Types of research design
A research design is the structure of your study — and the type you pick is dictated by two things: the kind of question you’re asking, and how much you can manipulate and control. The single biggest divide is whether you can establish causation (you intervene and randomize) or only association (you observe).
The main quantitative designs
| Design | What it does | Best for |
|---|---|---|
| Experimental | Manipulate an IV, use a control group, randomly assign | Cause and effect (strongest) |
| Quasi-experimental | Intervention, but no random assignment | Causal-ish questions when randomization isn’t possible |
| Correlational | Measure relationships without manipulating | Association between variables |
| Descriptive | Characterise a population or phenomenon | “What is happening / how common” |
| Cross-sectional | One point in time (a snapshot) | Prevalence, quick comparisons |
| Longitudinal | Same subjects measured over time | Change, development, trajectory |
Common in-depth (often qualitative) designs
- Case study — an intensive examination of one (or a few) bounded cases.
- Ethnography — immersion in a culture or setting over time.
- Phenomenology — the lived experience of a phenomenon.
- Grounded theory — building theory from the data itself.
The one distinction examiners care about
Only a true experiment — manipulation + control group + random assignment — gives a strong basis for saying X causes Y. Quasi-experiments are weaker; correlational and cross-sectional designs show that things go together, not that one causes the other. Claiming causation from a correlational design is the classic over-reach. (Choosing qual vs quant first? See qualitative vs quantitative.)
Get the free Research Design toolkit
A design-type decision chart, worked examples, and a validity checklist from Research Design Simplified. We’ll email you the download link.
Frequently asked questions
What are the main types of research design?
Experimental, quasi-experimental, correlational, descriptive, cross-sectional, and longitudinal (quantitative); case study, ethnography, phenomenology, grounded theory (qualitative).
Cross-sectional vs longitudinal?
Cross-sectional = a snapshot at one time; longitudinal = the same subjects over time, so it can show change.
Which design shows causation?
A true experiment (manipulation + control + random assignment). Correlational/cross-sectional designs show association only.
How do I pick one?
Match it to your question and what you can ethically/practically do — manipulate & randomize for cause-and-effect, observe for association, longitudinal for change.