✦ScienexySCIENCE KNOWLEDGE LIBRARY
concept · SCI-PSY-001-120

Research Design and Method Selection

Research design is the overall plan that connects a research question to participants, measurements, data collection and analysis. Method selection involves choosing procedures that can answer the question accurately, ethically and realistically.

Explanation

A fuller explanation is being developed. Verify this concept with cited sources before relying on it.

Example or application

A worked example has not yet been added.

Conceptual Coverage

A research design specifies the structure of a study and the logic through which evidence will address its questions. Major design families include descriptive, correlational, experimental, quasi-experimental, qualitative, longitudinal, cross-sectional and mixed-methods designs. A design may be exploratory, explanatory, evaluative or predictive. Method selection concerns the specific tools used within the design, such as questionnaires, interviews, observations, cognitive tasks, physiological measures or document analysis. Researchers should align the design with the research objective, type of variables, feasibility, ethical constraints and desired scope of inference. A research protocol records these decisions in advance.

Scientific Foundation

Research design draws on experimental methodology, survey science, qualitative inquiry, statistics and causal inference. The development of randomized trials strengthened the evaluation of interventions, while longitudinal and observational designs expanded research into real-world development and social behaviour. Modern design guidance emphasizes internal validity, external validity, construct validity and statistical conclusion validity, alongside qualitative criteria such as credibility and reflexivity. No single design is best for every psychological question.

Deep Understanding

The research question should determine the design, not the other way around. To estimate the frequency of a behaviour, a representative descriptive survey may be suitable. To investigate an association, a correlational design may be appropriate. To test whether an intervention changes an outcome, an experiment may be strongest when random assignment is ethical and feasible. To understand lived experience, interviews or ethnography may be more appropriate. If the question requires both estimating a pattern and explaining participants' experiences, mixed methods may add value. Researchers should also decide the unit of analysis, measurement timing, comparison conditions, recruitment strategy and analysis plan before collecting data.

Mathematical Coverage

Design selection does not have one universal formula. Quantitative designs may require sample-size and power calculations. For a simple comparison of two independent means, sample-size planning depends on the expected standardized effect size d = (μ₁ − μ₂)/σ, the significance level, desired power and allocation ratio. Repeated-measures designs must account for within-person correlation. Clustered designs may require a design effect, often approximated as DEFF = 1 + (m − 1)ICC, where m is average cluster size and ICC is the intraclass correlation. These are planning tools, not substitutes for a justified design. Qualitative sample planning is commonly based on the scope of the question, participant diversity, analytic approach and information adequacy rather than a universal numerical formula.

Practical Learning

For a question about whether a short educational programme improves delayed recall, compare possible designs. A randomized experiment may provide stronger causal evidence; a quasi-experiment may be more practical in an existing classroom; a qualitative interview study may explain how participants experience the programme but cannot alone estimate its causal effect. Write a short protocol identifying the research question, design, population, sampling, measures, procedure, analysis and ethical protections. Explain why alternative designs were not selected.

Critical Thinking

A sophisticated statistical method cannot rescue a design that fails to measure the intended construct or address the research question. A design may be internally strong but difficult to generalize beyond its setting. Practical constraints should be acknowledged, not hidden. Researchers should avoid selecting methods because they are fashionable or convenient and should not treat a single validity label as a complete quality assessment. The design should also anticipate missing data, participant burden and unintended consequences.

Knowledge Assessment

1. What is the main role of research design? Answer: To organize how evidence will answer the research questions. 2. Which design is generally suitable for testing a causal intervention effect when feasible and ethical? Answer: A randomized experiment. 3. Why might interviews be preferable to a closed survey for exploring unfamiliar experiences? Answer: They allow participants to explain meanings and issues not anticipated by the researcher. 4. What does ICC represent in clustered research? Answer: Intraclass correlation, a measure of similarity within clusters. 5. Application: Select a design for studying the prevalence, predictors and personal experiences of academic stress, explaining whether a mixed-methods approach is justified.

Further Learning

Study research paradigms, validity, power analysis, sampling, measurement, protocol development, preregistration and causal inference. Then progress to advanced experimental, qualitative, longitudinal and mixed-methods design.

Sources and verification

Sources are shown as provided in the project data. Not every source has yet been independently verified. Please report errors or missing citations.

  • Source verification pending; check an authoritative textbook, university, standards body, or peer-reviewed source.

Suggest a correction or source · More in Psychology