Your research design is the blueprint of your study — the overall plan that connects your research question to the evidence you will collect. Choosing the right one is one of the most consequential decisions in your methodology, because the design determines what kinds of conclusions you can legitimately draw. A descriptive design can tell you what is happening; only an experimental design can credibly tell you what causes it.
This guide walks through the main types of research design, with examples, and shows how to pick the one that fits your question.
What is a research design?
A research design is the structured framework that specifies how you will collect and analyse data to answer your question — including who or what you study, when, and under what conditions. It is distinct from your methods (the specific tools) and broader than them: the design is the logic; the methods are the instruments that serve it.
Designs are commonly grouped by their purpose, mapping onto the exploratory–descriptive–explanatory progression you may know from the types of research.
Exploratory research design
Use an exploratory design when little is known about a topic and your goal is to understand it better, clarify the problem, and generate hypotheses rather than test them. Methods are flexible: open-ended interviews, focus groups, literature scoping, pilot studies.
Example: Before designing a national programme, a team runs open interviews with users of a brand-new digital-savings product to learn what matters to them.
Exploratory designs do not produce conclusive answers — and they are not meant to. Their output is a sharper question and a set of candidate variables.
Descriptive research design
A descriptive design sets out to describe a population, situation, or phenomenon accurately — the what, who, where, and when, but not the why. Surveys and observational studies are the workhorses here.
Example: A cross-sectional survey reporting the prevalence of hypertension across age groups in a district.
Descriptive designs are excellent for establishing the lay of the land, but they cannot establish causation on their own.
Correlational research design
A correlational design measures the relationship between two or more variables without manipulating them — quantifying whether, and how strongly, they move together.
Example: Measuring whether students' study hours correlate with their exam scores.
The crucial caveat: correlation is not causation. A correlational design can reveal that two things are associated, but not which causes which, or whether a third factor drives both.
Experimental research design
An experimental design is the gold standard for establishing cause and effect. You manipulate an independent variable, control other conditions, and measure the effect on a dependent variable — ideally with random assignment to treatment and control groups.
Example: Randomly assigning patients to a new treatment or a placebo and comparing outcomes.
Because experiments control for confounding factors, they support the strongest causal claims. Quasi-experimental designs are a practical cousin: they compare groups without full random assignment (often because randomization is impossible or unethical), trading some causal certainty for real-world feasibility.
Cross-sectional vs longitudinal designs
This pair concerns time, and can overlay any of the above:
- Cross-sectional: data collected at a single point in time — a snapshot. Fast and cheap, but cannot show change.
- Longitudinal: the same subjects studied repeatedly over time — able to reveal trends, development, and (with care) cause. More powerful, but slower and costlier, with the risk of participants dropping out.
Example: A cross-sectional survey measures literacy in Primary 4 today; a longitudinal study follows the same pupils from Primary 1 to Primary 6.
Case study design
A case study investigates one (or a few) instances — a person, organization, community, or event — in depth and in context. It excels at rich, holistic understanding and is common in business, education, and the social sciences.
Example: An in-depth study of how a single rural clinic implemented a new patient-records system.
The trade-off is generalizability: insights are deep but may not transfer directly to other cases.
How to choose the right design
The design follows from the question, not the other way around:
- What is the question really asking? Description, association, or causation?
- What is happening? → descriptive
- Are these related? → correlational
- Does X cause Y? → experimental / quasi-experimental
- What is going on here, in depth? → case study
- We barely know the area → exploratory
- Over what time frame? A snapshot (cross-sectional) or change over time (longitudinal)?
- What is feasible and ethical? You may want an experiment but only be able to run an observational study — say so, and acknowledge the limit.
Tie the design back to your methodology
A research design is not a standalone choice — it has to cohere with your philosophy, your sampling, and your analysis. An experimental design implies probability sampling and statistical testing; an exploratory case study implies purposive sampling and qualitative analysis. Keeping that whole chain aligned is the essence of a strong methodology.
PaceReseacher's Methodology Copilot helps you select a design that fits your question and then carries that choice consistently through your sampling and analysis sections, citing real methodological literature as it goes. Return to the research methodology hub to see how design connects to the rest of your chapter, or sharpen your sample with our guide to sampling techniques.