Draft a Research Methodology Section with Justified Design and Validity Threats

You get a methodology section that ties each design, sampling and analysis choice to your research question, plus a validity-threat table with a mitigation for each.

Prompt

Prompt
Act as a research methods advisor who reviews thesis and grant proposals.

Your job: draft the methodology section for my study. Before writing, ask me these questions in one message and wait for my answers:
1. What is your research question or hypothesis?
2. What data or participants can you access, and roughly how many?
3. What field and level is this for (e.g., master's thesis, grant), and what length is expected?
4. What constraints do you have (time, budget, ethics approval, software)?

Rules:
- Tie every major choice (design, sampling, instruments, analysis) to the research question in one explicit sentence.
- For each major choice, name one alternative method and say in one line why it was not chosen.
- List threats to validity by type (internal, external, construct, statistical; or credibility and transferability for qualitative work) and give one concrete, doable mitigation for each.
- Do not invent citations, sample sizes or results; write "(source needed)" where I must add a reference.
- Use plain academic prose in the tense that fits my stage.

Output format:
1. Research design
2. Participants and sampling
3. Data collection and instruments
4. Data analysis
5. Table: Threat | Type | Why it matters here | Mitigation
6. Limitations (3-5 bullets)

End with one line naming the biggest unresolved weakness and what information would fix it.

Example output

At a glanceMethodology sections drafted for the vaccine-text RCTEach choice tied to the research question, with an alternative rejected
  1. Research designTwo-arm randomized controlled trial
  2. Participants and samplingAbout 600 staff randomized 1:1, stratified by campus
  3. Data collectionWeeks 1, 3, 6Three reminder texts; uptake from health records
  4. Data analysisLogistic regression, intention-to-treat
  5. Threats to validity4 threats, each with a mitigation
  6. LimitationsSample size not yet justified is the biggest gap

Example for: MPH student testing whether text-message reminders raise flu vaccine uptake among university staff, about 600 staff on 2 campuses, 10-week timeline.

1. Research design

This study will use a two-arm randomized controlled trial. Randomization supports the question because it isolates the effect of reminders on uptake from staff motivation and prior habits. Alternative not chosen: a before-after comparison, rejected because seasonal campaigns would confound any change.

2. Participants and sampling

All benefits-eligible staff with a mobile number on file (approx. 600) will be invited; those who opt out of texts are excluded. Individuals will be randomized 1:1, stratified by campus. Stratifying supports the question because the two campuses run separate clinics. Alternative not chosen: cluster randomization by department, rejected because about 30 departments gives too few clusters. A power calculation should be added (source needed).

3. Data collection and instruments

The intervention group receives three texts (weeks 1, 3, 6) with clinic hours and a booking link. The outcome is vaccination recorded in occupational health records by week 10. Record-based outcomes support the question because self-report overstates uptake (source needed). Alternative not chosen: survey follow-up, rejected for recall bias and low response.

4. Data analysis

The primary analysis compares uptake proportions using logistic regression adjusted for campus, intention-to-treat. This supports the question because it estimates the effect of offering reminders, which is the policy decision. Alternative not chosen: per-protocol analysis only, kept as a sensitivity check.

5. Threats to validity
ThreatTypeWhy it matters hereMitigation
Contamination between armsInternalColleagues may share reminder textsAsk in exit question whether staff saw others' texts; report rate
Off-site vaccinations missedConstructPharmacy shots are not in recordsAllow staff to upload proof; analyse with and without uploads
Single employerExternalUniversity staff may differ from other workplacesDescribe sample demographics; limit claims to similar settings
Low baseline varianceStatisticalIf uptake is already high, effect is hard to detectCheck last year's uptake before finalizing sample size
6. Limitations
  • One flu season only; effect may differ in high-publicity years.
  • Staff without mobile numbers are excluded, which may skew toward office roles.
  • Text wording is fixed, so the study cannot show which message elements matter.
  • No cost data, so cost-effectiveness cannot be assessed.

Biggest unresolved weakness: the sample size is not yet justified; last season's uptake rate and a target effect size would fix this.

Written by AI from this prompt. Your results will vary.

Tips and common edits

Why this works

  • Requiring one sentence linking each choice to the research question makes the design defensible to supervisors and reviewers.
  • Naming a rejected alternative per choice anticipates the most common examiner question: why not another method?
  • The threat table forces a concrete mitigation for every validity risk instead of a vague limitations paragraph.

Use when you have a research question but need a defensible methods draft for a thesis, proposal or report. Have your question, sample access and constraints ready. Works in ChatGPT, Claude or Gemini.

Common edits

  • Switch the validity types to qualitative criteria (credibility, dependability) for interview or case-study work.
  • Ask for a 300-word condensed version for a conference abstract or grant summary.

Make it repeatable

Keep using this prompt without pasting it in every time.

Prompt details

By
Prompts-GPT editors
Published
How prompts are written and checked

Did this prompt work?

Keep exploring

More in Research