Dissertation Methodology

The eight components

Not every dissertation needs all eight as separate sections, and disciplines vary. But an examiner will look for each of these ideas somewhere, and missing one is noticed.

1. 1Research philosophy. Positivism, interpretivism, pragmatism or critical realism, and why it fits your question rather than merely being named.
2. 2Approach. Deductive, testing existing theory, or inductive, building it from data.
3. 3Strategy. Survey, experiment, case study, ethnography, action research, systematic review.
4. 4Sampling. Frame, technique, size, and justification. Include a power analysis for quantitative work and a saturation argument for qualitative.
5. 5Data collection. The instrument, its provenance, its reliability, and the procedure in enough detail to replicate.
6. 6Analysis. The specific technique, the software, and how you handled assumptions or coding reliability.
7. 7Ethics. Approval reference, consent, anonymisation, data storage, and any vulnerable participant considerations.
8. 8Limitations. What the design cannot establish, stated before an examiner has to point it out.

Research philosophy, used rather than named

The single most common weakness in this chapter is naming a philosophical position in one paragraph and never referring to it again. If your stated philosophy does not constrain any subsequent choice, it is decoration and an examiner will treat it as such.

Used properly, the philosophy explains why your method is appropriate. An interpretivist position implies you are interested in how participants construct meaning, which rules out a closed-question survey as a primary instrument. A positivist position implies measurable variables and generalisable relationships, which rules out a sample of six. Say that explicitly.

Sampling, and defending your numbers

Sample size is the point examiners press on most reliably, and the defence differs completely between quantitative and qualitative work.

For quantitative designs, justify with a power analysis: state the test, the expected effect size with a citation rather than a guess, alpha, desired power, and the resulting minimum sample. Reporting that you recruited 214 when power analysis indicated 128 is a strength worth stating.

For qualitative designs, sample size is defended on saturation and on the depth the design requires, not on numbers. Fourteen interviews is defensible if you can describe the point at which new themes stopped appearing. It is not defensible as an arbitrary target, and "due to time constraints" is not a methodological justification.

Ethics, written properly

Ethics sections are often reduced to a sentence saying approval was obtained. That is the minimum and it reads as the minimum.

• The approval reference and the granting committee.
• How informed consent was obtained, and what participants were told they were consenting to.
• How anonymity or confidentiality was maintained, and the difference between them in your study.
• Where data was stored, in what form, for how long, and who had access.
• Right to withdraw: how it was communicated and up to what point it applied.
• Any specific considerations: vulnerable participants, sensitive topics, power relationships where you researched your own workplace.

Limitations, stated before you are asked

Naming your own limitations precisely is a strength. Having an examiner name them for you is not. The section should identify what your design genuinely cannot establish and what that means for your claims.

Be specific rather than ritual. "The sample was small" is generic. "A cross-sectional design cannot establish the direction of the relationship between contact frequency and cohesion, so the causal reading offered in section 5.2 is proposed rather than demonstrated" tells an examiner you understand your own study.

Quantitative, qualitative and mixed methods

The choice follows from the question, not from preference or from which you find easier. Getting this wrong is the most expensive error in the chapter because it cannot be fixed by editing.

A question asking whether one variable predicts another, or whether groups differ, is quantitative. It needs a sample large enough to detect the effect, measurement instruments with known reliability, and inferential statistics. A question asking how people experience, describe or make sense of something is qualitative. It needs depth rather than scale, and defends its sample on saturation rather than power.

Mixed methods are legitimate but they double the workload, which at taught level is rarely the right trade. If you use them, state the design explicitly: convergent, where both strands run in parallel and are compared; explanatory sequential, where quantitative results are explained by follow-up qualitative work; or exploratory sequential, where qualitative work informs instrument design. Saying "mixed methods" without naming the design signals the choice was not deliberate.

Reliability, validity and their qualitative equivalents

Every methodology chapter has to address quality, and the vocabulary differs between paradigms in a way that markers notice.

• Quantitative reliability: internal consistency reported as Cronbach alpha, test-retest where relevant, and inter-rater agreement where coding is involved.
• Quantitative validity: construct validity from the instrument literature, internal validity from the design, external validity from the sampling.
• Qualitative credibility: member checking, prolonged engagement, or triangulation across data sources.
• Qualitative transferability: thick description of the context, so a reader can judge whether findings apply to theirs.
• Qualitative dependability and confirmability: an audit trail of coding decisions, and reflexivity about your own position relative to participants.