Stage 3 / Prompt Toolkit

STAGE 3 · WRITE · HANDS-ON PROMPT BANK

Stage 3 Prompt Toolkit

Copyable prompts for synthesis, research gaps, tables and manuscript drafting. Replace bracketed fields, check every suggestion against your evidence and record only verified outputs in the Stage 3 worksheets.

Open Stage 3 workspace →Stage 3 overview Gemini Notebook guide

Authors remain responsible for every word. AI tools cannot be listed as authors. Do not upload unpublished manuscripts or copyrighted full texts unless permitted. Disclose AI assistance as required by the target journal.

How to use the prompts

Open the matching step and gather your verified inputs (matrix, themes, draft text).
Replace every bracketed field; share only material you are permitted to process.
Ask for evidence and uncertainty; compare the output with your sources.
Edit the output in your own words and record what you used in the AI-assistance log.

Use with Step 1

1 · Make the evidence base analysis-ready

Check an extraction matrix for consistencyReplace the bracketed fields and verify all output against your evidence.View prompt +

Copy and replace the bracketed fields

I am preparing the extraction matrix of a systematic literature review for synthesis. Review the table below for consistency only. Do not add or change study findings.

Research questions: [paste RQs]
Extraction matrix (CSV or table, one row per included study): [paste]

Return a table with: column name, inconsistent or non-standard values found, a proposed controlled value list, and the study IDs affected. List cells that are empty and should be marked “not reported” or “unclear”. Mark every proposal as requiring my verification.

Propose a study-level coding schemeReplace the bracketed fields and verify all output against your evidence.View prompt +

Copy and replace the bracketed fields

Using these research questions and the extraction fields below, propose a study-level coding scheme (variables and controlled values) suitable for descriptive counts and grouping in a systematic review.

Research questions: [paste]
Extraction fields currently used: [paste field names]
Discipline / topic: [describe]

For each variable give: name, definition, allowed values, a rule for ambiguous cases, and which RQ it serves. Do not code any studies.

Use with Step 2

2 · Choose and justify the synthesis method

Compare synthesis options for my evidenceReplace the bracketed fields and verify all output against your evidence.View prompt +

Copy and replace the bracketed fields

Act as a systematic review methods adviser. Based on the description below, compare narrative synthesis, thematic synthesis, framework synthesis, synthesis without meta-analysis (SWiM) and meta-analysis for each research question.

Research questions: [paste]
Number of included studies: [n]
Study designs and outcome types (summary): [paste]
Heterogeneity observed: [describe]

Return a table: RQ, suitable approach(es), why, conditions that would make it unsuitable, and a methodological reference I should look up and verify. Do not fabricate references; if unsure, say so.

Draft the synthesis-method paragraphReplace the bracketed fields and verify all output against your evidence.View prompt +

Copy and replace the bracketed fields

Draft a concise Method sub-section titled “Data synthesis” for a systematic review, using only the decisions below. Use past tense and formal academic style.

Synthesis approach per RQ: [paste]
Grouping structure: [paste]
How quality appraisal informs the synthesis: [paste]
Software used: [paste]

Place [CITATION NEEDED] where a methods reference should go; do not insert references yourself.

Use with Step 3

3 · Describe the evidence base

Interpret descriptive countsReplace the bracketed fields and verify all output against your evidence.View prompt +

Copy and replace the bracketed fields

Below are descriptive counts from the included studies of my systematic review. Write 2–3 paragraphs that interpret the patterns (not just restate numbers). Report numbers as “n of N (x%)”.

Total included studies (N): [n]
Counts by year: [paste]
Counts by country/region: [paste]
Counts by design: [paste]
Counts by domain / technique: [paste]

Use only these numbers. Flag any count that does not add up to N.

Plan a bibliometric mapReplace the bracketed fields and verify all output against your evidence.View prompt +

Copy and replace the bracketed fields

I want to add a science-mapping figure (e.g. keyword co-occurrence in VOSviewer or Bibliometrix) to my systematic review.

Research questions: [paste]
Data source and export format: [Scopus/WoS, file type]
Number of records: [n]

Advise whether a bibliometric map would answer any RQ, which analysis type fits, which settings and thresholds I must report, and how to describe the method transparently. Do not interpret clusters you have not seen.

Use with Step 4

4 · Develop themes by research question

Suggest initial codes from extracted findingsReplace the bracketed fields and verify all output against your evidence.View prompt +

Copy and replace the bracketed fields

I will paste extracted findings from included studies of a systematic review (with study IDs). Suggest initial descriptive codes.

Research question: [paste one RQ]
Extracted findings (Study ID — finding — page): [paste]

Return a table: code name, short definition, study IDs and the exact phrase from my data that supports the code. Do not use information that is not in the pasted text. I will review and revise all codes.

Group codes into themes and find contradictionsReplace the bracketed fields and verify all output against your evidence.View prompt +

Copy and replace the bracketed fields

Here are my verified codes with study IDs for one research question. Propose how they could be grouped into descriptive themes and possible analytical themes.

Research question: [paste]
Codes (code — definition — study IDs): [paste]
Quality ratings per study: [paste]

For each proposed theme give: name, definition, codes included, supporting studies, contradicting or divergent studies, and a note on strength of evidence (number and quality of studies). Mark which themes rest on a single study.

Turn a theme into a synthesised paragraphReplace the bracketed fields and verify all output against your evidence.View prompt +

Copy and replace the bracketed fields

Write one results paragraph for the theme below. Compare studies rather than summarising them one by one. Start with a topic sentence stating the synthesised finding, then support it with study IDs in brackets, then note contradictions or limitations of the evidence.

Theme and definition: [paste]
Supporting evidence (Study ID — finding — design — quality): [paste]
Contradicting evidence: [paste]

Use only the evidence provided; do not add claims or citations.

Use with Step 5

5 · Derive research gaps and an agenda

Identify evidence-based gapsReplace the bracketed fields and verify all output against your evidence.View prompt +

Copy and replace the bracketed fields

Using the synthesis summary below, identify research gaps for a systematic review. Classify each gap as evidence, knowledge, methodological, population/contextual, theoretical or practical.

Research questions: [paste]
Descriptive summary (counts): [paste]
Themes and strength of evidence: [paste]
Common limitations reported by included studies: [paste]

Return a table: gap type, gap statement, supporting evidence from my summary (with counts or study IDs), and why it matters. Do not state a gap that is not supported by the information I provided.

Convert gaps into future research questionsReplace the bracketed fields and verify all output against your evidence.View prompt +

Copy and replace the bracketed fields

For each verified research gap below, propose 1–2 specific future research questions and an appropriate study design (population, setting, method, outcome measures).

Gaps: [paste table]

Avoid generic recommendations such as “more research is needed”. Keep each question answerable by a single study.

Use with Step 6

6 · Design tables and figures

Plan the core tables and figuresReplace the bracketed fields and verify all output against your evidence.View prompt +

Copy and replace the bracketed fields

Recommend a set of tables and figures for my systematic review manuscript and map each to a PRISMA 2020 item or research question.

Research questions: [paste]
Number of included studies: [n]
Synthesis approach: [paste]
Target journal word / figure limits (if known): [paste]

Return a table: number, title, content/columns, purpose (RQ or PRISMA item), main text or supplementary. Keep the set concise.

Write a caption and check a tableReplace the bracketed fields and verify all output against your evidence.View prompt +

Copy and replace the bracketed fields

Review this table/figure description for clarity and write a self-explanatory caption. Define all abbreviations and state the data source.

Table or figure content: [paste]
Abbreviations used: [paste]

Also flag any values that seem inconsistent (e.g. totals not matching N = [n]). Do not change the data.

Use with Step 7

7 · Draft the manuscript

Convert protocol text into the Method sectionReplace the bracketed fields and verify all output against your evidence.View prompt +

Copy and replace the bracketed fields

Convert the protocol and portfolio notes below into the Method section of a systematic review. Use past tense, report what was actually done, and follow PRISMA 2020 items 5–15 as sub-headings where relevant.

Protocol text: [paste]
Stage 1 search details (sources, dates, strings): [paste]
Stage 2 selection, appraisal and extraction details: [paste]
Deviations from the protocol: [paste]

Insert [CHECK] wherever information is missing. Do not invent numbers, dates or tools.

Draft the Discussion openingReplace the bracketed fields and verify all output against your evidence.View prompt +

Copy and replace the bracketed fields

Draft the first two paragraphs of a Discussion section for a systematic review.

Research questions: [paste]
Principal findings per RQ (verified): [paste]
Closest prior reviews and how they differ (from my notes): [paste]

Paragraph 1: summarise principal findings and the main contribution. Paragraph 2: compare with prior reviews. Use [CITATION] placeholders; do not add references.

Claim–evidence audit of a draft sectionReplace the bracketed fields and verify all output against your evidence.View prompt +

Copy and replace the bracketed fields

Audit the draft section below. List every factual or evaluative claim and indicate whether it is supported by a cited included study (study ID or citation present), supported only by background literature, or unsupported.

Draft section: [paste]
List of included-study IDs/citations: [paste]

Return a table: claim, support status, citation(s) found, suggested action (keep / add evidence / soften / remove). Do not rewrite the section.

Use with any AI-assisted Stage 3 task

AI Output Verification Checklist

Tick items as you work; this checklist is not saved.

Evidence traceableEvery claim about the literature is linked to an included study and an evidence location.No invented sourcesEvery citation, number and quotation was checked against the original source.Meaning preservedEdits did not change technical meaning, numbers or the strength of claims.Human judgement retainedThemes, gaps, journal choice and responses were decided by the authors.Use disclosedAI assistance is logged and will be disclosed according to the journal’s policy.Privacy consideredOnly content permitted by policy and copyright was entered into the AI service.

An AI response is a draft aid, never a source to cite.

Responsible-use instruction

Add this to a prompt when you need stronger safeguards:

Use only the information I provide. Separate direct evidence from suggestions. Do not invent citations, study IDs, counts, quotations or page numbers. If information is missing, state “not reported” or ask me. Do not make final analytical or authorship decisions; I will verify and decide.

Use only the information I provide. Separate direct evidence from suggestions. Do not invent citations, study IDs, counts, quotations or page numbers. If information is missing, state “not reported” or ask me. Do not make final analytical or authorship decisions; I will verify and decide.

Start Stage 3 Step 1 →Open Stage 3 Portfolio