Science for Everybody

Contents

• Introduction
• What is experimental design?
• Independent groups design
• Repeated measures design
• Matched pairs design
• Choosing between the three designs
• Evaluation
• Strengths
• Weaknesses
• Summary
• Quick Questions
• Key Terms

---

Introduction

Experimental design describes how participants are allocated to the different conditions of an experiment. The choice between independent groups, repeated measures and matched pairs affects how well researchers control participant differences, order effects and practical demands.

A good design helps psychologists make a fair comparison between conditions so that changes in the dependent variable are more likely to reflect the independent variable.

What is experimental design?

In an experiment, researchers manipulate an independent variable (IV) and measure its effect on a dependent variable (DV). Experimental design refers specifically to how participants are allocated to the conditions created by the IV. It is different from the type of experiment, such as a laboratory, field, natural or quasi-experiment.

Imagine a researcher wants to test whether an energy drink changes reaction time compared with water. The IV is the drink condition and the DV could be reaction time measured in milliseconds. The researcher could organise the participants in three main ways.

Independent Groups Design

In an independent groups design, separate groups of participants take part in separate conditions. Each participant experiences only one condition of the IV.

For the energy-drink example, Group 1 could drink the energy drink and Group 2 could drink water. The researcher would then compare the average reaction time of the two groups.

Because different people are used in each condition, independent groups designs avoid order effects. Participants cannot become practised, tired or bored because they do not complete the same task under both conditions. They may also be less likely to work out the aim of the study.

However, participant variables can become a confounding influence. If one group happens to contain more naturally alert or faster-reacting participants, differences in reaction time may be caused by the people in each group rather than the drink.

Researchers therefore often use random allocation. Each participant should have the same chance of entering either condition. Random allocation does not guarantee identical groups, but it reduces systematic allocation bias.

Repeated Measures Design

In a repeated measures design, the same participants take part in every condition of the experiment. Each person therefore acts as their own comparison.

In the energy-drink study, every participant could complete the reaction-time task after the energy drink and after water. The researcher would compare each person’s performance across the two conditions.

This design controls participant variables because the same people appear in both conditions. It can also be more economical because fewer participants are needed.

The main problem is order effects. Performance in the second condition may improve because participants have practised the task, or it may become worse because they are tired, bored or less motivated. Participants also experience all conditions, so demand characteristics may be stronger because the aim can be easier to guess.

Counterbalancing helps control order effects. Half the participants complete Condition A then Condition B, while the other half complete B then A. This is often described as an AB/BA design. Counterbalancing spreads order effects more evenly across conditions, although it does not remove every possible carry-over effect.

Matched Pairs Design

Matched pairs is a compromise between the other two designs. Different participants take part in each condition, but researchers match participants into pairs on variables that are likely to affect the dependent variable.

For a memory study, researchers might first measure participants’ memory ability and then pair together people with similar scores. One person from each pair would enter Condition A and the other would enter Condition B.

Because different people complete each condition, matched pairs designs avoid order effects. Matching can also reduce participant variables if the researcher identifies and measures the most relevant characteristics.

However, it is difficult to create truly identical pairs. People differ in many ways, and the researcher may match participants on one variable while overlooking another important difference. Pre-testing and matching also take time and may be expensive. If one member of a pair withdraws, the matched comparison can be weakened.

Choosing Between the Three Designs

The best design depends on the research aim, the task and the practical context. If practice or fatigue would seriously distort performance, independent groups or matched pairs may be preferable. If individual differences are likely to have a large effect and the task can be repeated safely, repeated measures may provide stronger control.

Matched pairs can be useful when both participant variables and order effects are major concerns, but only when researchers can identify suitable matching variables and have enough time to create the pairs.

Reflection question: Suppose a psychologist wants to investigate whether background music affects recall of a 20-word list. Which experimental design would you choose, and what would be the most important source of bias or error to control in that design?

Quick Comparison

• Independent groups: different participants in each condition; no order effects, but participant variables may differ between groups.
• Repeated measures: the same participants in all conditions; participant variables are controlled, but order effects and demand characteristics may occur.
• Matched pairs: different participants in each condition are matched on relevant variables; order effects are avoided and participant differences are reduced, but matching is difficult and time-consuming.
• Random allocation is especially useful in independent groups designs.
• Counterbalancing is especially useful in repeated measures designs.

---

Evaluation

Strengths

Repeated Measures Can Improve Internal Validity

A major strength of repeated measures is that participant variables are controlled because the same people take part in all conditions. This makes differences in the DV less likely to be caused by stable individual differences such as intelligence, memory ability or motivation. Therefore, if order effects are controlled effectively, repeated measures can provide a more internally valid test of the IV.

Independent Groups Avoid Order Effects

Independent groups designs avoid practice, fatigue and carry-over effects because each person completes only one condition. This is especially useful when the first condition would permanently change performance in the second. As a result, independent groups can produce a cleaner comparison when repeating the task would itself change behaviour.

Matched Pairs Can Balance Control and Independence

Matched pairs designs combine an important advantage of independent groups with some control over participant variables. Different participants complete each condition, so there are no order effects, but matching can make the groups more comparable on characteristics linked to the DV. Therefore, matched pairs can be a strong choice when order effects would be serious and individual differences are also likely to matter.

Control Procedures Can Strengthen Design Quality

Random allocation and counterbalancing make experimental designs more systematic. Random allocation reduces researcher bias when assigning people to independent groups, while counterbalancing distributes order effects in repeated measures. These procedures also make the method easier to describe and replicate, which supports reliability.

Weaknesses

Independent Groups Remain Vulnerable to Participant Variables

A key weakness of independent groups is that the conditions contain different people. Even with random allocation, one group may differ by chance in characteristics that influence the DV. This can reduce internal validity because an apparent effect of the IV may actually reflect participant variables. Larger samples can reduce this problem, but they increase the time and resources needed.

Repeated Measures Can Create Order Effects and Demand Characteristics

Repeated measures exposes each participant to every condition. Practice may improve later performance, while fatigue or boredom may reduce it. Participants may also compare conditions and infer the research aim, increasing demand characteristics. Counterbalancing reduces systematic order effects, but it may not remove carry-over effects completely.

Matched Pairs Can Never Produce Perfectly Matched Participants

Matching is based on selected characteristics, so researchers must decide which variables matter most. Two people matched for age and memory score may still differ in sleep, motivation, stress or prior experience. This means participant variables can remain. In addition, pre-testing and pairing take time, and losing one participant can reduce the usefulness of the matched partner’s data.

Experimental Design Does Not Solve Every Validity or Ethics Problem

The choice of design cannot by itself remove sample bias, cultural bias or low temporal validity. A perfectly counterbalanced repeated measures study can still use an unrepresentative sample or a culturally narrow task. Ethical issues also depend on the procedure and participant burden. Therefore, researchers must evaluate experimental design alongside sampling, ethics, operationalisation and the wider context of the study.

Quick Questions

What does experimental design describe?

It describes how participants are allocated to the different conditions of an experiment.

How are participants organised in an independent groups design?

Different participants take part in each condition, so each participant experiences only one level of the IV.

What are order effects?

Changes in performance caused by the order of conditions, such as practice, fatigue or boredom.

How does counterbalancing usually work in a two-condition repeated measures experiment?

Half the participants complete A then B and the other half complete B then A.

Why can matched pairs reduce participant variables?

Participants are paired on relevant characteristics, making the groups more similar on variables that could influence the DV.

Key Terms

Experimental design – The way participants are organised or allocated across the conditions of an experiment.

Independent groups design – Different participants take part in each condition of the experiment.

Repeated measures design – The same participants take part in every condition of the experiment.

Matched pairs design – Different participants take part in each condition but are matched on important characteristics.

Participant variables – Individual differences between participants that may influence performance.

Order effects – Changes in performance caused by the order in which conditions are completed.

Random allocation – Assigning participants so each has an equal chance of entering any condition.

Counterbalancing – Varying the order of conditions to control order effects, commonly using AB/BA.

Demand characteristics – Cues that allow participants to guess the aim of a study and change their behaviour.

Matching variable – A characteristic used to pair participants because it may influence the dependent variable.