Your dissertation research design is the overall plan for answering your research questions: what kind of data you collect, from whom, and how you compare or interpret it. It decides what kind of conclusion your study is allowed to draw.
This guide sets out the main designs, the claims each can support, how to choose between them and how to justify the choice. The methodology chapter then describes the design in detail; this page is about making the decision well.
How to Choose a Research Design
- Decide what your question needs: a cause, a relationship, a description, or an understanding of meaning and experience.
- Pick the family that delivers it: quantitative for measurement, qualitative for meaning, mixed methods when you genuinely need both.
- Choose the specific design within that family, checking access, time, ethics and skills.
- Justify it by showing fit, explaining rejected alternatives and stating what the design cannot claim.
What Research Design Means in a Dissertation
Terms overlap between textbooks, so it helps to separate three levels. Your methodology is the overall approach and its philosophical stance. Your design is the plan, such as a quasi-experiment or a case study. Your methods are the specific tools, such as a survey or interviews.
Design comes before method. Choosing "a survey" first and building a study around it often produces questions the data cannot answer. Start with the question, choose the design, then choose methods that serve it.
Quantitative Research Designs
Quantitative designs measure variables numerically and analyze them statistically. They differ mainly in whether you manipulate anything and whether groups are formed randomly.
| Design | What it does | Claim it supports |
|---|---|---|
| Experimental | Randomly assigns participants to conditions and manipulates the independent variable | Causal, within the study's limits |
| Quasi-experimental | Compares existing groups or time points without random assignment | Cautious association; causal claims limited by group differences |
| Correlational | Measures variables as they occur and tests relationships or prediction | Association and prediction, not causation |
| Descriptive or survey | Summarizes characteristics, prevalence or attitudes | Description of the sample and, with good sampling, the population |
| Longitudinal | Measures the same people repeatedly over time | Change over time and temporal order |
Random assignment is what lets an experiment rule out most rival explanations. In applied settings it is often impractical or unethical, which is why quasi-experimental and correlational designs are so common in doctoral work.
Correlational studies are the most frequently overstated. A regression supports language such as "predicts" or "is associated with", never "causes" or "impacts".
Qualitative Research Designs
Qualitative designs explore meaning, experience and process through text, talk and observation. Samples are small and purposeful, chosen because participants can speak richly to the phenomenon.
| Design | Central question | Typical data |
|---|---|---|
| Phenomenology | What is the lived experience of this phenomenon? | In-depth interviews |
| Grounded theory | What theory explains this process? | Interviews with theoretical sampling and constant comparison |
| Case study | How and why does this happen in this bounded case? | Several sources: interviews, documents, observation |
| Ethnography | How does this group's culture shape behavior? | Extended fieldwork and observation |
| Narrative inquiry | How do people story their experiences? | Life stories and extended interviews |
| Generic qualitative | How do people describe or understand this issue? | Interviews analyzed thematically |
Qualitative findings are transferable to similar settings through detailed description rather than statistically generalizable. Naming a specific design, instead of writing "qualitative" alone, signals to your committee that you know what tradition you are working in.
Mixed Methods Designs
Mixed methods combine quantitative and qualitative strands with a deliberate point of integration. The three core designs are widely described in the methods literature.
- Convergent: both strands collected in the same phase and compared to see where they agree or differ.
- Explanatory sequential: quantitative first, then qualitative to explain the results.
- Exploratory sequential: qualitative first, then a quantitative phase that builds on it, such as a new instrument.
Mixed methods roughly doubles the work. Choose it only when one strand alone cannot answer your question, and plan exactly where integration happens.
Matching Your Question to a Design
| If your question asks | Consider |
|---|---|
| Does this intervention cause better outcomes? | Experimental, or quasi-experimental if randomization is impossible |
| Is X related to Y, or does X predict Y? | Correlational |
| How common is X in this population? | Descriptive survey with a sound sampling plan |
| What is it like to experience X? | Phenomenology |
| How and why did X unfold in this organization? | Case study |
| How much, and why? | Explanatory sequential mixed methods |
The wording of your research questions should signal the design. "What is the effect of" implies a causal design, so a correlational study usually needs "What is the relationship between" instead.
How to Justify Your Design Choice
A justification that satisfies a committee has four parts. Weak justifications usually skip one of them.
- Fit: state what the question requires and show the design delivers it.
- Feasibility: explain access, time and ethics, including why a stronger design was not possible.
- Alternatives: name at least one design you considered and why you rejected it.
- Limits: say plainly what the design cannot claim.
Support each step with methods literature in your field. A design justified only by "it was convenient" invites hard questions in your proposal defense.
A Worked Example: Choosing a Design
Question. Does a new onboarding program improve 12-month retention of new hires at a mid-sized firm?
Ruled out. A true experiment, because the firm will not withhold onboarding from half its new hires.
Chosen. A quasi-experimental design comparing hires who joined after the program began with a historical comparison group from the previous year, adjusting for role and department.
Stated limits. Other changes between the two years, such as the labor market, could explain differences, so results are reported as associations. A power analysis justifies the sample size.
Every choice is explained before the committee has to ask, which is what makes a design defensible.
Validity, Reliability and Trustworthiness
Every design has threats to the quality of its conclusions, and your committee expects you to name them and show how you reduced them.
| Approach | Key quality criteria | Example safeguards |
|---|---|---|
| Quantitative | Internal validity, external validity, construct validity, reliability | Validated instruments, control variables, power analysis, pilot test |
| Qualitative | Credibility, transferability, dependability, confirmability | Member checking, thick description, audit trail, reflexive memos |
| Mixed methods | Quality of each strand plus quality of integration | Joint displays, clear integration points, explaining divergence |
The qualitative criteria above come from Lincoln and Guba's widely used framework. Some programs prefer other terms, so follow your department's language.
Common Research Design Pitfalls
- Causal language in a correlational or quasi-experimental study.
- Calling a study "qualitative" without naming a design or tradition.
- Mixed methods with no clear reason for combining strands or no integration point.
- A sample too small to detect the expected effect, or too broad for in-depth qualitative work.
- A design that needs access you do not yet have, such as organizational records.
Settle the design early. It is the decision that shapes every later chapter, and changing it after ethics approval means amendments and lost months on what is already a long run.
How DissertationHorse Helps with Research Design
Send your topic, draft research questions, program requirements and practical constraints such as access and timeline. A methods specialist at your chosen tier recommends a design, explains the alternatives and drafts a justification as a custom paper for you to study and adapt with your supervisor.
You see the price before you pay, your identity is never shared with the writer, and free revisions within the scope of your original request run.
Want a Design Your Committee Will Approve?
Send your topic, draft research questions and practical constraints. You get a reasoned design recommendation and justification written for your field.
Order Research Design HelpFree revisions · Full refund if late · Written from scratch for your order
Frequently Asked Questions
It is the overall plan for answering your research questions: the type of study, who takes part, what data you collect and how you compare or interpret it.
Methodology is the overall approach and its rationale. Design is the specific plan, such as a case study or quasi-experiment. Textbooks use the terms loosely, so define them in your chapter.
None is best in general. The right design is the one that can answer your question with the access, time and ethics approval you have.
No. It can show association and prediction. Causal claims need a design that controls rival explanations, ideally through random assignment.
Usually qualitative, but case studies can use quantitative data or both. What defines a case study is the focus on a bounded case, not the type of data.
When one strand alone cannot answer your question, for example when you need to measure a pattern and explain why it happens. Plan the integration point from the start.
It is not always required, but a pilot is useful for testing new instruments, recruitment and procedures. Some programs expect one for new surveys.
Possibly, with your committee's agreement, and with an ethics amendment if procedures change. It is far easier to get the design right before approval.