Experimental designs: independent groups, repeated measures and matched pairs. Design of investigations, including control of variables, randomisation and counterbalancing.

What this dot point is asking

AQA wants you to describe the three experimental designs, their strengths and limitations, and how to control problems using randomisation and counterbalancing. The exam skill is to match each design to its characteristic problem and the technique that addresses it.

The three designs

The three designs differ in how participants are allocated to conditions, and each has a characteristic strength and weakness. In an independent groups design, different people take part in each condition, so there are no order effects (each person does only one condition), but the design is vulnerable to participant variables, because differences between the groups of people (rather than the independent variable) might cause any difference in results; it also requires more participants overall. In a repeated measures design, the same people take part in every condition, which controls participant variables perfectly (each person is compared with themselves) and needs fewer participants, but it introduces order effects (practice, fatigue or boredom) and a greater risk of demand characteristics, since participants experience both conditions and may guess the aim. The matched pairs design is a compromise: different participants are used (avoiding order effects) but they are matched in pairs on variables relevant to the study (such as age and IQ), with one member of each pair in each condition, which reduces participant variables. Its drawback is that matching is time-consuming and never perfect, since participants cannot be matched on every relevant variable.

Controlling problems

Knowing which technique solves which problem is exactly what AQA tests. Order effects in a repeated measures design are tackled by counterbalancing: the order of conditions is varied across participants so that practice and fatigue effects are spread evenly and cancel out, for example by having half the participants do condition A then B and the other half do B then A (or by using an ABBA order within each participant). Participant variables in an independent groups design are tackled by random allocation: assigning participants to conditions by chance (for example by drawing names) so that individual differences are distributed evenly across the groups. Two broader controls apply to any design: randomisation, the use of chance methods to decide the order of materials or trials so the researcher's bias cannot influence them, and standardisation, keeping the procedure and instructions identical for every participant so the study is replicable and confounding is reduced.