Methodology Chapter: Design, Decisions, and Rigor

Explain how the question can be answered

A methodology chapter shows how a study’s evidence was obtained and how the chosen approach supports the claims made from it. It is an argument for the research design, not a catalogue of definitions. Begin with the question and the type of answer sought. A study estimating prevalence needs a credible population and measure; an inquiry into lived experience needs access to participants’ accounts and a transparent interpretive process. The design should follow the question rather than a method the writer happens to prefer.

State the setting, period and unit of analysis. Are observations about people, organizations, events or documents? A multi-site project may collect individual interviews but draw conclusions about institutional practice; explain the link between those levels. Describe the researcher’s access and any change in the setting during the study. A method performed during an unusual disruption may yield evidence with different limits than the same method under routine conditions.

Justify the design against realistic alternatives. A survey can compare responses across a larger group but may not explain why respondents interpret a question differently. Interviews can illuminate processes but do not by themselves establish population percentages. An experiment may support a causal comparison when allocation and implementation are credible, but it may be infeasible or inappropriate in a particular setting. Show why the selected approach is the best available for this question and what it cannot establish.

Make selection and collection traceable

Define the population or corpus and how cases entered the study. For a survey, report the sampling frame, recruitment, eligibility, response and reasons for missing data. For a qualitative study, describe the selection logic, variation sought and the point at which the collected material was sufficient for the research purpose. Avoid claiming a convenience sample represents everyone merely because it contains many participants.

Describe instruments and procedures at a level another researcher can understand. Include how a questionnaire was developed or adapted, what an interview guide covered, how documents were located or how measurements were made. Note piloting and changes. A short appendix can hold a full instrument while the chapter explains the decisions most relevant to validity and interpretation.

Give the sequence. Who recruited participants, where did data collection take place, how long did it last and how were responses recorded? If the researcher is also a manager, teacher or clinician in the setting, explain how that relationship could affect participation and answers. Document deviations from the planned method. A polished account that hides changes leaves the reader unable to judge their consequences.

Describe analysis rather than naming it

Explain what happened to the evidence after collection. A quantitative analysis should define variables, exclusions, transformations, missing-data handling, comparisons and uncertainty. If a model is used, state the assumptions relevant to its interpretation. A statistically significant coefficient does not by itself establish a causal effect; the design and potential confounding matter.

For qualitative analysis, explain how material was read, coded, compared and interpreted. Were categories derived from a theoretical framework, developed from the data or both? Give an example of a theme and the sort of evidence that supports it. Discuss contradictory cases rather than treating all quotations as confirmation. Software may help organize records but does not perform the researcher’s interpretive judgment.

For mixed methods, specify when each kind of data was collected and how findings were integrated. Placing survey results and interviews in separate chapters is not automatically integration. Explain whether interviews explored an unexpected pattern, whether one dataset informed sampling for the other and how disagreement between them was handled.

Protect participants and evaluate rigor

Describe consent, confidentiality and data protection in relation to the actual study. Identify any approval or permission obtained and the limits of privacy promised. A focus group cannot guarantee that participants will keep others’ statements confidential. If data are de-identified, explain how potentially identifying combinations of details are handled. Do not imply that ethical approval eliminates every risk.

Evaluate the strongest threats to a trustworthy conclusion. Measurement error, nonresponse, selection, confounding and attrition may matter in quantitative designs. In qualitative work, positionality, recruitment, context and the handling of alternative interpretations may matter. Use relevant checks: a validated instrument, sensitivity analysis, an audit trail, comparison of cases or participant feedback where appropriate. A list of generic “reliability and validity” definitions does not show what was actually done.

State limitations proportionately. A single site can provide a useful explanation of a process without representing every institution. Self-report may be well suited to perceptions but weak evidence for actual behavior. Explain how each limitation bears on the question and whether another source of evidence reduces it. Avoid declaring a study useless because it cannot answer questions it was never designed to address.

Connect design to the claims that follow

End by reminding the reader which conclusions the method can support. If the design compares two periods without a control group, later results should be described as observed change rather than a proven effect of the intervention. If interviews sampled only service users, later claims should not speak for people who never enrolled. The methodology chapter prepares the reader to assess the results and discussion on their merits.

Revise the chapter after completing analysis so it accurately reports the work performed. Preserve the distinction between planned and actual procedures, and include enough detail for scrutiny without turning the chapter into a log of irrelevant steps. A strong methodology makes the path from question to evidence to inference visible.