Have a research question but not a design your methodologist will accept? This guide compares quantitative, qualitative and mixed-methods designs, explains validity and rigor, and gives decision rules for matching a method to your question.
If your methodologist has asked you to "justify your design" and you are not sure why a survey, an interview study or a trial is the right choice, this guide is for you. It gives a plain-language map of the main design families, the words markers expect for validity and rigor, and a way to match method to question.
Students often pick a familiar method first, such as a survey, and then bend the question to suit it. Reverse the order.
Quick answer. Ask what kind of answer your question needs. A count or comparison points to quantitative work. An experience or process points to qualitative work. A question that needs both a measure and an explanation may need mixed methods. Then check that you can actually run the design with your access, time, skills and ethics approval.
Behind the methods sit different assumptions about knowledge. A post-positivist view treats reality as something that can be measured, with error, and tested. An interpretivist or constructivist view holds that people construct meaning, so understanding requires their accounts. A pragmatic view chooses whichever methods best answer the question, which is the usual stance behind mixed-methods work.
You rarely need to write an essay on philosophy, but a sentence stating your stance and how it fits your design shows a committee that your choices are coherent. If a program requires a stronger statement, see the theoretical framework guide.
Quantitative research measures variables numerically and analyses them statistically. It suits questions about how much, how many, whether groups differ and whether variables are related.
| Design | What it does | Main strength | Main weakness |
|---|---|---|---|
| Randomized controlled trial | Randomly assigns participants to conditions | Best basis for causal claims | Costly, sometimes impractical or unethical |
| Quasi-experimental | Tests an intervention without random assignment | Feasible in practice settings | Confounding and selection bias |
| Cohort | Follows groups over time to see who develops an outcome | Shows sequence of exposure and outcome | Time, attrition, cost |
| Case-control | Compares people with and without an outcome, looking back at exposures | Efficient for rare outcomes | Recall and selection bias |
| Cross-sectional | Measures everything at one time point | Fast, good for description and association | Cannot show cause and effect |
Randomized trials are reported with CONSORT, and observational designs with STROBE. The EQUATOR Network hosts these and other reporting guidelines. Reading the checklist for your design before you collect data shows you what reviewers will expect.
Name your independent and dependent variables, and say how each is measured. Also name likely confounders and how you will handle them, whether by design (matching, restriction, randomization) or by analysis (adjusting in a regression). If the concept in your study is not measured well by existing scales, say so as a limitation.
Qualitative research explores meaning, experience and process, usually through interviews, focus groups, observation or documents. It suits "how" and "what is it like" questions where the important variables are not yet known.
| Tradition | Best for | Typical data | Example question form |
|---|---|---|---|
| Phenomenology | The essence of a lived experience | In-depth interviews | What is it like to care for a dying patient for the first time? |
| Grounded theory | Building a theory of a social process | Interviews, observation, iterative sampling | How do nurses manage moral distress over time? |
| Ethnography | The culture of a group or setting | Fieldwork, observation, interviews | How does unit culture shape how staff speak up? |
| Qualitative description | A close, plain account of an experience | Interviews, focus groups | What do patients say helped them after discharge? |
The questions above are examples of form, not studies that have been done. Whichever tradition you choose, name your analytic approach, such as thematic analysis, and describe the steps.
Reporting guidance comes from COREQ (interviews and focus groups) and SRQR (qualitative research generally). Explain your position as a researcher and how your assumptions might shape the analysis. Note how you kept an audit trail of decisions, so a reader can follow how data became themes.
Mixed methods combine both strands so that each answers a part of the question the other cannot. Use it only when the question genuinely needs both, because the design costs more time and demands skill in two traditions.
Markers expect you to show how you protected the quality of your findings, using the vocabulary of your tradition.
| Tradition | Quality concept | What it asks |
|---|---|---|
| Quantitative | Internal validity | Could something other than the intervention or exposure explain the result? |
| Quantitative | External validity | Do the findings apply beyond this sample and setting? |
| Quantitative | Reliability | Would the instrument give consistent results? |
| Qualitative | Credibility | Do the findings ring true to participants? (member checking, prolonged engagement) |
| Qualitative | Transferability | Is the context described richly enough for readers to judge fit? |
| Qualitative | Dependability and confirmability | Is there an audit trail, and is reflexivity documented? |
Name the specific threats relevant to your design and say what you did about each.
Probability sampling gives every member of the population a known chance of selection, which supports generalizing. Non-probability sampling is common in nursing research and is acceptable when you are honest about its limits.
| Approach | How it works | Use when |
|---|---|---|
| Simple random | Everyone has an equal chance, drawn from a complete list | You have a sampling frame and want to generalize |
| Stratified | Random sampling within subgroups | Subgroups must be represented |
| Convenience | Recruit whoever is accessible | Access is limited, and you state the limits |
| Purposeful | Choose people who can speak to the topic | Qualitative work that needs rich data |
| Snowball | Participants refer others | Hard-to-reach groups |
Qualitative studies typically judge sample adequacy by the richness and relevance of the data rather than a fixed number. Whatever you choose, describe the population, the recruitment route and the criteria for inclusion and exclusion. For quantitative sample size, explain how the number was reached, often by a power analysis.
The way you collect data limits the way you can analyze it. Decide both together, before you build instruments or recruit anyone.
| Design | Typical analysis | What you report |
|---|---|---|
| Two-group comparison | Independent-samples t-test, or a non-parametric alternative | Group statistics, test result, effect size, confidence interval |
| Association between variables | Correlation or regression | Coefficients, confidence intervals, model fit |
| Binary outcome with several predictors | Logistic regression | Odds ratios with confidence intervals |
| Interview study | Thematic or content analysis | Themes, illustrative quotes, audit trail |
| Grounded theory | Constant comparison and coding | A theory with its categories and their links |
If you plan a statistical analysis and are unsure which one applies, the biostatistics guide explains the common tests in plain terms.
A small pilot can test recruitment, instruments and procedures before the main study. Present it honestly as a pilot: it estimates feasibility, not effectiveness, and a pilot with a tiny sample cannot prove that an intervention works. Say what you would change based on the results.
Illustrative example, not a real client.
Problem. A doctoral student wanted to understand why new graduate nurses leave their first unit. She planned a large survey because surveys were what her program taught.
Tension. Her methodologist asked one question: "Can a rating scale tell you why someone left?" She could not answer, and the proposal meeting was days away.
Turn. She rewrote her question in plain words and saw that it had two halves: how common are intentions to leave, and what explains them. She chose an explanatory sequential design, with a short survey to measure intent and follow-up interviews to explain it.
Proof. The methodologist accepted the rationale because each strand answered a different half of the question. The committee's remaining questions were about how the interview sample would be drawn from survey respondents, which is a normal, solvable question.
Payoff. Her methods chapter was easier to write because every choice in it traced back to one sentence: what kind of answer does the question need?
Send your question and program requirements, and get a quote for a design rationale or a complete methods section. The price is shown before you pay, and revisions are free for 14 days.
Try translating your question into one of these forms and follow the decision rule.
Then check feasibility: access, time, skills, cost and ethics. A perfect design you cannot run is not the right design. See the IRB and research ethics guide for the ethics step.
When two designs both fit your question, choose the one that scores better on these checks.
A smaller, well-executed study nearly always earns more credit than an ambitious design that runs out of time.
The scenario below is fictional and only shows the reasoning.
| Step | What the student writes |
|---|---|
| Topic | Nurse-led discharge teaching for patients with heart failure |
| Draft question | Does a nurse-led teaching session improve patient confidence in self-care? |
| Type of answer needed | A comparison between groups, and possibly the reasons behind it |
| Design options | Randomized trial (not feasible in one clinic), quasi-experimental comparison of two units, or a mixed-methods study |
| Chosen design | Quasi-experimental, with a short interview strand to explain results |
| Threats named | Differences between units, staff turnover, self-report bias |
| Handling | Collect baseline data on both units, adjust for differences in analysis, use a validated scale |
Each row follows from the one before. That chain is what a committee is looking for when it says "justify your design".
A design rationale is usually two to four sentences that link question, design and limits. Show the reasoning rather than just naming the design.
| Weak | Stronger |
|---|---|
| A survey was used because it is easy. | A cross-sectional survey was chosen because the aim is to describe and compare groups at one point in time. It cannot show cause, so results are described as associations. |
| The study is qualitative because I want to hear their voices. | A qualitative description design was chosen because little is known about how patients experience the transition home, and participants' own accounts are needed to identify what matters to them. |
| Mixed methods will give a complete picture. | An explanatory sequential design was chosen because survey results are expected to show whether confidence changed, and interviews are needed to explain why. |
Use the methods section guide to turn the rationale into a complete chapter, and the results and discussion guide to report what comes next.
You can sharpen your sense of design by classifying every study you read for your literature review. Ask six questions of each paper, and write the answers in your synthesis matrix.
| If the abstract says | The design is probably |
|---|---|
| Participants were randomly assigned | Randomized controlled trial |
| A comparison group without random assignment | Quasi-experimental |
| People were followed over time to see who developed the outcome | Cohort |
| People with the outcome were compared with people without it | Case-control |
| Data were collected at one point in time | Cross-sectional |
| Interviews were analyzed for themes | Qualitative description or thematic analysis |
| Both statistical results and interviews were reported | Mixed methods, if planned together |
Authors sometimes label a design incorrectly, so trust the methods section over the title. Practicing this classification makes your own design rationale more precise, and it helps you criticize a study's limitations accurately. The synthesis matrix guide shows how to record it.
Reporting guidelines list what readers need to judge a study. Reading the one for your design early tells you what to record while you work, so you do not have to reconstruct it later.
| Study type | Reporting guideline |
|---|---|
| Randomized trial | CONSORT |
| Cohort, case-control or cross-sectional study | STROBE |
| Qualitative interviews or focus groups | COREQ |
| Qualitative research more broadly | SRQR |
| Systematic review | PRISMA 2020 |
| Quality improvement study | SQUIRE 2.0 |
The EQUATOR Network hosts the full set, and the PRISMA statement site hosts the review guideline. Naming the guideline in your methods, and following its checklist, shows that you know what a complete report looks like. Journals commonly ask for the completed checklist on submission, so keeping it as you go saves time.
A checklist item you cannot answer, such as how participants were allocated or how missing data were handled, is a sign that your design has a gap. Fix it in the plan rather than in the discussion of limitations. For a proposal that puts the whole plan together, see the research proposal guide.
Yes, if the sample is well matched to the question and the data are rich. Adequacy is judged by depth and how well the analysis is supported, not by a target number.
Reliability is consistency of measurement. Validity is whether the instrument or study measures or concludes what it claims to. An instrument can be reliable without being valid.
Many programs expect one. Check your handbook, and see the theoretical framework guide for how to apply it.
It depends on the questions and the analysis. Closed-ended items analyzed statistically are quantitative. Open-ended responses analyzed for meaning are qualitative. A survey with both can support a mixed-methods approach if planned that way.
Only cautiously. Without random assignment, differences between groups may explain the result, so describe findings as evidence consistent with an effect and state the alternative explanations you could not rule out.
Your design is an argument that your evidence can answer your question. If you make that argument in plain terms and admit its limits, your methods section will read as confident rather than defensive. For a full plan around your chosen design, read the research proposal guide.
To get a methods section drafted or reviewed, get your instant quote. The price is shown before you pay. Delivered work comes with 14 days of free revisions, and refund terms are on the money-back guarantee page. Use any support in line with your institution's academic-integrity rules.