Research Design Selector
Answer a few questions about your research purpose, evidence, timing, and level of control to identify study designs that may fit your project.
Find a suitable research design
Recommendations update as you describe the study.
Describe your study
Your suggested design and alternatives will appear here.
Choose a design from the question—not the software
A research design is the logic connecting the question, evidence, sampling, measurement, and analysis.
The same statistical method can appear in several designs, but the strength of the conclusion depends on how the evidence was produced. Random assignment, comparison groups, temporal ordering, repeated measurement, and triangulation each solve different methodological problems.
Use the selector during proposal development, then refine the recommendation into a complete protocol.
Research purpose
Description, explanation, evaluation, interpretation, development, and synthesis require different strategies.
Level of control
Assignment and manipulation determine whether experimental reasoning is possible.
Timing
Cross-sectional, longitudinal, prospective, retrospective, and time-series designs answer different questions.
Evidence integration
Mixed methods can combine breadth, explanation, development, and contextual depth.
Common research designs at a glance
Use these labels as starting points, not substitutes for a full protocol.
| Purpose | Typical design | Best suited for |
|---|---|---|
| Describe a population | Descriptive cross-sectionalSurvey | Prevalence, characteristics, attitudes, practices, or current conditions |
| Study natural associations | CorrelationalAnalytical cross-sectional | Relationships among variables measured without intervention |
| Observe outcomes over time | Prospective cohortRetrospective cohort | Temporal sequence, incidence, trajectories, and risk factors |
| Compare prior exposures | Case-control | Rare outcomes or outcomes with long development periods |
| Estimate causal effects | Randomized controlled trialCluster RCT | Interventions that can be assigned ethically and practically |
| Evaluate without randomization | Quasi-experimentalInterrupted time seriesRegression discontinuity | Programs, policies, natural experiments, and implementation settings |
| Understand lived experience | Phenomenology | Meaning and structure of a shared human experience |
| Develop an explanatory theory | Grounded theory | Processes, interactions, stages, and actions |
| Understand culture or practice | Ethnography | Groups, communities, norms, routines, and social worlds |
| Investigate a bounded case | Case study | Programs, organizations, events, classrooms, or individuals in context |
| Integrate qualitative and quantitative evidence | ConvergentExplanatory sequentialExploratory sequential | Triangulation, explanation, instrument development, and implementation |
| Synthesize existing evidence | Systematic reviewScoping reviewMeta-analysis | Evidence summaries, gaps, pooled effects, and research landscapes |
Before finalizing the design
- Specify the unit of analysis. Individuals, classrooms, institutions, documents, and events require different sampling and analysis plans.
- Establish temporal order. Decide whether the proposed timing can show that an exposure preceded an outcome.
- Plan a comparison. For impact evaluation, define the counterfactual or comparison condition explicitly.
- Match sampling to claims. The sampling strategy limits which population or context the findings can represent.
- Align analysis with design. Repeated, clustered, weighted, nested, and longitudinal data require methods that preserve their structure.
Design labels are not enough
Calling a study experimental, phenomenological, mixed-methods, or systematic does not by itself establish rigor. The protocol must explain recruitment, data sources, measurement quality, researcher positioning, analytic procedures, integration, missing data, fidelity, and threats to inference.
Hybrid designs are common. A project may combine a quasi-experiment with interviews, embed a process evaluation within a trial, or use a sequential mixed-methods approach to develop and validate an instrument.
Frequently asked questions
Start with the research question, the kind of evidence needed, whether variables can be manipulated, how participants can be assigned, when measurements occur, and whether the goal is description, explanation, evaluation, prediction, or deep understanding.
No. It provides an educational starting point. Feasibility, ethics, sampling, measurement quality, disciplinary conventions, and the exact research question may change the most appropriate design.
Experimental research deliberately assigns or manipulates an intervention. Observational research measures naturally occurring exposures, characteristics, or outcomes without assigning the exposure.
A randomized controlled trial is appropriate when an intervention can be assigned ethically and practically, a comparison condition is available, and random assignment can be implemented at the participant or cluster level.
A quasi-experimental design evaluates an intervention without full random assignment. Common forms include nonequivalent control groups, interrupted time series, regression discontinuity, and difference-in-differences designs.
Use a cross-sectional design when variables are measured at one period to describe a population, estimate prevalence, or examine associations. It is generally limited for establishing temporal order.
Use a cohort design when participants are grouped by an exposure or characteristic and followed over time to observe outcomes. Cohorts may be prospective or retrospective.
Phenomenology explores the meaning and structure of lived experience. Grounded theory develops an explanatory process or theory from systematically collected and compared data.
Use a case study for an in-depth investigation of a bounded case, such as a program, institution, event, community, or individual, using multiple sources of evidence.
Use convergent design when qualitative and quantitative strands are collected in parallel, explanatory sequential when quantitative findings need qualitative explanation, and exploratory sequential when qualitative findings guide instrument or model development.
Action research may use qualitative, quantitative, or mixed methods. Its defining feature is a collaborative cycle of planning, action, observation, and reflection to improve local practice.
Report the design label, rationale, setting, participants, sampling, timing, intervention or exposure, comparison condition, measurement schedule, data sources, analytic approach, and design-specific limitations.