Research methods are the systematic procedures researchers use to collect, analyze, and interpret evidence in order to answer a research question. They include quantitative, qualitative, mixed-methods, experimental, observational, survey, interview, case-study, and evidence-synthesis techniques. The appropriate method depends on the question, intended claim, data, ethics, and available resources.
Introduction
Every defensible research conclusion depends on the methods used to produce it. Researchers must decide what evidence they need, where that evidence will come from, how it will be collected, how it will be analyzed, and what conclusions the design can reasonably support.
Research methods are therefore more than a list of tools such as surveys, interviews, experiments, or statistical tests. They form a connected system of decisions. A strong study aligns its research question, design, sample, measurements, collection procedures, analysis, ethical safeguards, and interpretation.
This guide explains the major types of research methods, how they differ, how to select an appropriate method, how research quality is evaluated, and how modern practices such as open research and artificial intelligence affect methodological work.
Key Takeaways
- Research methods are the procedures used to produce and analyze evidence.
- Quantitative methods examine numerical patterns; qualitative methods explore meanings and experiences; mixed methods integrate both.
- A method should be selected according to the research question and intended conclusion, not because it is familiar or convenient.
- Research design, sampling, measurement, collection, and analysis must be aligned.
- Ethical conduct, transparent reporting, and responsible data management apply throughout the research lifecycle.
- No research method is universally best; every method has assumptions, strengths, and limitations.
What Are Research Methods?
Research methods are organized procedures for obtaining and analyzing evidence. They explain how a researcher moves from a question to a defensible answer.
A complete set of research methods may specify:
- Who or what will be studied.
- How participants, cases, documents, or records will be selected.
- Which variables, experiences, processes, or meanings will be examined.
- How concepts will be measured or explored.
- How evidence will be collected.
- How the evidence will be analyzed.
- How bias, error, and alternative explanations will be addressed.
- How participants, data, and communities will be protected.
- How the study will be documented and reported.
For example, “student engagement” is not directly observable as a single object. A quantitative researcher might measure it using attendance records and a validated scale. A qualitative researcher might explore how students describe belonging, participation, and motivation during interviews. Both studies concern engagement, but they operationalize and investigate it differently.
Research Methods, Methodology, Approach, and Design
These terms are related but not interchangeable.
| Term | Meaning | Example |
|---|---|---|
| Research method | A specific procedure for collecting or analyzing evidence | Interview, questionnaire, experiment, regression, thematic analysis |
| Research methodology | The rationale, assumptions, and logic supporting the choice and use of methods | Explaining why an interpretive methodology is appropriate for studying lived experience |
| Research approach | A broad orientation to evidence and inquiry | Quantitative, qualitative, or mixed methods |
| Research design | The overall plan connecting the question, evidence, procedures, and inference | Randomized experiment, cross-sectional survey, ethnography, case study |
| Research technique or instrument | A practical mechanism used within a method | Interview guide, questionnaire scale, laboratory sensor, coding framework |
A researcher may use semi-structured interviews as a method, within a qualitative case-study design, supported by an interpretive methodology.
This distinction matters because merely naming an instrument does not explain the logic of a study. Saying “a questionnaire was used” does not identify the sampling strategy, measurement quality, timing, analytical model, or conclusions the design can support.
The Three Major Research Approaches
Quantitative research methods
Quantitative research uses numerical data and formal analytical procedures to measure variables, estimate patterns, test hypotheses, compare groups, or model relationships.
Common quantitative methods include:
- Structured surveys
- Experiments
- Quasi-experiments
- Standardized tests
- Physiological or laboratory measurements
- Analysis of administrative records
- Statistical analysis of existing datasets
Quantitative research is useful when the question asks:
- How many?
- How much?
- How frequently?
- Is there a measurable difference?
- Are two variables associated?
- Does an intervention change an outcome?
- How accurately can an outcome be predicted?
A quantitative study might examine whether weekly study time predicts examination performance. The researcher could collect numerical study-time and assessment data and analyze the relationship using regression.
Quantitative evidence can support precise estimates and formal comparisons, but the strength of the conclusion depends on measurement quality, sample selection, design, assumptions, missing data, and potential confounding.
Qualitative research methods
Qualitative research uses nonnumerical evidence to understand meanings, experiences, practices, interactions, contexts, and social processes.
Common qualitative methods include:
- In-depth interviews
- Focus groups
- Participant or nonparticipant observation
- Ethnographic fieldwork
- Document analysis
- Diaries
- Visual and audiovisual methods
- Qualitative case studies
Qualitative research is useful when the question asks:
- How do people experience a situation?
- Why does a process occur in a particular context?
- How do individuals interpret an event?
- How are identities, practices, or decisions constructed?
- What explanations emerge from participants’ accounts?
For example, interviews could explore how first-generation university students experience academic belonging. The goal would not be to reduce every response to a numerical score, but to examine patterns of meaning while preserving relevant context and variation.
Qualitative quality does not come from informality. It requires a clear sampling rationale, careful data production, systematic analysis, reflexivity, transparent interpretation, and evidence connecting findings with the source material.
Mixed-methods research
Mixed-methods research intentionally combines and integrates quantitative and qualitative evidence to answer a question more completely than either approach could answer alone.
A mixed-methods study is not created merely by including a few open-ended questions in a survey. The quantitative and qualitative components must be connected and their findings meaningfully integrated.
Common mixed-methods designs include:
| Design | Sequence | Purpose |
|---|---|---|
| Convergent design | Quantitative and qualitative evidence collected during a similar period | Compare or combine findings |
| Explanatory sequential design | Quantitative phase followed by qualitative phase | Use qualitative evidence to explain numerical findings |
| Exploratory sequential design | Qualitative phase followed by quantitative phase | Use qualitative findings to develop measures, hypotheses, or a larger study |
| Embedded design | One form of evidence placed within a larger design | Address a secondary question or process |
| Multiphase design | Several connected studies or phases | Investigate a complex program or problem over time |
Suppose a university survey shows that commuting students report lower participation. Follow-up interviews could explore timetable constraints, transportation difficulties, employment responsibilities, and perceptions of campus belonging. Integration occurs when both sets of findings are used to construct the final explanation.
Qualitative vs. Quantitative vs. Mixed Methods
| Feature | Quantitative | Qualitative | Mixed methods |
|---|---|---|---|
| Principal focus | Measurement, frequency, comparison, association, prediction, effects | Meaning, experience, context, process, interpretation | Complementary measurement and explanation |
| Typical evidence | Numerical variables | Words, images, observations, documents | Numerical and qualitative evidence |
| Common questions | How many? Is there a difference? What predicts an outcome? | How? Why? What does an experience mean? | What happened, and how or why did it happen? |
| Typical sampling | Probability or structured nonprobability sampling | Purposive, theoretical, maximum-variation, or case-based sampling | One or more connected sampling strategies |
| Common analysis | Descriptive and inferential statistics | Thematic, content, narrative, discourse, or grounded-theory analysis | Separate analyses followed by integration |
| Main quality concerns | Reliability, validity, bias, precision, model assumptions | Credibility, reflexivity, dependability, transferability, interpretive grounding | Quality of each component and quality of integration |
| Common limitation | Important context may be reduced or inadequately measured | Findings may not estimate population prevalence | Requires more time, expertise, and integration planning |
The table describes common patterns rather than rigid rules. Qualitative studies can count features, and quantitative projects can examine complex contextual relationships. The decisive issue is the logic connecting the question, evidence, and conclusion.
Common Types of Research Design
A research approach is broad, while a research design organizes the study as a whole.
Experimental research
Experimental research examines the effect of an intervention by actively assigning or manipulating exposure conditions and comparing outcomes.
In a randomized controlled experiment, participants or units are assigned to conditions by a random process. Randomization can improve the balance of competing explanations, although attrition, noncompliance, measurement problems, and implementation differences may still threaten validity.
Experiments are useful for causal questions but may be infeasible or unethical when researchers cannot appropriately assign exposure.
Quasi-experimental research
Quasi-experimental research evaluates an intervention or event without full random assignment.
Examples include:
- Interrupted time-series designs
- Regression discontinuity designs
- Difference-in-differences studies
- Natural experiments
- Matched comparison groups
- Controlled before-and-after studies
These designs can support stronger causal reasoning than a simple cross-sectional association when their assumptions are appropriate. The researcher must identify and investigate alternative explanations rather than treating every pre–post difference as an intervention effect.
Observational and correlational research
Observational research examines variables without assigning participants to exposure conditions.
A correlational study may estimate whether variables change together. However, an association alone does not establish that one variable caused the other. Reverse direction, confounding, selection effects, and measurement error may explain the pattern.
Observational designs include cross-sectional, cohort, case-control, panel, and longitudinal studies.
Survey research
Survey research collects standardized responses from a sample. It may be:
- Cross-sectional, collecting evidence at one time.
- Repeated cross-sectional, sampling a population at several times.
- Longitudinal, following the same participants or units.
- Descriptive, estimating characteristics or opinions.
- Analytical, examining relationships among variables.
The quality of a survey depends on more than questionnaire design. Coverage, sampling, response rates, nonresponse patterns, question wording, scale quality, administration mode, and weighting can affect the findings.
Case-study research
A case study is an intensive examination of a bounded case in its real-world context.
A case may be a person, organization, community, event, program, policy, classroom, or implementation process. Case studies can use qualitative evidence, quantitative evidence, or both.
A case study is therefore not synonymous with an interview. Interviews may be one source within a case study, alongside documents, observations, records, and other evidence.
Ethnography
Ethnography investigates cultural practices, interactions, meanings, and social organization, commonly through sustained engagement and observation in a setting.
Researchers usually maintain field notes and may also conduct interviews, collect artifacts, or analyze documents. Reflexivity is especially important because the researcher’s presence, relationships, and standpoint influence access and interpretation.
Phenomenological research
Phenomenological research examines how people experience and make sense of a phenomenon. Data are commonly generated through detailed interviews, reflective accounts, or other first-person materials.
Different phenomenological traditions make different philosophical and analytical commitments, so researchers should identify the specific tradition rather than using “phenomenology” as a general label for any interview study.
Grounded theory
Grounded-theory research aims to develop an explanatory theory grounded in systematically collected and analyzed evidence.
Data collection and analysis usually interact. Researchers compare incidents, categories, cases, and emerging ideas as the study develops. Grounded theory should not be claimed merely because themes emerged from interviews; it involves a distinctive logic of theoretical development.
Action research
Action research combines systematic inquiry with efforts to improve practice. It is often cyclical:
- Identify a practical issue.
- Plan an action.
- Implement the action.
- observe outcomes.
- Reflect and revise.
Teachers, health professionals, community organizations, and practitioners frequently use action research. Researchers must still address evidence quality, participant interests, role conflicts, and the limits of local findings.
Systematic review and evidence synthesis
A systematic review uses explicit, reproducible procedures to identify, select, appraise, and synthesize existing studies.
Evidence-synthesis methods include:
- Systematic review
- Scoping review
- Meta-analysis
- Qualitative evidence synthesis
- Realist review
- Rapid review
- Umbrella review
A traditional narrative literature review can support background and interpretation, but it should not automatically be described as systematic. The label depends on the transparency and rigor of the search, eligibility, appraisal, and synthesis procedures.
Data-Collection Methods
Surveys and questionnaires
Questionnaires collect structured self-reported information. They can measure:
- Behaviors
- Knowledge
- Attitudes
- Experiences
- Preferences
- Demographic characteristics
- Psychological or social constructs
Strong questionnaires use clear wording, suitable response options, accessible presentation, appropriate scales, and pilot testing. A long questionnaire is not necessarily comprehensive; unnecessary length can increase burden and incomplete responses.
Interviews
Interviews may be structured, semi-structured, or unstructured.
- Structured interviews use standardized questions and ordering.
- Semi-structured interviews use a guide while allowing relevant probing.
- Unstructured interviews are more conversational but still require a clear research purpose.
Interview evidence is jointly produced through the interaction. Researchers should consider the interviewer’s role, setting, power relationships, recording procedures, transcription decisions, and how interpretations were developed.
Focus groups
A focus group is a facilitated group discussion used to explore shared and differing views. Its distinctive value comes partly from participant interaction.
Focus groups are not simply inexpensive group interviews. Group composition, confidentiality limits, dominant speakers, sensitive topics, facilitator behavior, and disagreement all require careful consideration.
Observation
Observation records behavior, interaction, activity, or context as it occurs. It can be:
- Structured or unstructured
- Participant or nonparticipant
- Overt or covert
- Conducted in naturalistic or controlled settings
Researchers need clear observation protocols and ethical safeguards. Recorded behavior should not automatically be treated as self-explanatory; its meaning may depend on context.
Experiments and measurements
Experimental collection may involve:
- Behavioral tasks
- Laboratory instruments
- Biological measures
- Sensors
- Standardized assessments
- Manipulated messages or environments
Researchers should report calibration, protocol consistency, intervention fidelity, exclusions, adverse events where relevant, and the reliability and validity of measurements.
Documents and archival records
Documents can include:
- Policy materials
- Historical records
- Meeting minutes
- News reports
- Court documents
- Academic publications
- Organizational records
- Social-media posts
- Administrative databases
Documents were produced for particular purposes and audiences. Researchers should examine provenance, authenticity, missingness, access, context, and the difference between the document’s content and the reality it claims to represent.
Digital-trace and computational data
Modern research may use:
- Platform activity
- Search behavior
- Website logs
- Mobile or wearable-sensor data
- Geographic data
- Network data
- Digitized text or images
- Large administrative databases
The availability of digital data does not make its use automatically ethical or representative. Platform populations, algorithmic filtering, consent expectations, re-identification risk, deleted material, automated accounts, and terms of access can affect the study.
Sampling Methods
Sampling determines which people, cases, records, locations, texts, or observations contribute evidence.
Probability sampling
In probability sampling, units have a known or calculable selection probability.
| Type | Description | Appropriate use |
|---|---|---|
| Simple random sampling | Units selected using a random process | When a suitable population list is available |
| Systematic sampling | Every specified interval selected after a random start | Ordered lists without harmful periodic patterns |
| Stratified sampling | Population divided into strata before sampling | Ensuring representation of important subgroups |
| Cluster sampling | Groups such as schools or neighborhoods selected | Geographically dispersed populations or natural groups |
| Multistage sampling | Selection occurs through several levels | Large regional or national studies |
Probability sampling can support population estimation when the sampling frame, selection process, response patterns, and analysis are handled appropriately.
Nonprobability sampling
| Type | Description | Appropriate use |
|---|---|---|
| Purposive sampling | Cases selected because they can illuminate the question | Qualitative, case-based, or specialist research |
| Convenience sampling | Accessible cases recruited | Pilot or exploratory work with explicit limitations |
| Quota sampling | Recruitment fills specified categories | When subgroup balance is needed without probability selection |
| Snowball or chain-referral sampling | Participants assist with further recruitment | Hard-to-reach or networked populations |
| Theoretical sampling | Sampling guided by developing theory | Grounded-theory research |
| Maximum-variation sampling | Diverse cases selected across relevant characteristics | Exploring variation and common patterns |
Nonprobability sampling is not automatically inferior. It may be appropriate when the aim is conceptual depth, case comparison, specialist knowledge, theory development, or access to a difficult-to-identify population. However, researchers should not make probability-based population claims from a convenience sample without a defensible basis.
How is sample size determined?
There is no universal sample-size rule.
For quantitative research, sample-size planning may depend on:
- Expected effect or clinically meaningful difference
- Desired precision
- Statistical power
- Significance threshold
- Outcome variability
- Number of predictors
- Group allocation
- Clustering or repeated measures
- Expected attrition or nonresponse
- Planned statistical model
For qualitative research, adequacy depends on the specificity of the sample, richness of the evidence, complexity of the question, diversity of experiences, analytical strategy, and depth of engagement. “Saturation” should not be treated as an unexplained numerical rule. Researchers should describe what they considered sufficiently developed and how that judgment was reached.
For case studies, archival research, and evidence synthesis, selection is justified through case boundaries, inclusion criteria, search coverage, theoretical relevance, or information needs rather than one universal formula.
Data-Analysis Methods
Quantitative analysis
Common quantitative procedures include:
- Frequencies, percentages, means, medians, and measures of dispersion
- Confidence intervals
- t tests and analysis of variance
- Nonparametric comparisons
- Correlation
- Linear, logistic, count, and survival regression
- Multilevel and repeated-measures models
- Time-series analysis
- Structural equation models
- Classification and prediction models
- Sensitivity and robustness analysis
The statistical test should follow from the design, measurement level, dependency structure, distribution, assumptions, and intended inference. Researchers should not select a test merely because it produces a small p value.
A p value does not measure the probability that a hypothesis is true, and statistical significance does not establish practical importance, causality, or absence of bias. Effect estimates, uncertainty intervals, assumptions, missing-data decisions, and substantive interpretation should be reported.
Qualitative analysis
Common qualitative approaches include:
| Analytical approach | Principal purpose |
|---|---|
| Thematic analysis | Identify and interpret patterns of meaning |
| Qualitative content analysis | Systematically categorize content, sometimes with descriptive counting |
| Grounded-theory analysis | Develop an explanatory theory through comparison and category development |
| Discourse analysis | Examine how language constructs meanings, identities, or social realities |
| Narrative analysis | Study how experiences and identities are organized through stories |
| Interpretative phenomenological analysis | Examine detailed personal meaning and lived experience |
| Framework analysis | Organize evidence in a structured matrix, often for applied research |
| Conversation analysis | Examine the detailed organization of interaction |
Coding is not the complete analysis. Codes must be interpreted and connected to categories, themes, mechanisms, concepts, or claims according to the selected analytical approach.
Researchers should explain who conducted the analysis, how interpretations developed, how disagreements or alternative readings were considered, and how quoted or observed evidence supports the findings.
Mixed-methods integration
Mixed-methods analysis has two requirements:
- Analyze each evidence type using suitable procedures.
- Integrate the findings in relation to the overall question.
Integration can occur by:
- Connecting: Results from one phase guide sampling or data collection in another.
- Building: One phase develops instruments or hypotheses for the next.
- Merging: Quantitative and qualitative findings are compared or combined.
- Embedding: One evidence type is included within another design.
- Joint display: Results are organized in a table or figure showing convergence, complementarity, or contradiction.
Conflicting findings should be investigated rather than hidden. Disagreement may reveal measurement limitations, subgroup differences, contextual variation, or genuinely different dimensions of the phenomenon.
How to Choose a Research Method
Choose a research method by starting with the research question and the conclusion you need to support. Then align the design, sample, measurement, data collection, analysis, ethics, and available resources.
Step 1: Define the research problem
State what is unknown, uncertain, contested, or practically important. A broad topic such as “social media and students” is not yet a research problem.
A clearer problem might be:
Universities lack evidence about how first-year students perceive the influence of course-based social-media groups on belonging and academic help-seeking.
Step 2: Write an answerable question
Different questions require different evidence.
| Question form | Likely methodological direction |
|---|---|
| What proportion or how frequently? | Descriptive quantitative study |
| Is there a measurable difference? | Comparative quantitative design |
| Are variables associated? | Correlational or analytical observational design |
| Did an intervention cause an outcome? | Experimental or strong quasi-experimental design |
| How do people experience something? | Qualitative interview, observation, or phenomenological design |
| How does a process operate in context? | Case study, ethnography, process evaluation, or qualitative design |
| What is known across existing studies? | Systematic, scoping, or other evidence review |
| What occurred, and why? | Mixed-methods or multi-method design |
This table is a starting point, not an automatic classifier.
Step 3: Specify the intended claim
Decide whether the study aims to:
- Describe
- Compare
- Estimate
- Explore
- Interpret
- Predict
- Explain
- Evaluate
- Develop theory
- Support a causal claim
The method must be capable of supporting that type of claim. A cross-sectional correlation may identify an association but usually cannot, by itself, establish the direction or cause of the relationship.
Step 4: Identify the required evidence
Determine:
- Which population, setting, case, or body of literature is relevant?
- What information would count as evidence?
- Are existing records sufficient?
- Must participants describe experiences in their own words?
- Is an intervention involved?
- Is change over time important?
- Are direct observations needed?
Step 5: Select the design and sampling strategy
Choose a design that links the evidence to the intended conclusion. Then select participants, cases, or materials using a sampling strategy that matches the study’s purpose.
A nationally representative prevalence estimate and an in-depth study of specialist experiences require different forms of sampling.
Step 6: Define concepts and measures
Explain how key concepts will be observed, measured, or interpreted.
For a construct such as “academic stress,” researchers might use:
- A validated scale
- Physiological indicators
- Interview accounts
- Observed behavior
- Administrative records
- A combination of evidence
Each option captures different dimensions and carries different limitations.
Step 7: Plan the analysis before collecting evidence
A useful analysis plan identifies:
- Variables, outcomes, or qualitative focus
- Data-cleaning procedures
- Inclusion and exclusion decisions
- Statistical models or qualitative analytical approach
- Treatment of missing data
- Subgroup or sensitivity analyses
- Integration procedures for mixed methods
- Software and version-control procedures where relevant
Early planning reduces the temptation to choose analyses only after seeing favorable results.
Step 8: Evaluate ethics and feasibility
Consider:
- Participant burden and risk
- Informed consent
- Privacy and confidentiality
- Data security
- Recruitment access
- Researcher safety
- Time and budget
- Required skills
- Cultural and linguistic appropriateness
- Institutional or regulatory review
- Potential effects on communities
A technically impressive method is unsuitable if it is unethical, inaccessible, or impossible to execute well.
Step 9: Pilot and revise
Pilot studies can test:
- Recruitment
- Question wording
- Interview prompts
- Measurement procedures
- Technical systems
- Timing
- Data quality
- Analysis workflows
A pilot does not guarantee success, but it can expose avoidable problems before the main study.
Research-Method Alignment Test
Before finalizing a study, ask:
- Does the question identify what the study will investigate?
- Can the design support the intended type of conclusion?
- Does the sample provide relevant evidence?
- Are the concepts measured or explored appropriately?
- Does the collection method produce the evidence required?
- Does the analysis answer the stated question?
- Have plausible biases and alternative explanations been considered?
- Are ethical protections built into each stage?
- Are the limitations consistent with the claims?
- Could another researcher understand what was done?
If any answer is no, the study may be methodologically misaligned.
Worked Examples
Example 1: Quantitative education study
Question: Is a structured formative-feedback program associated with improved first-year examination performance?
A defensible study might use:
- A randomized experiment where feasible, or a justified quasi-experimental design.
- A defined population of eligible first-year students.
- Baseline and follow-up achievement measures.
- Consistent implementation monitoring.
- Analysis accounting for initial performance, clustering, attrition, and uncertainty.
- A preregistered primary outcome and analysis plan where appropriate.
A simple comparison of final averages may be misleading if groups differed before the intervention or if participation was self-selected.
Example 2: Qualitative health study
Question: How do rural patients experience follow-up consultations delivered through telehealth?
Possible methods include:
- Purposive sampling across age, location, digital access, and health needs.
- Semi-structured interviews.
- Reflexive thematic analysis.
- Documentation of the researchers’ assumptions and relationships to the topic.
- Attention to negative or divergent cases.
- Secure handling of health-related information.
The goal is not to estimate the percentage of all rural patients holding each view. It is to develop a contextualized interpretation of experiences and variation.
Example 3: Mixed-methods workplace study
Question: How is hybrid working related to employee retention intentions, and how do employees explain that relationship?
An explanatory sequential design might include:
- A survey measuring hybrid-work arrangements, job characteristics, and retention intentions.
- Statistical analysis of relationships and subgroup differences.
- Interview sampling informed by the survey findings.
- Interviews exploring autonomy, isolation, supervision, commuting, and work-life boundaries.
- A joint display integrating numerical patterns and qualitative explanations.
The interviews should not be treated as decorative quotations. Their purpose is to explain, complicate, or challenge the quantitative findings.
Advantages and Limitations of Major Approaches
Quantitative methods
Advantages:
- Can estimate frequencies, differences, and relationships.
- Support formal uncertainty assessment.
- Can analyze large datasets.
- May support population estimates when sampling is appropriate.
- Can evaluate specified hypotheses and models.
Limitations:
- Measures may inadequately represent complex concepts.
- Important context can be lost.
- Large samples do not correct poor measurement or biased selection.
- Statistical associations can be mistaken for causal effects.
- Analytical flexibility can produce misleading conclusions if undisclosed.
Qualitative methods
Advantages:
- Provide contextual and detailed evidence.
- Explore meanings, experiences, and processes.
- Can investigate poorly understood topics.
- Allow unexpected issues to emerge.
- Help explain how and why outcomes differ.
Limitations:
- Do not normally estimate population prevalence.
- Interpretation is influenced by context and researcher standpoint.
- Data collection and analysis can be time-intensive.
- Superficial coding may remove meaning from context.
- Readers need sufficient evidence to evaluate the interpretation.
Mixed methods
Advantages:
- Connects measurement with explanation.
- Can address several dimensions of a problem.
- Allows one method to inform another.
- Can reveal convergence and contradiction.
- Useful for complex interventions and implementation questions.
Limitations:
- Requires competence in more than one tradition.
- Integration may be weak or omitted.
- Designs can become too ambitious for the available resources.
- Conflicting findings require careful interpretation.
- A weak component can undermine the overall conclusion.
How Is Research Quality Evaluated?
Quality criteria should match the study’s design and methodological assumptions.
Quantitative quality
Important considerations include:
- Measurement reliability: Are measurements sufficiently consistent for their intended use?
- Construct validity: Does the measure represent the intended concept?
- Internal validity: Are alternative explanations for the observed effect adequately addressed?
- External validity: To what settings, populations, conditions, or periods might the findings apply?
- Statistical-conclusion validity: Are the analysis, assumptions, precision, and interpretation appropriate?
- Reproducibility: Can the computational procedures be understood and repeated using the relevant materials?
Reliability is necessary for many forms of measurement, but a measure can be consistent without measuring the intended construct.
Qualitative trustworthiness
Common considerations include:
- Credibility: Is the interpretation convincingly grounded in suitable evidence?
- Dependability: Is the research process transparent and logically documented?
- Confirmability: Can readers trace interpretations to the evidence and analytical process?
- Transferability: Is sufficient contextual detail provided for readers to judge relevance elsewhere?
- Reflexivity: Has the researcher examined how their assumptions, identity, role, and relationships influenced the study?
Techniques may include prolonged engagement, triangulation, reflexive notes, transparent coding decisions, examination of divergent cases, participant engagement, or peer discussion. None should be treated as a mechanical guarantee of quality.
Mixed-methods quality
Mixed-methods quality requires:
- A clear reason for combining approaches.
- Appropriate quality within each component.
- A documented point of integration.
- An explanation of how findings relate.
- Attention to inconsistencies.
- Conclusions that genuinely draw on the integrated evidence.
Research Ethics and Governance
Research ethics protects participants, communities, researchers, and the integrity of knowledge.
The Belmont Report identifies respect for persons, beneficence, and justice as foundational principles for human-subject research. These principles inform issues such as voluntary participation, risk minimization, fair participant selection, and appropriate consent.
Researchers should consider:
- Whether consent is informed, voluntary, and understandable.
- Whether the potential benefits justify the burdens and risks.
- Whether recruitment is fair.
- Whether vulnerable or dependent participants require additional safeguards.
- How privacy and confidentiality will be protected.
- Whether identities could be inferred from combined data.
- How unexpected findings or distress will be managed.
- Whether incentives could become undue influence.
- How conflicts of interest will be disclosed.
- How communities and stakeholders may be affected by publication.
In the United States, some human-subject research falls under the Common Rule and related institutional requirements. In the United Kingdom and other jurisdictions, applicable ethics, health-research, data-protection, professional, and institutional arrangements differ. Researchers should consult their institution rather than assuming one rule applies everywhere.
Approval from an Institutional Review Board or Research Ethics Committee does not transfer all ethical responsibility to the committee. Researchers remain responsible for following the approved protocol, reporting relevant changes, protecting data, and responding appropriately to unforeseen problems.
Research Methods in Modern Research
Open and transparent research
Open research can include:
- Preregistering questions and analysis plans.
- Using registered reports.
- Sharing protocols and materials.
- Maintaining reproducible analysis scripts.
- Providing data-access statements.
- Sharing data where lawful and ethical.
- Reporting deviations from planned procedures.
- Publishing negative or inconclusive findings.
- Using appropriate reporting guidelines.
Preregistration records a research plan before the relevant data are collected or analyzed. It helps readers distinguish planned confirmatory analysis from later exploratory analysis. It does not prevent justified changes, but changes should be transparently documented.
Open research does not mean placing every dataset online. Sensitive, confidential, licensed, culturally restricted, or re-identifiable evidence may need controlled access or may not be shareable. A useful principle is to make research as open as possible and as restricted as necessary.
Data management
A data-management plan should address:
- Data formats and documentation
- File naming and organization
- Storage and backup
- Access control
- Consent and permitted reuse
- De-identification or pseudonymization
- Retention and secure destruction
- Version control
- Metadata
- Repository selection
- Licensing
- Long-term preservation
The FAIR principles encourage digital research objects to be findable, accessible, interoperable, and reusable. “Accessible” does not always mean openly downloadable; metadata and controlled-access procedures may be appropriate.
Reporting guidelines
Reporting guidelines help authors describe research transparently. Examples include:
- CONSORT for randomized trials
- STROBE for observational studies
- PRISMA for systematic reviews
- COREQ and SRQR for qualitative studies
- SPIRIT for trial protocols
- ARRIVE for animal research
- SQUIRE for quality-improvement studies
A reporting guideline does not repair a weak design. It helps researchers report what was actually planned and conducted so readers can evaluate it.
Digital Research Tools
| Research task | Example tools | Important caution |
|---|---|---|
| Reference management | Zotero, EndNote, Mendeley | Check imported metadata and citation formatting |
| Surveys | Qualtrics, REDCap, LimeSurvey | Protect personal data and test accessibility |
| Statistical analysis | R, Python, SPSS, Stata, SAS | Document code, assumptions, versions, and decisions |
| Qualitative analysis | NVivo, ATLAS.ti, MAXQDA, Taguette | Software organizes evidence but does not perform interpretation responsibly on its own |
| Evidence screening | Rayyan, Covidence | Screening decisions still require explicit eligibility criteria |
| Open workflows | OSF and institutional repositories | Confirm consent, licensing, and disclosure risk before sharing |
| Reproducible writing | Quarto, R Markdown, Jupyter | Preserve dependencies and computational environments |
| Version control | Git and approved repository platforms | Avoid exposing confidential files, keys, or identifiable data |
Software should follow the methodological plan. Selecting software first and then designing a study around its available buttons reverses the proper order.
Can Artificial Intelligence Be Used in Research?
AI can assist with selected research tasks, but researchers remain responsible for the accuracy, legality, ethics, originality, and transparency of the work.
Potential uses include:
- Generating provisional search terms
- Supporting code development
- Checking data formats
- Assisting transcription where consent and security permit
- Translating material for preliminary review
- Organizing notes
- Exploring alternative interpretations
- Editing researcher-written text
- Producing synthetic examples for training or testing
AI output should not be treated as evidence merely because it sounds plausible.
Researchers should:
- Verify every factual statement, quotation, citation, and calculation.
- Avoid uploading confidential, identifiable, copyrighted, embargoed, or culturally restricted material to unapproved systems.
- Review potential bias and unequal performance across groups or languages.
- Maintain records of consequential AI-assisted procedures.
- Disclose AI use when required by institutions, funders, publishers, or disciplinary standards.
- Ensure that a human researcher makes and can defend methodological and interpretive decisions.
- Never list an AI system as an author or use it to fabricate data, participants, references, ethics approvals, or findings.
- Confirm current local guidance because institutional and publication policies continue to develop.
AI may increase efficiency, but it does not possess responsibility for the research. The researcher cannot delegate accountability to a tool.
How to Write a Research Methods Section
A methods section should enable readers to understand how the evidence was produced and evaluate whether the conclusions are justified.
1. Research design and rationale
State:
- The approach and design.
- The research setting.
- Why the design was suitable for the question.
- Whether the study was exploratory, descriptive, explanatory, evaluative, or confirmatory.
2. Participants, cases, or sources
Report:
- Target population or case boundaries.
- Inclusion and exclusion criteria.
- Recruitment or source-identification procedures.
- Sampling strategy.
- Sample-size rationale.
- Relevant participant or case characteristics.
- Response, attrition, or exclusion information.
3. Materials, measures, or instruments
Describe:
- Questionnaires and scales.
- Interview or observation guides.
- Equipment and software.
- Operational definitions.
- Reliability and validity evidence where relevant.
- Translation or adaptation procedures.
- Pilot testing.
4. Procedure
Explain what occurred and in what order, including:
- Recruitment
- Consent
- Group allocation where applicable
- Data collection
- Duration and timing
- Intervention delivery
- Follow-up
- Quality-control procedures
5. Data analysis
Specify:
- Data preparation and exclusions.
- Statistical models or qualitative analytical approach.
- Assumption checks.
- Missing-data procedures.
- Coding and interpretation process.
- Software and versions.
- Integration procedures for mixed methods.
- Sensitivity or robustness analyses where applicable.
6. Ethics and transparency
Report:
- Ethics review or the institutional basis for an exemption or nonreview determination.
- Consent procedures.
- Privacy and data-security protections.
- Registration or preregistration.
- Data, material, and code availability.
- Deviations from the original plan.
- Relevant conflicts of interest.
Copyable methods-section template
This study used a [approach] [design] to investigate [research question] in [setting]. The design was selected because [methodological rationale].
Participants/cases were selected using [sampling method] from [population or source]. Eligibility criteria included [criteria]. The final sample comprised [number and relevant characteristics], justified by [power, precision, information needs, case logic, or other rationale].
Evidence was collected using [instruments or procedures]. [Describe measurement quality, piloting, timing, and implementation.]
Data were analyzed using [statistical or qualitative approach]. [Describe preparation, coding, models, assumptions, missing-data procedures, and mixed-methods integration where applicable.]
Ethical approval or review was provided by [body and reference, where appropriate]. Participants [describe consent], and data were protected through [security and confidentiality measures]. [State registration, materials, data, code, and disclosure information.]
The template should be adapted to the discipline and reporting standard. It should not be completed with vague phrases merely to create the appearance of rigor.
Common Research-Method Mistakes
Choosing the method before defining the question
Familiarity with a survey, interview platform, or statistical package is not a sufficient methodological reason.
Confusing random sampling with random assignment
Random sampling concerns selection from a population. Random assignment concerns placement into study conditions. They serve different purposes.
Making causal claims from simple associations
Correlation, prediction, and causation are not interchangeable. Causal interpretation requires a design and assumptions capable of addressing competing explanations.
Treating convenience as representativeness
A large convenience sample can still be systematically different from the target population.
Using unvalidated measures without justification
A questionnaire created quickly for one project may not measure the intended construct reliably or validly.
Treating qualitative coding as automatic
Software, code frequencies, or AI-generated categories do not replace interpretive reasoning, contextual understanding, and reflexivity.
Claiming saturation without explanation
Researchers should describe what form of saturation or adequacy they considered, how it was assessed, and why the evidence was sufficient for the analytical goal.
Collecting two evidence types without integrating them
A project is not meaningfully mixed methods when the qualitative and quantitative components never inform one another or the final conclusion.
Reporting only favorable analyses
Undisclosed exclusions, outcome changes, repeated testing, and selective reporting can distort findings.
Treating ethics as a one-time form
Consent, privacy, risk, community effects, data security, and responsible reporting require attention throughout the study.
Conclusion
Research methods are the practical and analytical procedures through which research questions are answered. Strong research does not begin by selecting a fashionable tool. It begins by defining the problem, identifying the intended claim, and aligning the design, sample, measurement, collection, analysis, ethics, and interpretation.
Quantitative, qualitative, and mixed methods each serve different purposes. The most appropriate method is the one that produces relevant evidence, respects the study context and participants, addresses plausible sources of error, and supports conclusions no stronger than the design allows.
References
- Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77–101. https://doi.org/10.1191/1478088706qp063oa
- Center for Open Science. (n.d.). Preregistration.
- Creswell, J. W., & Plano Clark, V. L. (2018). Designing and conducting mixed methods research (3rd ed.). SAGE.
- Jhangiani, R. S., Chiang, I.-C. A., Cuttler, C., & Leighton, D. C. (2019). Research methods in psychology (4th ed.). Kwantlen Polytechnic University. https://doi.org/10.17605/OSF.IO/HF7DQ
- Lincoln, Y. S., & Guba, E. G. (1985). Naturalistic inquiry. SAGE.
- Malterud, K., Siersma, V. D., & Guassora, A. D. (2016). Sample size in qualitative interview studies: Guided by information power. Qualitative Health Research, 26(13), 1753–1760. https://doi.org/10.1177/1049732315617444
- National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research. (1979). The Belmont report. U.S. Department of Health and Human Services.
- Nosek, B. A., et al. (2015). Promoting an open research culture. Science, 348(6242), 1422–1425. https://doi.org/10.1126/science.aab2374
- Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and quasi-experimental designs for generalized causal inference. Houghton Mifflin.
- UK Research and Innovation. (2026). Open research.
- UK Research Integrity Office. (2025). Code of practice for research (Version 3.5).
- Wilkinson, M. D., et al. (2016). The FAIR Guiding Principles for scientific data management and stewardship. Scientific Data, 3, Article 160018. https://doi.org/10.1038/sdata.2016.18
