A dissertation methodology is the chapter that explains and justifies how a study was designed, how data were selected or collected and analysed, and how quality, ethics, and limitations were handled. Its purpose is to show that the methods fit the research question and that the findings can be evaluated.
Introduction
The methodology chapter is where you demonstrate that your research process was logical, ethical, transparent, and appropriate for the problem you investigated.
A strong methodology does more than list a survey, interview, experiment, dataset, or statistical test. It explains why those choices were suitable, how they were implemented, what assumptions they involved, and what their limitations mean for the conclusions.
This guide explains how to plan and write a dissertation methodology for qualitative, quantitative, mixed-methods, secondary-data, review-based, humanities, and computational research. It also covers sample-size justification, research quality, data management, reporting standards, digital tools, and responsible use of artificial intelligence.
Key Takeaways
- A methodology chapter explains both what you did and why the approach was appropriate.
- Every major methodological choice should connect to a research question or objective.
- There is no universal methodology structure; discipline, design, degree level, and institutional guidance matter.
- Sampling, data collection, analysis, quality procedures, ethics, and limitations should be reported transparently.
- Quantitative validity, qualitative trustworthiness, mixed-methods integration, and review quality require different forms of justification.
- AI tools may assist limited parts of the workflow, but the researcher remains responsible for accuracy, confidentiality, disclosure, and methodological reasoning.
What Is a Dissertation Methodology?
A dissertation methodology is a reasoned account of the strategy, design, methods, procedures, and analytical techniques used to answer a study’s research questions. It allows readers to judge whether the research process was suitable, rigorous, ethical, and sufficiently transparent.
The exact content depends on the discipline and project. An experimental psychology dissertation may explain randomisation, measures, statistical power, and regression assumptions. A qualitative education study may discuss an interpretivist position, purposive sampling, interviews, reflexive thematic analysis, and researcher positionality. A history dissertation may justify its archive selection, source criticism, periodisation, and interpretive framework.
The central principle is the same: the chapter should make the logic of the investigation visible.
What should a dissertation methodology include?
Most empirical methodology chapters address:
- Research aims or questions.
- Research philosophy or theoretical orientation, when relevant.
- Research approach and design.
- Setting, population, cases, texts, or datasets.
- Sampling or source-selection strategy.
- Data collection or evidence-generation procedures.
- Instruments, materials, or technical systems.
- Data preparation and analysis.
- Quality, rigor, or validation procedures.
- Ethical and data-management considerations.
- Methodological limitations.
- A brief chapter summary.
Not every dissertation needs these sections as separate headings. The order and emphasis should follow the logic of the project rather than a rigid template.
Methodology vs. Methods
Methods are the specific procedures used to generate or analyse evidence. Methodology is the reasoning that connects those procedures to the research question, assumptions, and overall design.
Examples of methods include:
- Interviews
- Questionnaires
- Experiments
- Observations
- Archival searches
- Statistical tests
- Thematic analysis
- Simulation
- Machine-learning evaluation
- Legal or textual analysis
Methodology addresses broader questions:
- What kind of knowledge is the study trying to produce?
- Why is this design appropriate?
- Why were these participants, cases, sources, or variables selected?
- What can the chosen evidence establish?
- What assumptions does the analysis make?
- What alternative approaches were considered?
- How were quality and ethics protected?
The distinction is not used identically in every discipline. Some scientific departments use “methods” for the entire chapter, while some social-science departments expect an explicit methodology or research-philosophy discussion. Follow the terminology in your dissertation handbook and recent successful dissertations in your field.
What Is the Purpose of the Methodology Chapter?
The methodology chapter has four main purposes.
1. To demonstrate alignment
The design should answer the research questions. A causal question may require an experimental or strong quasi-experimental design. A question about lived experience may require qualitative evidence. A question about prevalence may require an appropriate sampling frame and quantitative estimates.
2. To justify decisions
Readers need to understand why one approach was chosen over realistic alternatives. “Interviews were convenient” is not a methodological justification. A stronger explanation would show that interviews were appropriate because the study required detailed accounts of participants’ interpretations and experiences.
3. To support evaluation and reproducibility
The chapter should contain enough detail for a knowledgeable reader to evaluate the process and, where the research tradition permits, reproduce or meaningfully compare the procedure.
Exact replication may not be the goal of every qualitative or interpretive study. Nevertheless, transparent reporting allows readers to understand how evidence was generated and how interpretations were developed.
4. To define the boundaries of the conclusions
Methods determine what a study can reasonably claim. A cross-sectional survey may identify associations but cannot by itself demonstrate temporal order or causation. A small purposive interview sample can provide depth but may not support statistical generalisation.
A credible methodology states these boundaries openly.
Where Does the Methodology Chapter Go?
In a conventional five-chapter empirical dissertation, methodology is usually Chapter 3:
- Introduction
- Literature review
- Methodology
- Results or findings
- Discussion and conclusion
This arrangement is common but not universal.
A dissertation containing several experiments may describe methods separately for each study. An article-based thesis may include a general methodology chapter plus study-specific methods. Humanities research may integrate methodology into the introduction. Creative-practice research may combine contextual, reflective, and practice-based methods across chapters.
What tense should a methodology use?
Use the tense that matches the stage of the project:
- Proposal before data collection: future tense — “Interviews will be conducted.”
- Completed dissertation: past tense — “Interviews were conducted.”
- Established knowledge or current methodological literature: present tense — “Reflexive thematic analysis treats the researcher as an active interpreter.”
Do not mechanically convert every sentence to past tense. Methodological principles and published arguments can remain in the present tense.
How long should the methodology chapter be?
There is no universal word count or fixed percentage.
Length depends on:
- Degree level
- Total dissertation length
- Number of studies
- Complexity of the design
- Disciplinary convention
- Novelty of the methods
- Ethical complexity
- Institutional requirements
The chapter is long enough when every consequential decision is explained without turning into a general research-methods textbook.
Check your handbook before following an online word-count estimate. A 10,000-word undergraduate project and an 80,000-word doctoral thesis cannot use the same methodology allowance.
Prepare Before Writing the Methodology
Writing is easier when the research process has been documented from the beginning.
Check institutional requirements
Review:
- Dissertation handbook
- Marking rubric
- Ethics approval conditions
- Data-management requirements
- Formatting rules
- Expected chapter structure
- Examples held in the institutional repository
Institutional guidance takes priority over a generic online template.
Build a research alignment matrix
An alignment matrix reveals whether each research question is supported by appropriate evidence.
| Research question | Evidence needed | Source or sample | Collection method | Analysis | Quality check |
|---|---|---|---|---|---|
| What factors predict student retention? | Numerical measures of retention and predictors | Student records or representative sample | Administrative dataset | Logistic regression | Model diagnostics and sensitivity analysis |
| How do first-generation students experience transition? | Detailed accounts and meanings | Purposively selected students | Semi-structured interviews | Reflexive thematic analysis | Reflexivity, rich extracts, transparent analytic process |
| Why do statistical and experiential findings differ? | Integrated quantitative and qualitative evidence | Participants from both phases | Sequential mixed methods | Joint display and integrated interpretation | Integration rationale and analysis of divergence |
Every method should have a visible purpose. A method that does not help answer a research question may not belong in the study.
Keep an audit record
During the project, record:
- Protocol versions
- Ethics amendments
- Recruitment dates
- Inclusion and exclusion decisions
- Instrument revisions
- Pilot-test outcomes
- Software and package versions
- Data-cleaning rules
- Coding-framework changes
- Deviations from the original plan
- AI tools used and their role, where relevant
These records reduce reliance on memory when the chapter is written.
How to Write a Dissertation Methodology
Step 1: Restate the research aim and questions
Begin with a concise reminder of what the study investigated.
Do not repeat the entire introduction. State enough to establish the link between the research problem and the methodological choices.
Example:
This study examined how remote-working arrangements influenced employee collaboration in small technology firms. It addressed two questions: whether the frequency of remote work was associated with reported collaboration quality and how employees explained the advantages and difficulties of remote collaboration.
This opening prepares the reader for a mixed design involving numerical measurement and qualitative explanation.
Step 2: Explain the research philosophy when it matters
A research philosophy describes assumptions about reality, knowledge, evidence, and the relationship between the researcher and the subject of inquiry.
Common positions include:
- Positivism or postpositivism: often associated with measurement, hypothesis testing, and attempts to estimate relationships while recognizing uncertainty and possible error.
- Interpretivism or constructivism: focuses on how people construct meanings within social, cultural, or historical contexts.
- Critical approaches: examine power, inequality, ideology, and the possibility of transformative research.
- Pragmatism: selects and integrates approaches according to what best addresses the research problem.
- Realism: accepts a reality independent of individual perception while recognizing that knowledge of it may be partial or mediated.
Do not add a lengthy philosophy section merely because an online template includes one. In some laboratory, engineering, mathematical, or computing disciplines, the assumptions may be better explained through the design, model, or validation strategy.
Where philosophy is required, connect it to practical decisions.
Weak explanation:
The study used interpretivism because interpretivism is common in qualitative research.
Stronger explanation:
An interpretivist orientation was appropriate because the study examined how teachers understood and negotiated a new assessment policy. The purpose was not to estimate a single objective effect but to analyse context-dependent meanings constructed through professional experience.
Step 3: Identify the research approach
The research approach explains how theory and evidence relate.
Deductive approach
A deductive study begins with theory or hypotheses and examines whether observations support the proposed relationships.
Typical sequence:
- Develop theory or hypotheses.
- Operationalise concepts.
- Collect relevant data.
- Test the hypotheses.
- Evaluate or refine the theory.
Inductive approach
An inductive study develops concepts or explanations from patterns in the evidence rather than beginning with a tightly specified hypothesis.
Abductive approach
Abduction moves iteratively between theory and unexpected evidence to develop the most plausible explanation. It is useful when existing theory does not fully explain the observations.
Avoid claiming that all quantitative studies are deductive or all qualitative studies are inductive. Either type of data can be used within different reasoning strategies.
Step 4: Name and justify the research design
The research design is the overall plan linking the question, evidence, collection process, and analysis.
Possible designs include:
- Experiment
- Quasi-experiment
- Cross-sectional survey
- Longitudinal study
- Case study
- Cohort study
- Ethnography
- Phenomenology
- Grounded theory
- Action research
- Evaluation research
- Design science
- Comparative historical research
- Corpus study
- Systematic review
- Scoping review
- Simulation study
- Mixed-methods design
State the design precisely.
“Quantitative methodology” is not a complete design. “A cross-sectional correlational survey” is more informative.
Then explain:
- Why the design fits the question.
- What the design can establish.
- Why plausible alternatives were less appropriate.
- Whether the study is exploratory, descriptive, explanatory, evaluative, or predictive.
- Whether data were collected at one time or over time.
- Whether intervention allocation, comparison groups, or controls were used.
Step 5: Describe the setting, population, cases, or source universe
Specify where the research took place and what body of people, documents, events, systems, or records the study concerns.
Depending on the project, report:
- Country, institution, organization, community, archive, platform, laboratory, or database.
- Relevant dates.
- Target population.
- Unit of analysis.
- Eligibility criteria.
- Case boundaries.
- Access arrangements.
- Context that affects interpretation.
Protect confidentiality. Do not name a participating organization if the ethics agreement requires anonymity.
Step 6: Explain sampling and sample-size justification
Describe how participants, cases, documents, records, or observations were selected.
Probability sampling
Probability methods give eligible units a known probability of selection. Examples include:
- Simple random sampling
- Systematic sampling
- Stratified sampling
- Cluster sampling
- Multistage sampling
Explain the sampling frame, selection procedure, strata or clusters, expected response rate, and any weighting.
Non-probability sampling
Examples include:
- Purposive sampling
- Convenience sampling
- Quota sampling
- Snowball or chain-referral sampling
- Theoretical sampling
- Criterion sampling
- Maximum-variation sampling
Explain why the selected strategy fits the research purpose and acknowledge its consequences for transferability or generalisation.
Sample-size justification
Do not write only that the sample was “large enough” or copied from a previous dissertation.
For quantitative research, possible rationales include:
- A priori power analysis
- Desired precision or confidence-interval width
- A census or near-census
- A smallest effect size of interest
- Model complexity and expected event counts
- Resource constraints accompanied by sensitivity analysis
- The full available eligible dataset
For a simple estimate of a population proportion, an initial precision-based calculation may use:
[
n_0 = \frac{Z^2p(1-p)}{e^2}
]
where (Z) is the selected normal critical value, (p) is the anticipated proportion, and (e) is the desired margin of error.
This formula is not appropriate for every survey or analysis. Complex sampling, finite populations, clustering, nonresponse, subgroup analysis, regression, repeated measures, and non-normal outcomes require additional considerations.
For qualitative interviews, justify the sample using the study aim, participant specificity, theoretical framework, quality of dialogue, analytic strategy, information power, or an appropriate concept of saturation rather than a universal numerical rule (Malterud et al., 2016).
For mixed methods, justify the sample for each strand and explain how participants or cases connect across phases.
Step 7: Describe data collection or evidence generation
Report exactly how the evidence was obtained.
For interviews or focus groups, include:
- Interview format
- Topic-guide development
- Mode and location
- Duration
- Recording
- Transcription
- Language and translation
- Interviewer role
- Field notes
- Whether repeat interviews occurred
For surveys, include:
- Administration mode
- Recruitment and reminders
- Question order
- Response formats
- Validated scales
- Adaptations
- Pilot testing
- Accessibility
- Response-quality checks
For experiments, include:
- Conditions
- Randomisation or allocation
- Blinding, where applicable
- Apparatus and materials
- Intervention or stimulus
- Outcome measurement
- Timing
- Controls
- Protocol deviations
For documentary or archival research, include:
- Repositories and collections
- Search dates
- Selection criteria
- Source provenance
- Authenticity and completeness checks
- Languages
- Excluded material
- Source criticism
For computational research, include:
- Hardware and software environment
- Dataset source
- Preprocessing
- Feature construction
- Training, validation, and test splits
- Baselines
- Hyperparameter selection
- Evaluation metrics
- Random seeds
- Repeated runs
- Repository or code availability
Step 8: Explain instruments, measures, and pilot testing
Name the tools used to generate evidence.
These may include:
- Questionnaire scales
- Interview guides
- Observation protocols
- Laboratory instruments
- Sensors
- Coding manuals
- Extraction forms
- Rubrics
- Software systems
- Algorithms
For established measures, cite their original development and relevant validation evidence. Explain any adaptation, translation, scoring change, or shortened version.
For a new instrument, describe:
- How items were generated.
- How content was reviewed.
- Whether cognitive interviewing or expert assessment was used.
- How it was piloted.
- What revisions followed.
- How reliability or validity evidence was assessed.
Pilot testing is not simply proof that a questionnaire “worked.” Report what was tested and what changed.
Step 9: Describe data preparation and analysis
The analysis section should be specific enough for a knowledgeable reader to understand how raw evidence became findings.
Quantitative analysis
Report:
- Software and version
- Coding and variable construction
- Scoring procedures
- Exclusions
- Missing-data handling
- Outlier treatment
- Descriptive statistics
- Statistical models or tests
- Assumption checks
- Effect sizes
- Confidence intervals
- Multiple-testing adjustments, when relevant
- Robustness or sensitivity analyses
Do not write only “SPSS was used.” Software is a tool, not an analysis method.
Better example:
Multiple linear regression was used to estimate the association between weekly study time and final assessment score while adjusting for prior attainment and attendance. Residual plots, variance inflation factors, and influence statistics were examined to assess model assumptions and sensitivity to influential observations.
Qualitative analysis
Name the analytic approach and version or tradition.
Examples include:
- Reflexive thematic analysis
- Framework analysis
- Qualitative content analysis
- Grounded-theory coding
- Interpretative phenomenological analysis
- Narrative analysis
- Discourse analysis
- Conversation analysis
For thematic analysis, explain familiarisation, coding, theme development, theme review, definition, and reporting. Clarify whether coding was primarily inductive, deductive, semantic, or latent and how the researcher’s interpretive role was handled. Merely stating that “themes emerged” can hide the analytical decisions made by the researcher (Braun & Clarke, 2006).
Mixed-methods analysis
Describe:
- Analysis of each strand.
- Timing and priority.
- The point at which integration occurred.
- How datasets were connected, merged, embedded, or compared.
- Use of joint displays.
- How agreement, complementarity, and divergence were interpreted.
A mixed-methods study is not complete if the qualitative and quantitative findings remain in separate sections without integration.
Secondary-data analysis
Report:
- Dataset name and custodian.
- Original study purpose.
- Original population and sampling design.
- Data-collection period.
- Access conditions.
- Variables selected.
- Recoding and derived variables.
- Weighting.
- Missingness.
- Linkage.
- Cleaning.
- Disclosure controls.
- Limitations created by using data collected for another purpose.
Systematic or structured review
Describe:
- Protocol or registration.
- Databases and other sources.
- Complete search strategy.
- Search dates.
- Eligibility criteria.
- Deduplication.
- Screening process.
- Data extraction.
- Critical appraisal or risk-of-bias assessment.
- Synthesis method.
- Handling of heterogeneity.
- Certainty assessment, if applicable.
PRISMA 2020 can guide transparent reporting of eligible systematic reviews, but it does not replace the methodological guidance appropriate to the review question (Page et al., 2021).
Step 10: Explain quality, rigor, and validation
Use criteria that match the research tradition.
Quantitative research
Possible considerations include:
- Measurement reliability
- Construct validity
- Internal validity
- External validity
- Statistical conclusion validity
- Calibration
- Model fit
- Robustness
- Sensitivity analysis
- Measurement invariance
- Transparent reporting
Reliability does not prove validity. A measure can produce consistent scores while failing to measure the intended construct.
Qualitative research
Depending on the methodology, quality may involve:
- Credibility
- Dependability
- Confirmability
- Transferability
- Reflexivity
- Contextual depth
- Coherence
- Transparent interpretation
- Negative or deviant case analysis
- Appropriate participant engagement
- An audit trail
Member checking, inter-coder agreement, and triangulation are not mandatory quality tests for every qualitative tradition. Use them only when they are philosophically and analytically appropriate.
Mixed-methods research
Quality also depends on:
- A clear reason for mixing.
- Quality within each component.
- Appropriate timing and priority.
- Meaningful integration.
- Transparent treatment of conflicting findings.
- Defensible mixed-methods inferences.
Computational research
Consider:
- Data leakage
- Benchmark suitability
- Repeated evaluation
- Reproducible environments
- Appropriate baselines
- Error analysis
- Fairness or subgroup performance
- External validation
- Model uncertainty
- Availability of code and configuration details
Step 11: Address ethics and data management
Do not reduce ethics to “participants signed a consent form.”
Explain:
- Ethics or institutional-review approval.
- Approval body and reference number, when permitted.
- Participant information.
- Consent process.
- Voluntary participation.
- Withdrawal arrangements.
- Risks and benefits.
- Compensation or incentives.
- Safeguards for vulnerable participants.
- Confidentiality and anonymity.
- Data access.
- Storage location.
- Encryption.
- Retention.
- Sharing.
- Secure destruction.
- Management of unexpected disclosures.
- Conflicts of interest.
In US research covered by the Common Rule, institutional review and informed-consent requirements may apply. In the UK, ethical consent to participate should not automatically be treated as the same thing as the lawful basis for processing personal data under the UK GDPR.
Use the requirements of your institution, jurisdiction, professional body, funder, and research setting.
Anonymisation vs. pseudonymisation
- Anonymised data can no longer reasonably be linked to an identifiable person.
- Pseudonymised data replace direct identifiers with codes, but re-identification may remain possible through a key or additional information.
Do not promise anonymity if the research team can reconnect responses to participants.
Step 12: Acknowledge limitations and researcher influence
Methodological limitations should show critical awareness, not destroy confidence in the study.
For each important limitation, explain:
- What the limitation is.
- Why it arose.
- How it may affect interpretation.
- What was done to reduce its effect.
- What readers should avoid concluding.
Examples include:
- Selection bias
- Nonresponse
- Measurement error
- Confounding
- Small or unbalanced groups
- Limited geographic scope
- Recall bias
- Researcher influence
- Translation loss
- Missing variables
- Dataset age
- Model instability
- Incomplete archives
Qualitative researchers should also explain relevant positionality and reflexivity. This may include the researcher’s relationship to the setting, assumptions, professional role, insider or outsider status, and influence on collection and interpretation.
Step 13: End with a chapter summary
Conclude by summarising:
- The design.
- The principal evidence sources.
- The collection procedure.
- The analysis.
- The quality and ethical safeguards.
Do not introduce new methodological decisions in the summary.
Dissertation Methodology Template
The following structure can be adapted rather than copied mechanically.
3.1 Chapter introduction
- Restate the aim or questions.
- Identify the overall approach.
- Preview the chapter.
3.2 Research philosophy or theoretical orientation
- State the relevant assumptions.
- Connect them to the question and design.
- Avoid an unrelated textbook history.
3.3 Research approach
- Deductive, inductive, abductive, or iterative.
- Explain the relationship between theory and evidence.
3.4 Research design
- Name the design precisely.
- Explain time horizon, comparison structure, setting, and unit of analysis.
- Justify the design over alternatives.
3.5 Population, cases, sources, or dataset
- Define the target population or evidence universe.
- State eligibility criteria and boundaries.
3.6 Sampling and sample size
- Explain selection.
- Report recruitment or source identification.
- Justify size.
- Report final included numbers and exclusions.
3.7 Data collection or evidence generation
- Describe the procedure chronologically.
- Identify instruments, materials, platforms, and dates.
- Report pilot testing and changes.
3.8 Data preparation and analysis
- Explain cleaning, coding, transformation, modelling, or interpretation.
- Name software and versions.
- Report assumptions and quality checks.
3.9 Quality and rigor
- Address validity, reliability, trustworthiness, integration, robustness, or reproducibility as appropriate.
3.10 Ethics and data management
- Report approval, consent, risk management, confidentiality, storage, access, retention, and sharing.
3.11 Methodological limitations and reflexivity
- Explain important constraints and their consequences.
3.12 Chapter summary
- Summarise the methodological logic and transition to the findings.
Qualitative vs. Quantitative vs. Mixed Methods
| Dimension | Qualitative | Quantitative | Mixed methods |
|---|---|---|---|
| Typical purpose | Explore meanings, experiences, processes, or context | Measure variables, estimate patterns, test hypotheses, or predict outcomes | Combine breadth and depth or explain one form of evidence with another |
| Data | Words, images, observations, documents, interactions | Counts, measurements, scores, categories, or numerical records | Both qualitative and quantitative data |
| Sampling | Often purposive, theoretical, criterion-based, or maximum variation | Often probability-based, census-based, or analytically selected | Separate or connected sampling for each component |
| Common designs | Case study, ethnography, phenomenology, grounded theory, narrative inquiry | Experiment, quasi-experiment, survey, cohort, correlational study | Convergent, explanatory sequential, exploratory sequential, embedded |
| Analysis | Thematic, framework, content, discourse, narrative, grounded-theory analysis | Descriptive statistics, hypothesis tests, regression, multilevel models, machine learning | Separate analyses followed by explicit integration |
| Quality emphasis | Credibility, reflexivity, coherence, contextual depth, transparency | Reliability, validity, precision, bias control, assumptions, robustness | Component quality plus integration and defensible meta-inferences |
| Main limitation risk | Decontextualised coding or unsupported interpretation | Invalid measurement, bias, violated assumptions, or overclaiming causation | Superficial integration or an unmanageable design |
How to choose
Choose the approach that best matches the question.
- Use qualitative research when the question asks how, why, or what an experience means.
- Use quantitative research when the question requires measurement, estimation, comparison, prediction, or hypothesis testing.
- Use mixed methods when integrating different forms of evidence produces an answer that neither component could provide alone.
Do not choose mixed methods merely because it appears more comprehensive. It requires competence, time, and a genuine integration plan.
Methodology Structures by Dissertation Type
Primary empirical dissertation
A primary empirical methodology normally emphasizes:
- Design
- Participants or cases
- Sampling
- Instruments
- Procedure
- Analysis
- Quality
- Ethics
Secondary-data dissertation
A secondary-data methodology should emphasize:
- Dataset provenance
- Original sampling and collection
- Access conditions
- Variables
- Data quality
- Cleaning
- Weighting
- Missingness
- Analysis
- Disclosure risk
- Fitness for the new question
Systematic-review dissertation
A review methodology should emphasize:
- Review question and framework
- Protocol
- Sources
- Search strategy
- Eligibility criteria
- Screening
- Extraction
- Appraisal
- Synthesis
- Reporting flow
Humanities or theoretical dissertation
Relevant sections may include:
- Interpretive or theoretical framework
- Corpus or source selection
- Archive or edition
- Periodisation
- Source criticism
- Close-reading procedure
- Comparative logic
- Translation decisions
- Researcher standpoint
- Limitations of the evidence base
Legal dissertation
Depending on the project, the methodology may explain:
- Doctrinal analysis
- Selection of legislation and cases
- Jurisdiction
- Date boundaries
- Hierarchy of authority
- Comparative framework
- Policy analysis
- Empirical legal methods
- Interpretive principles
Computational or machine-learning dissertation
A computational methodology should explain:
- Dataset
- Licensing and provenance
- Preprocessing
- Model architecture
- Training procedure
- Baselines
- Validation design
- Metrics
- Hyperparameters
- Hardware and software
- Random seeds
- Error analysis
- Reproducibility
- Ethical or fairness considerations
Creative-practice or design dissertation
The methodology may integrate:
- Practice-based inquiry
- Iterative design
- Reflective documentation
- Prototyping
- User research
- Material experimentation
- Evaluation criteria
- Researcher-practitioner position
- Relationship between the artefact and written analysis
How to Justify Sample Size
A strong sample-size paragraph answers four questions:
- What decision rule was used?
- What assumptions informed it?
- What adjustment was made for loss, missingness, clustering, or exclusions?
- What can the achieved sample support?
Quantitative example
An a priori power analysis was conducted for the primary multiple-regression model. The analysis specified the smallest effect considered practically meaningful, the planned number of predictors, the significance threshold, and the desired statistical power. The recruitment target was increased to allow for incomplete responses and exclusions. A sensitivity analysis was also reported for the final analytic sample.
Qualitative example
Sampling was guided by information power rather than a predetermined universal interview total. The study had a focused aim, a relatively specific participant group, and an in-depth interview format. Recruitment continued while additional interviews contributed substantively relevant variation to the developing analysis.
Secondary-data example
The study used all eligible observations in the dataset. The methodology therefore justified the effective analytic sample after eligibility restrictions, missing-data exclusions, and weighting rather than presenting a prospective recruitment calculation.
Reliability, Validity, Trustworthiness, and Rigor
| Research tradition | Questions to answer |
|---|---|
| Quantitative | Were constructs measured appropriately? Were procedures consistent? Were sources of bias controlled? Were model assumptions checked? Are estimates precise and robust? |
| Qualitative | Is the interpretation grounded in the evidence? Is the analytic process transparent and coherent? Is researcher influence addressed? Is sufficient context provided for readers to assess transferability? |
| Mixed methods | Was mixing necessary? Were both components rigorous? Where did integration occur? How were conflicting results handled? |
| Systematic review | Was the search reproducible? Were eligibility decisions transparent? Was appraisal appropriate? Did the synthesis match the evidence? |
| Computational | Were leakage and overfitting controlled? Were baselines fair? Are code, versions, parameters, and evaluation procedures sufficiently documented? |
Quality procedures should be designed into the research rather than added as decorative terminology after the study is complete.
Dissertation Methodology Examples
The following examples are hypothetical and should be adapted to the actual project.
Qualitative methodology example
The study adopted an interpretivist qualitative design to explore how first-generation university students understood their transition into postgraduate education. Participants were selected through purposive maximum-variation sampling to capture differences in discipline, age, mode of study, and prior educational pathway. Semi-structured interviews were recorded, transcribed, and analysed using reflexive thematic analysis. Analysis involved familiarisation, initial coding, development and revision of candidate themes, and production of an interpretive account supported by representative extracts. A reflexive journal documented the researcher’s assumptions and analytical decisions.
Why this works:
- The philosophy matches the aim.
- The sampling has a clear purpose.
- Collection and analysis are named.
- Researcher interpretation is acknowledged.
Quantitative methodology example
A cross-sectional survey design was used to examine associations between academic self-efficacy, weekly study time, and assessment performance among undergraduate students. Participants were selected using stratified random sampling across faculties. Established multi-item measures were scored according to their published guidance. Multiple regression estimated the association between each predictor and performance after adjustment for prior attainment and attendance. Missingness, influential observations, residual patterns, multicollinearity, effect sizes, and confidence intervals were examined.
Why this works:
- The design does not claim causality.
- Sampling and measurement are specified.
- The model is connected to the question.
- Assumption checks and uncertainty are included.
Mixed-methods methodology example
An explanatory sequential mixed-methods design was used. In the first phase, administrative records were analysed to estimate whether participation in a mentoring programme was associated with first-year retention. In the second phase, purposively selected participants were interviewed to explain mechanisms that could account for the statistical patterns. Integration occurred during participant selection, development of the interview guide, and construction of a joint display comparing numerical trends with qualitative explanations.
Why this works:
- The reason for mixing is explicit.
- The phases are connected.
- Integration is described rather than implied.
Secondary-data methodology example
The analysis used a nationally archived labour-force dataset collected through a stratified multistage design. The dissertation documented the original population, survey period, response process, access licence, weighting variables, and known coverage limitations. Variables were recoded according to a preregistered analysis plan. Complete-case estimates were compared with multiply imputed estimates as a sensitivity analysis. Survey weights and clustering were incorporated into the models.
Why this works:
- It reports where the data came from.
- It recognizes the original sampling design.
- Data preparation and missingness are transparent.
- The analysis respects survey structure.
Modern Research Practices
Use design-specific reporting guidelines
A reporting guideline is a structured list of information that should be reported transparently. Relevant examples include:
- CONSORT 2025 for randomized trials
- STROBE for observational health research
- PRISMA 2020 for systematic reviews
- COREQ or SRQR for qualitative health research
- APA JARS for quantitative, qualitative, and mixed-methods psychology research
Use the guideline while planning and writing, but do not treat checklist completion as proof that the design is valid.
Consider preregistration or a protocol
Preregistration records research questions, hypotheses, outcomes, exclusions, and analysis plans before the relevant evidence is examined.
It can help distinguish:
- Confirmatory analysis
- Exploratory analysis
- Planned decisions
- Data-dependent decisions
Not every dissertation requires preregistration, and its form should match the research tradition. Qualitative projects may benefit more from a transparent protocol, reflexive record, or time-stamped analytic plan than from a rigid hypothesis-testing template.
Document data and code
Where ethics, consent, law, licensing, and institutional policy permit, improve transparency through:
- Data dictionaries
- Codebooks
- Analysis scripts
- Version-controlled code
- Environment files
- Readme documents
- Search strategies
- Extraction templates
- Synthetic or de-identified examples
- Controlled-access repositories
“Data available on request” should not be promised when the researcher lacks the authority to share the data.
Responsible use of artificial intelligence
Generative AI may help with limited tasks such as:
- Brainstorming search terms
- Explaining unfamiliar statistical syntax
- Checking code for obvious errors
- Reformatting non-sensitive notes
- Improving grammar in researcher-written text
- Creating preliminary organizational suggestions
It should not replace:
- Research-design decisions
- Verification of sources
- Ethical judgment
- Interpretation of participant experiences
- Statistical expertise
- Critical appraisal
- Researcher accountability
Before using an AI system:
- Check university and supervisor rules.
- Check ethics approval and participant information.
- Do not upload confidential, identifiable, restricted, or unpublished material unless the system and authorization explicitly permit it.
- Verify every output.
- Record the tool, version or access date, task, and level of human review when the use materially affected the research.
- Disclose use when required.
- Never list an AI system as an author.
- Retain responsibility for plagiarism, fabricated citations, coding errors, and misleading interpretations.
Institutional policies differ, and rapidly changing platform terms may affect confidentiality and data use. Current local guidance should always be checked.
Useful Digital Research Tools
| Task | Possible tools | Methodological caution |
|---|---|---|
| Reference management | Zotero, EndNote, Mendeley | Check imported metadata and citation accuracy |
| Survey collection | Qualtrics, REDCap, LimeSurvey, institution-approved forms | Review privacy, server location, accessibility, and consent |
| Quantitative analysis | R, Python, Stata, SPSS, SAS, JASP | Report versions, packages, settings, and diagnostic procedures |
| Qualitative analysis | NVivo, MAXQDA, ATLAS.ti, Taguette | Software organises material; it does not perform the researcher’s interpretation |
| Review management | Covidence, Rayyan, EPPI-Reviewer | Document screening configuration and conflict resolution |
| Reproducible documents | Quarto, R Markdown, Jupyter | Freeze dependencies and separate confidential data from shareable code |
| Version control | Git and approved repositories | Avoid committing identifiable or restricted data |
| Power analysis | G*Power, R packages, simulation code | Select assumptions based on the actual model rather than a convenient default |
| Data documentation | Codebooks, README files, data dictionaries | Keep documentation synchronized with the final dataset |
Software names should not dominate the methodology. The chapter must describe the methodological operation performed with the software.
Common Methodology Mistakes
Choosing methods because they are easy
Feasibility matters, but convenience alone does not establish suitability. Explain the trade-off between the ideal design and the feasible design.
Writing a research-methods textbook
Do not provide pages of generic definitions without applying them to the study. Every conceptual discussion should lead to a project-specific decision.
Treating qualitative, quantitative, and mixed methods as complete designs
These are broad approaches. Name the specific design, sampling strategy, collection method, and analysis.
Using vague sampling language
“Participants were selected randomly” is insufficient unless the sampling frame and random procedure are explained.
Providing no sample-size rationale
Avoid unsupported rules such as “30 participants is always enough” or “ten interviews guarantee saturation.”
Reporting only the software
“Data were analysed using SPSS” does not identify the variables, models, assumptions, exclusions, or uncertainty.
Mixing results into methodology
The methodology may report the achieved sample and procedural deviations, but substantive findings belong in the results or findings chapter.
Confusing anonymity and confidentiality
Confidential data may still be identifiable to the research team. Describe the protection accurately.
Hiding protocol deviations
Explain changes, their reasons, approval status, and possible consequences. Transparent adaptation is more credible than pretending that the original plan was followed exactly.
Applying the wrong quality vocabulary
Do not add inter-coder reliability to reflexive thematic analysis simply because a generic checklist mentions reliability. Do not use “trustworthiness” as a substitute for quantitative model diagnostics.
Overclaiming causality or generalisation
The claims in the discussion must remain within the design’s capabilities.
Using AI without an audit trail
Unrecorded AI use can create problems involving confidentiality, invented sources, hidden transformations, and unclear authorship responsibility.
Final Methodology Checklist
Before submission, confirm that the chapter:
- Restates the research aim and questions.
- Names the design precisely.
- Connects every major method to a question or objective.
- Explains relevant philosophical or theoretical assumptions.
- Defines the population, cases, corpus, or dataset.
- States inclusion and exclusion criteria.
- Describes sampling and recruitment.
- Justifies the sample size.
- Describes instruments, materials, or systems.
- Reports pilot testing and modifications.
- Explains data collection chronologically.
- Identifies software and versions.
- Describes cleaning, coding, and analysis.
- Reports assumptions, diagnostics, or reflexive procedures.
- Addresses reliability, validity, trustworthiness, integration, or reproducibility appropriately.
- Reports ethics approval and consent where applicable.
- Explains confidentiality and data management.
- Distinguishes anonymisation from pseudonymisation.
- Acknowledges limitations and deviations.
- Avoids presenting substantive results.
- Uses past tense for completed procedures.
- Uses methodological citations rather than unsupported personal preference.
- Follows the university handbook.
- Provides enough detail for informed evaluation.
- Records and discloses material AI use when required.
Conclusion
A dissertation methodology should present a coherent chain from the research question to the evidence, analysis, and conclusions. The strongest chapters are not necessarily the longest or most philosophical. They are specific, justified, transparent, ethically responsible, and adapted to the discipline.
Describe what was done, explain why it was appropriate, acknowledge what it cannot establish, and provide enough detail for readers to evaluate the integrity of the research.
