What is experimental design?

Experimental design is the process of deciding how to allocate participants to different conditions in a psychological experiment. This allocation helps researchers test the effect of an independent variable (IV) – the factor being manipulated – on a dependent variable (DV) – the factor being measured.

In a typical experiment, participants are divided into groups. One common approach is to have an experimental group, which experiences the change or treatment, and a control group, which does not.

The key decision in experimental design is how participants from the sample are assigned to conditions. For example, if there are fifteen participants, the researcher must choose whether all will experience every condition or if they will be split into separate groups for different conditions. This choice leads to different types of designs, each with its own strengths and limitations.

Types of experimental designs

There are three main types of experimental designs used in psychology: independent measures, repeated measures, and matched pairs. Each type determines how participants are grouped and exposed to the IV, affecting the experiment's validity and efficiency.

Independent measures design

In an independent measures design, different groups of participants are used for each condition of the experiment. Each participant experiences only one level of the IV.

Key features of independent measures design:

• How it works - The sample is divided into separate groups. For instance, to test if gender affects reading ability, one group of 25 boys might take a reading test, while a separate group of 25 girls takes the same test. The results are then compared.
• Advantages - This design avoids order effects, as participants are not exposed to multiple conditions.
• Disadvantages - Individual differences between groups, such as varying intelligence or motivation, can affect results. It often requires more participants, making it resource-intensive.

Repeated measures design

A repeated measures design uses the same participants in all conditions of the experiment. Each person is tested multiple times, once for each level of the IV.

Key features of repeated measures design:

• How it works - The same group experiences every condition. For example, twelve participants might complete two different IQ tests, and their scores on each are compared.
• Advantages - It controls for individual differences, as the same people are used throughout. This often requires fewer participants and can provide more sensitive measurements of the IV's effect.
• Disadvantages - Order effects can occur, where performance changes due to the sequence of conditions rather than the IV. This can skew results and reduce validity.

Matched pairs design

Matched pairs design uses different participants for each condition but matches them on key variables to minimise differences.

Key features of matched pairs design:

• How it works - Participants are paired based on relevant characteristics, such as age, gender, intelligence, or personality scores. One member of each pair is assigned to the experimental condition, and the other to the control. For example, pairs might be matched for age and IQ, then one tests a new learning method while the other uses a standard one.
• Advantages - It reduces the impact of individual differences, similar to repeated measures, while avoiding order effects since different people are used.
• Disadvantages - Matching can be time-consuming and may not account for all variables. It still requires more participants than repeated measures.

Order effects in repeated measures designs

Order effects are unwanted influences on participants' performance that arise from the sequence in which conditions are presented in a repeated measures design. These effects occur because the same participants experience multiple conditions, which can lead to changes unrelated to the IV.

Types of order effects

• Practice effect - Participants may improve on later tasks due to familiarity or learning from earlier ones.
• Fatigue effect - Performance may decline in later conditions due to tiredness or boredom.

These effects can confound results, making it seem like the IV caused changes when the order was actually responsible. As a result, repeated measures designs must include controls to minimise these issues.

Controlling order effects with counterbalancing

Counterbalancing is a technique used to manage order effects in repeated measures designs by varying the sequence of conditions across participants. This ensures that any effects from the order are evenly distributed and do not systematically bias the results.

How counterbalancing works

1. Divide the participants into subgroups.
2. Assign different orders of conditions to each subgroup. For example, with two conditions (A and B), half the participants do A first then B, while the other half do B first then A.
3. Compare results across subgroups to balance out practice or fatigue effects.

This method reduces the impact of order effects, improving the experiment's internal validity – the extent to which the IV truly causes the observed changes in the DV.

Using randomisation to reduce bias

Randomisation is a key method for minimising bias in experimental designs. Bias occurs when factors other than the IV influence the results, such as uneven group allocation.

The purpose and process of randomisation

• Reducing bias - Randomisation ensures every participant has an equal chance of being assigned to any group or condition. This helps create comparable groups and eliminates systematic differences.
• How it is applied - In independent measures, participants might be randomly allocated using a computer or drawing lots. In repeated measures with counterbalancing, the order of conditions can also be randomised for each participant.
• Benefits - It controls for confounding variables – factors that could affect the DV independently of the IV – making the experiment fairer and more reliable.

By incorporating randomisation, researchers enhance the experiment's internal validity, meaning that any observed changes in the DV can be more confidently attributed to the IV rather than confounding variables.