What is the best experimental design in psychology? When it comes to experimental designs, one size doesn't fit all, and choosing the right one for your research is crucial. A design that works great in one context won't necessarily be the appropriate choice for a different study. In this article, we'll go through experimental designs in psychology, consider the strengths and weaknesses of each one and consider what contexts they could be applied in.
Fig. 1 - It's important to consider what experimental design best fits your research question.
Experimental Design Definition
To investigate the effect of our independent variable on the dependent variable, we need to expose participants to experimental conditions. The independent variable is then manipulated between these conditions.
Let's say you want to measure sleep's impact on a student's academic performance. Your independent variable is the amount of sleep that students get. You need to expose your participants to at least two different levels of sleep to investigate its impact on their academic performance.
You decide you want participants to get only three hours (sleep-deprived experimental condition) of sleep in one condition and 8 in the second (well-rested experimental condition).
The question now should the same group be subjected to both levels of your independent variable? Or should you create two groups where each is assigned to only one condition? These are questions related to choosing your experimental design.
Experimental design refers to the ways participants are assigned to the different conditions of an experiment. Your experimental design can involve subjecting the same group of participants to all levels of your independent variable or just one level.
Types of Experimental Design
There are four types of experimental design: independent measures, repeated measures, matched pairs, and quasi-experimental designs.
Various factors determine the type of experimental design used, like:
- The research method used.
- Whether the researcher can manipulate the variables studied.
- Whether participants can be randomly allocated into the experimental (or control) groups.
- Characteristics of participants.
Let's now go through each experimental research design used in psychology.
Experimental Research Design: Independent Measures
The independent measures design is an experimental design in which you only assign participants to one of your experimental conditions. This experimental design is also known as the between-subjects design.
So, if you have two conditions (e.g. one involving getting three hours of sleep before an exam and one that involves getting eight hours), you will need two groups. Each of the participants taking part in your experiment would only be assigned to one of the conditions.
However, if you're comparing two different groups of people, you need to account for the potential individual differences between the groups to ensure your findings are valid. Randomly assigning participants to groups is one way to average out any between-group differences.
It might skew your results if your eight-hour sleep group is much more academically advanced than your three-hour sleep group. To control for this, you can allocate participants randomly to each of the two conditions.
Random assignment means that each participant has a 50% chance of being assigned to either group. When you have two conditions, this could be done by flipping a coin when assigning each participant to the group or using a random number generator.
Fig. 2 - Hypothetical results from the experiment investigating the impact of sleep on exam performance.
Using the independent measures design, you can compare the two experimental groups in your experiment. This design also allows you to avoid order effects and demand characteristics. Order effects occur when the order in which participants participate in the condition affects their performance.
By only taking the exam once, participants don't have a chance to practice or get bored with the task, which could skew results if not considered.
Demand characteristics occur when participants guess the aim of your experiment and adjust their behaviour to what they think is expected.
Experimental Research Design: Repeated Measures
In a repeated measures design, participants take part and are assessed in each experimental condition. Therefore, the data for each IV condition come from the same participants. This experimental design is also referred to as a within-subjects design.
This approach eliminates the potential individual differences between groups, which is a confounding variable. This way, repeated measures design increases the study's validity.
Suppose the same group of students participate in our sleep and academic performance experiment conditions. In that case, we know our results won't be affected by differences in academic ability, intelligence, or motivation between the experimental groups.
It also makes recruitment easier, as we need half the number of participants that we would have needed for the independent measures design. However, this design introduces the risk of order effects as well as demand characteristics.
For example, participants may perform better in the second condition because they know the task already (practise effect) or may not perform as well due to fatigue (fatigue effect).
Moreover, if the participants know what conditions you're investigating, they can modify their behaviour to fit what they think you expect.
Here, if we made the same participant take the test after eight hours of sleep and then again after three hours, they might guess that we expect their performance to be lower with less sleep. Therefore, they can put less effort into the second condition.
To minimise order effects in repeated-measures experimental designs, we can counterbalance the order in which participants participate in the two conditions. Counterbalancing involves subjecting half of the participants to the first condition first and the other half to the second condition first. This way, it is possible to determine how order effects influenced the results.
Experimental Research Design: Matched-Pairs
Similarly to the independent-measures design, the matched-pairs design involves subjecting participants to only one experimental condition. However, in this design, the assignment process is more complex. Participants are first paired based on specific characteristics that could be potential confounding variables. Then each individual in the matched pair is randomly assigned to an experimental or control group.
In our experiment, we could first match the students we recruited based on their IQ and past academic performance. Let's say that Jess and Fiona performed similarly on both of these dimensions. We would flip a coin to decide which group Fiona would be assigned to and assign Jess to the other.
In some way, this design combines the best of both words – it allows us to minimise both the order effects and the individual differences. However, this design can be more complex, costly and time-consuming.
Finding two groups of participants that match all the key characteristics that might influence your results can be difficult.
The ideal participants for the matched-pairs design would be monozygotic twins, likely to be similar in many personality characteristics and share 100% of their genes.
Quasi-experimental Research Design
The quasi-experimental design is similar to the independent measures design, as it tests different participants at each independent variable level, except that participants are not randomly assigned to conditions. Instead, this design utilises naturally existing groups and investigates differences between them.
For example, to investigate the difference in prejudice between Irish and English. It would be impossible to manipulate someone's nationality, so we must use already existing groups.
Similarly, if we wanted to study the impact of economic status on prosocial behaviour, we would also utilise this design. It would be unethical to manipulate the participant's economic status.
The advantage of this design is that it has high external validity, as it involves testing participants in real-life settings. But it can introduce individual differences that could confound our results.
Since the independent variable is not manipulated in quasi-experimental designs, findings from studies using this design are correlational.
Fig. 3 - Some variables like demographics, genes or personality cannot be manipulated. When investigating these variables, we need to rely on quasi-experimental designs.
Experimental Design Examples
Independent measures research example:
- Investigating how the amount of sleep influences attention by recruiting 20 participants and randomly assigning half to each experimental condition (4 hours and 10 hours of sleep).
Repeated measures research example:
- Investigating differences in depression scores before and after cognitive behavioural therapy treatment.
Matched-pairs research example:
- Investigating the effect of exercise on weight loss, 50 participants were recruited. The participants were paired by BMI, age and gender, two males aged 21 with a BMI of 28, two females aged 22 with a BMI of 26, etc. These pairs were randomly assigned to the control group (no exercise) and the experimental group (2h of moderate-intensity exercise a week) to measure weight loss.
Quasi-experimental research example:
- Investigating the effectiveness of CBT for participants with depression or schizophrenia diagnosis. Here, the naturally occurring independent variable is the psychiatric diagnosis.
Evaluation of Experimental Design Psychology
Let's compare the occurrence of potential confounding variables (order effects, demand characteristics and individual differences) in each experimental design type.
| Experimental design | Order effects and demand characteristics | Individual differences |
| Individual measures | no | yes |
| Repeated measures | yes | no |
| Matched-pairs | no | no (controlled) |
| Quasi-experimental | no | yes |
Experimental Designs - Key takeaways
- Experimental design refers to how participants are assigned to the different conditions of an experiment.
- There are four types of experimental designs: independent measures, repeated measures, matched pairs, and quasi-experimental designs.
- Independent measures design involves assigning participants to only one of your experimental conditions. This can introduce confounding variables due to individual differences between groups, which can be controlled using random assignment.
- Repeated measures design involves subjecting the same group of participants to all independent variable levels. This can introduce order effects that are controlled by counterbalancing the conditions.
- Matched-pairs design involves pairing participants with specific characteristics that could be potential confounding variables. The quasi-experimental design utilises naturally existing groups and investigates differences between them.
Explanations 
Exams 
Magazine 
