest when the task can be repeated without the first condition changing how participants respond in the second. It is weaker when exposure to one condition makes the next condition impossible to treat as fresh, such as seeing an ambiguous image or learning the answer to a puzzle.

Independent groups

In an independent groups design, each participant takes part in one condition only. In a memory study, one group might learn words with music, while a separate group learns words in silence. The researcher compares the average scores of the two groups.

The main strength is that order effects are avoided. Practice, tiredness or boredom from an earlier experimental condition cannot carry over, because each person completes only one condition. Participants can still get tired during their one task, so procedure length still matters. This is useful when the task would be spoiled by repetition. If a participant has already seen the answer to a problem-solving task, they cannot meaningfully complete a second condition using the same task.

Independent groups can also reduce demand characteristics. Participants only see one version of the study, so it may be harder for them to infer the aim. Data collection may also be practical if both groups can be tested at the same time.