ChatGPT prompt for scientific research

Help me plan or synthesize a scientific research investigation with transparent assumptions and uncertainty.

Research objective: [RESEARCH OBJECTIVE]
System, population, or phenomenon: [STUDY CONTEXT]
Available evidence, data, and sources: [AVAILABLE EVIDENCE]
Methods, tools, and constraints: [METHODS AND CONSTRAINTS]
Required deliverable and audience: [DELIVERABLE AND AUDIENCE]

Ask up to 3 clarifying questions only if a required input is missing. Do not invent experiments, datasets, measurements, citations, effect sizes, statistical results, mechanisms, or safety approvals. Do not present a hypothesis as an established finding.

Produce:
1. A concise research question and, where appropriate, a falsifiable hypothesis with a predicted direction and a competing explanation.
2. A study or analysis plan: variables, operational definitions, comparison group or baseline, sampling or data-selection approach, procedure, and primary outcome. Explain why each choice tests the question.
3. A table separating background claims, direct evidence, inference, and unknowns. For each supplied source, note design, relevant result, limitation, and applicability to this context.
4. A bias and validity review covering confounding, measurement error, selection bias, statistical power or sample-size uncertainty, reproducibility, and generalizability as relevant.
5. A results-reporting template that states effect estimates and uncertainty where data exist, avoids causal language unless justified, and names null or ambiguous findings.
6. The next three highest-value actions, such as locating a primary study, running a control, preregistering an analysis, or consulting a domain specialist.

Use SI units and standard scientific terminology appropriate to the field. Before answering, check that every factual assertion is linked to supplied evidence or labeled as a proposed method, and explicitly flag conclusions that cannot be drawn from the available data. For health, laboratory safety, regulated work, or decisions affecting people, identify when qualified supervision, institutional review, or a licensed professional is needed.

Critically appraise this scientific paper or study notes: [PAPER OR NOTES]. The claim I want to evaluate is [CLAIM]. My use context is [DECISION CONTEXT]. Produce a structured appraisal covering question, design, population or materials, comparator, outcomes, main result, uncertainty, sources of bias, external validity, and what the study cannot establish. Separate the authors’ findings from your interpretation. Do not infer results omitted from the paper. Ask up to 3 questions only if a required input is missing.

Create an analysis plan for this dataset and research question: [RESEARCH QUESTION]. Variables, units, collection method, and expected sample are [DATA DICTIONARY]. Known data-quality issues are [DATA LIMITATIONS]. Specify primary and secondary outcomes, inclusion rules, transformations, model or test options, assumption checks, missing-data handling, visualizations, and how to report uncertainty. Do not calculate results or claim significance without data. Flag any design issue that limits causal interpretation. Ask up to 3 questions only if a required input is missing.

Draft a structured scientific abstract from these verified study details: [STUDY DETAILS]. The target journal or conference rules are [VENUE RULES]. Produce Background, Methods, Results, and Conclusions in [WORD LIMIT] words. Include numerical results only exactly as supplied, state the study design and sample, and make conclusions proportionate to the design. Do not add significance claims, mechanisms, or citations not in my material. End with a list of statements that need verification before submission. Ask up to 3 questions only if a required input is missing.