3.9 Quasi-experimental designs | Quantitative methods | Research Designs | UvA

Summary

Quasi-experimental designs resemble true experimental designs but lack random assignment. They are often used when randomization is impractical or unethical, such as when studying the impact of violent video games on children’s aggression. The script explores various quasi-experimental designs including static group comparison, pre-test post-test non-equivalent control group, and interrupted time series. These designs allow researchers to examine causal relationships while addressing threats to internal validity, like pre-existing differences between groups. The design choice depends on the research context, aiming to improve causal inference despite the lack of random assignment.

Takeaways

• 🧪 Quasi-experimental designs resemble true experiments but lack random assignment, and sometimes manipulation or comparison.
• ⚖️ These designs are often used when random assignment is impractical or unethical, such as parental choice in exposing children to violent games.
• 🌱 Independent variables in quasi-experiments may be naturally occurring characteristics (e.g., gender, traumatic experiences) rather than manipulated factors.
• 🔗 Unlike correlational studies, quasi-experiments still aim to test causal hypotheses and control extraneous variables as much as possible.
• 👥 The static-group comparison design compares post-test scores between non-randomly assigned groups, but is vulnerable to selection threats.
• 🔍 Adding a pre-test creates a non-equivalent control group design, helping researchers detect preexisting group differences and clarify causal order.
• ⏳ When no control group is available, a one-group pre-test/post-test design can at least establish temporal precedence, though many validity threats remain.
• 📉 Interrupted time-series designs improve validity by collecting multiple observations before and after treatment to distinguish natural trends from treatment effects.
• 🧭 Sudden changes after treatment in an interrupted time series can indicate an effect, but researchers must still rule out history threats.
• 🔁 Replicated interrupted time-series designs add a comparison group or shift treatment timing to strengthen causal inference and reduce alternative explanations.

Q & A

What is a quasi-experimental design?

A quasi-experimental design is similar to an experimental design but lacks key features like random assignment. It is used to demonstrate causal relationships, often in situations where random assignment is not possible due to practical or ethical reasons.

Why can't random assignment always be used in quasi-experimental designs?

Random assignment may not be possible due to practical constraints or ethical considerations. For example, in studies involving children, parents may need to decide whether their child is exposed to a particular treatment, which prevents random assignment.

How does a quasi-experiment differ from a true experiment?

In a true experiment, participants are randomly assigned to different conditions, and the independent variable is manipulated. A quasi-experiment, on the other hand, may lack random assignment and sometimes the manipulation of the independent variable, making it less controlled.

What is an example of a natural experiment in a quasi-experimental design?

An example of a natural experiment is studying the effects of gender on aggression. Since we cannot assign children to be boys or girls for a study, gender becomes an independent variable that cannot be manipulated, making it a natural experiment.

What is a static group comparison design?

A static group comparison design involves comparing two groups after exposure to the independent variable, but without random assignment. This design is vulnerable to selection threats to internal validity, which can lead to ambiguity in causal direction.

How does the pre-test post-test non-equivalent control group design improve upon the static group comparison design?

The pre-test post-test non-equivalent control group design includes a pre-test to assess any preexisting differences between groups, which helps establish causal direction and control for maturation and history effects.

What limitations exist when using a one-group pre-test post-test design?

The one-group pre-test post-test design can only establish temporal precedence, meaning it shows that the cause occurred before the effect. It still faces threats to internal validity, such as maturation and history effects, since there is no control group.

How does an interrupted time series design improve the one-group pre-test post-test design?

An interrupted time series design adds multiple measurements of the dependent variable before and after exposure to the independent variable. This allows researchers to assess natural changes in the dependent variable, helping to rule out maturation and history effects.

What is a replicated interrupted time series design, and why is it useful?

A replicated interrupted time series design adds a second group of participants who are measured at the same time points but do not receive the treatment. This helps determine whether the observed changes are due to the treatment or if they are caused by a history effect.

What does it indicate if the second group in a replicated interrupted time series shows the same pattern of results?

If the second group shows the same pattern of results, it suggests that the observed changes are likely due to a history effect rather than the independent variable. This strengthens the conclusion that the intervention may not be the cause of the change.