Research Design

Research design is the overall plan that connects a research question to the evidence needed to answer it.
555
minutes read

What Is Research Design?

Research design is the overall plan that connects a research question to the evidence needed to answer it. A research design sets out what will be studied, who will be included, how data will be collected, and how comparisons will be made.

A good design makes the eventual conclusions defensible. A weak one can leave even careful data collection unable to answer the question, which is why peer reviewers, ethics committees, and funders such as the NIH scrutinize this part of a study so closely.

‍

What Are the Main Types of Research Design?

The main types of research design fall into six broad families:

  • Experimental design: the researcher assigns participants to conditions, ideally at random, to test cause and effect. Randomized controlled trials (RCTs) are the best-known example and are reported using the CONSORT guidelines.
  • Quasi-experimental design: an intervention is tested without random assignment, often using existing groups, before-and-after comparisons, or interrupted time series.
  • Observational study: the researcher measures exposures and outcomes without intervening, as in cohort, case-control, and cross-sectional studies, which are reported using the STROBE guidelines.
  • Longitudinal study: the same participants are followed over time to track change, in either observational or interventional work. The Framingham Heart Study, running since 1948, is a well-known example.
  • Mixed methods research: quantitative and qualitative approaches are combined to answer different parts of a question.
  • Descriptive and qualitative designs: surveys, case studies, and interview-based studies that describe experiences or phenomena in depth, often reported using COREQ or SRQR.

‍

How Do You Select a Design for Your Research Question?

Selecting a research design starts with the question: the question should drive the design, not the other way around. A quick walkthrough helps:

  1. Define what you want to know: a cause, an association, a prevalence, or an experience.
  2. Check whether you can ethically and practically assign the exposure. If not, a non-experimental approach may be the only option, which is why the evidence linking smoking to lung cancer comes from cohort and case-control studies such as the British Doctors Study.
  3. Decide whether timing matters. Questions about change or sequence usually need repeated measurements.
  4. Weigh feasibility, including budget, access to participants, timelines, and existing data.
  5. Match the design to the strength of evidence your audience expects. The Oxford CEBM Levels of Evidence, for example, place systematic reviews of randomized trials at the top for treatment questions.

Funders look for this logic explicitly: NIH peer review assesses the rigor and feasibility of a study's approach, and a mismatch between question and design is one of the common grant proposal weaknesses reviewers point out.

‍

What Is the Difference Between Research Design and Methodology?

Research design and methodology are often used interchangeably, but they describe different layers. Design is the blueprint: the type of study and the structure of its comparisons. Methodology covers the specific procedures used to carry out that blueprint, such as sampling techniques, measurement instruments, data collection protocols, and analysis methods.

For example, choosing a cohort study is a design decision. Deciding how to recruit participants, which questionnaire to use (such as the SF-36 health survey), and which regression model to fit (such as a Cox proportional hazards model) are methodological decisions. Keeping the distinction clear makes a methods section easier to follow.

‍

Which Design Flaws Do Reviewers Flag Most Often?

Peer reviewers flag a familiar set of design flaws. A frequent one is a design that cannot answer the stated question, such as causal claims drawn from cross-sectional data. Missing or poorly matched comparison groups are another common concern.

Reviewers also flag inadequate sample size, unaddressed confounding, selection bias in recruitment, and outcomes that changed after data collection began. Outcome switching is common even in leading journals: the COMPare project found that 58 of 67 trials (87%) published in the New England Journal of Medicine, The Lancet, JAMA, The BMJ, and Annals of Internal Medicine had discrepancies between pre-specified and reported outcomes that required a correction letter.

Most of these issues are far easier to prevent at the planning stage than to fix later: a 2015 BMJ study of 1,286 trials in Cochrane reviews found that 43% had at least one domain at high risk of bias, and estimated avoidable waste due to inadequate methods at 42% once incomplete reporting was accounted for.

‍

FAQ

Can a study combine more than one research design?

Yes. A study can combine more than one research design, for example a randomized trial with an embedded qualitative component, or a cohort study with a nested case-control analysis. Combining approaches can answer broader questions, but each component needs its own clear rationale, and authors should explain how the parts fit together, as the GRAMMS guideline for mixed methods studies recommends.

Who decides the research design for a study?

The principal investigator (PI) usually leads the decision, working with co-investigators, statisticians, and methodologists. Funders, ethics committees such as institutional review boards (IRBs), and sometimes regulators such as the FDA or the EMA shape the final plan through their requirements. In student research, supervisors play a major role in shaping and approving the design before data collection begins.

Can a research design change mid-study?

A research design can change mid-study, but changes should be justified, documented, and approved by the relevant ethics committee where required. Adaptive trial designs, covered in the FDA's 2019 guidance on adaptive designs, plan for specific changes in advance, such as dropping a treatment arm or adjusting the sample size at an interim analysis. Unplanned changes, especially to primary outcomes, can introduce bias and should be reported transparently, including any amendments to protocols registered on ClinicalTrials.gov or the ISRCTN registry.

Is a pilot study part of research design?

A pilot study is a small preliminary study that tests feasibility before the main project. A pilot study is not the main design itself, but it informs design decisions by revealing problems with recruitment, procedures, or measurement. The CONSORT extension for randomised pilot and feasibility trials, published in 2016, sets out how to report one. Many funders, including the NIHR in the UK, expect pilot data to show that the full study is feasible.

Where does research design appear in a paper?

Research design is usually described at the start of the methods section, often in the first sentence or under a “Study design” subheading. Reporting guidelines such as CONSORT and STROBE also ask authors to name the design in the title or abstract, which helps readers and indexing services such as PubMed identify the study type quickly.

Free access for academic researchers

Create your free QED account to validate your research, strengthen grant proposals, and uncover scientific insights.
We've sent you an access link.
Please check your inbox.

Didn't get your email? Check your spam folder or reach out to info@qedscience.com

Oops! Something went wrong while submitting the form.
Looking for QED for pharma, biotech or life science organizations?