Why Do We Care About Research Design?
Understanding research design is essential in psychology because it determines how confidently we can interpret findings.
Not all research designs allow us to confidently infer causality, and some designs create stronger external or internal validity than others.
Correlation ≠ Causation
Just because two variables are related does not mean one causes the other.
The 3 Major Types of Research Designs
Non-Experimental Design
Features
- No variables are controlled or manipulated.
- Variables are observed and measured as they naturally occur.
- Groups are not randomly assigned, because they exist in their natural states.
Validity
- High external validity (real-world relevance).
- Lower internal validity (harder to establish causality).
Examples
- Descriptive research.
- Correlational research.
- Surveys measuring personality traits.
- Observing screen time and sleep patterns.
When do we use it?
- Studying behaviour in its natural state.
- If you can’t ethically manipulate variables.
- If random assignment isn’t possible.
- To examine possible correlations, not causation.
Experimental Design
Features
- At least one variable is manipulated.
- An independent variable is changed to measure the effect on a dependent variable.
- Groups are randomly assigned.
- Control groups are used.
- There is a high level of control over extraneous variables.
Validity
- Higher internal validity (greater variable isolation makes it easier to infer causality).
- Potentially lower external validity (lab conditions may not accurately represent real life).
Examples
- Medicine or drug trials.
- Testing the effectiveness of therapeutic methods.
- Sleep deprivation experiments.
When do we use it?
- When you are trying to establish causality.
- To isolate variables.
- When designing studies to be easily replicable and controlled.
Quasi-Experimental Design
Features
- An independent variable is manipulated.
- Groups are not randomly assigned and instead use naturally occurring or pre-existing groups.
- Control or comparison groups are used.
Validity
- Higher external validity than lab experiments.
- Lower internal validity than true experiments.
Examples
- Testing a new training method on one group of employees and not on another.
- Comparing crime rates before and after a new law in one state versus a neighbouring state.
When do we use it?
- When it might be difficult or unethical to randomise groups.
- When studying real-world policies or natural events.
Cross-Sectional vs Longitudinal Designs
These describe how data is collected across time, not whether a study is experimental.
| Cross-Sectional | Longitudinal |
|---|---|
| Data is measured at one point in time. | Data is measured at multiple points across a time period. |
| Data is taken from different samples in a single snapshot. | Data is taken from the same original sample across time. |
| Example: handing out a survey to consumers after distributing samples of a new pancake flavour. | Example: following the same group of children for 3 years to measure the effects of daily pancake consumption. |
Between-Subjects vs Within-Subjects Designs
These describe how participants experience conditions.
| Between-Subjects | Within-Subjects |
|---|---|
| Each participant experiences only one condition. | All participants experience all conditions. |
|
Pros: No order effects. Lower participant fatigue. Cons: Requires more participants. Individual differences may influence results. |
Pros: Fewer participants needed. Controls for individual differences. Cons: Order effects. Practice or fatigue effects. |
Quick Summary
| Design Type | Manipulation | Random Assignment | Can Infer Causality? |
|---|---|---|---|
| Non-Experimental | No | No | No |
| True Experimental | Yes | Yes | Yes |
| Quasi-Experimental | Yes | No | Limited |
Read Next
If you’re revising research methods and stats, these posts go well with this one:
