Elicit Evidence Extraction Sheet (No Invented Effect Sizes)

Prompt

Act as an Elicit research-ops extractor who only uses a pasted paper-list excerpt. You write a evidence extraction sheet the excerpt already supports. You do not invent effect sizes, p-values, citation counts, or sample sizes not in Inputs. This is not medical advice; not an Elicit subscription upsell.
You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs.

Inputs:
- Paper-list excerpt I lock (titles, years, study types present): [PaperExcerpt]
- Extraction notes I lock (columns or outcomes named only if present): [ExtractionNotes]
- Elicit / workspace version notes I lock: [Version]
- Review or project label I may quote (or UNKNOWN): [ReviewLabel]
- Study pair labels already present (or UNKNOWN): [StudyPairs]
- Outcome field names already present (or UNKNOWN): [OutcomeFields]
- Flag rules already present (or UNKNOWN): [FlagRules]
- Words I must not use: [Banned]
- What I must never invent (effect sizes, p-values, citation counts, or sample sizes): [Never]
- Output format: [Format]
- Language: [Lang]

Generate:
1. Honesty ledger: PaperExcerpt nouns, ExtractionNotes, Version, ReviewLabel, StudyPairs, OutcomeFields, FlagRules, Lang. Forbidden: invented effect sizes, p-values, citation counts, sample sizes. Banner: not medical advice; not an Elicit subscription upsell.
2. Evidence extraction sheet: one row per StudyPairs item. Columns: study, PaperExcerpt facts, OutcomeFields, FlagRules, missing cells. Missing FlagRules write NOT IN INPUTS. Use Elicit Find Papers, Extract Data, and Research Reports language when PaperExcerpt supports it. Never print effect-size or p-value VALUES not in PaperExcerpt or ExtractionNotes.
3. Extraction lock: quote ExtractionNotes only. Unnamed outcomes stay NOT IN INPUTS.
4. Version lock: print Version. Refuse features newer than Version if Version is named.
5. Refuse list: inventing effect sizes, inventing p-values, inventing citation counts, inventing sample sizes.
6. Compliance pass: quote Banned and Never hits. Cut them. Format as Format.

Constraints:
- Sheet from PaperExcerpt and ExtractionNotes only. No invented effect-size VALUES. Teach Elicit evidence extraction, not a Consensus or Scite swap.
- Honor Version. No emojis. Add a one-line not-advice banner.

Instructions

Replace every [bracket] with your details before running. Works on ChatGPT, Claude, and Gemini.

Generated Output

This image was generated using the prompt above.

Elicit Evidence Extraction Sheet (No Invented Effect Sizes) - Result

Examples

Example Input

PaperExcerpt: Harbor Quay quay-noise RCT title as pasted year 2021 as pasted study type RCT as pasted. Effect invent NONE. P invent NONE.
ExtractionNotes: primary outcome sleep disruption as pasted; secondary UNKNOWN. Sample invent NONE.
Version: Elicit as pasted (do not invent unreleased auto-meta pack)
ReviewLabel: Harbor Quay noise evidence desk as pasted
StudyPairs: Harbor Quay quay-noise RCT as pasted; River Ops traffic cohort UNKNOWN.
OutcomeFields: sleep disruption as pasted. Citation count NOT IN INPUTS.
FlagRules: missing year flag as pasted.
Banned: guaranteed clinical effect, effect invent, p invent, citation invent
Never: invent effect sizes, invent p-values, invent citation counts, invent sample sizes
Format: ledger + extraction sheet + extraction lock + version + refuse + compliance
Lang: English

Example Output

1. Ledger. PaperExcerpt: Harbor Quay quay-noise RCT title year 2021 study type RCT. Effect invent NONE. P invent NONE. ExtractionNotes primary outcome sleep disruption; secondary UNKNOWN. Sample invent NONE. Version Elicit as pasted. ReviewLabel Harbor Quay noise evidence desk. StudyPairs Harbor Quay quay-noise RCT; River Ops traffic cohort UNKNOWN. OutcomeFields sleep disruption. Citation count NOT IN INPUTS. FlagRules missing year flag. Lang English. Banner: not medical advice; not an Elicit subscription upsell. Forbidden: invented effect sizes, p-values, citation counts, sample sizes.

2. Evidence extraction sheet.
1. Harbor Quay quay-noise RCT | PaperExcerpt year 2021 RCT | OutcomeFields sleep disruption | FlagRules missing year not triggered | effect VALUES NONE | Elicit extract only.
2. River Ops traffic cohort | StudyPairs UNKNOWN | write NOT IN INPUTS | do not invent a second study.
secondary UNKNOWN so write secondary outcomes NOT IN INPUTS. Citation count NOT IN INPUTS so write citation VALUES NOT IN INPUTS.

3. Extraction lock. primary outcome sleep disruption as pasted. secondary UNKNOWN so NOT IN INPUTS. Sample invent NONE.

4. Version lock. Elicit as pasted. Refuse unreleased auto-meta pack.

5. Refuse list. inventing effect sizes; inventing p-values; inventing citation counts; inventing sample sizes.

6. Compliance. Quoted Banned: guaranteed clinical effect, effect invent, p invent, citation invent. Cut. Quoted Never: invent effect sizes, invent p-values, invent citation counts, invent sample sizes. Cut. Format ledger + extraction sheet + extraction lock + version + refuse + compliance.

Missing-data policy: if a field was blank, write NOT IN INPUTS rather than guessing. Lock any tool version named in Inputs; if unnamed, write unknown. No invented testimonials, star ratings, or press logos. If legal, clinical, insurance, HR, education-plan, or veterinary content appears, add a one-line not-advice and de-identify banner. Quote banned-word hits and cut them. End with a gaps list of five bullets the user still owes you. Character and byte caps in the job are hard; print counts when relevant. Refuse to backfill DOIs, exam dumps, PHI, PII, or compensation promises not in Inputs.