| name | scientific-style |
| description | Calibrate scientific claims and integrate citations properly. Use when adjusting claim strength, adding hedging language, integrating literature, or matching evidence to conclusions. Complements human-writing by addressing when hedging IS appropriate. |
| argument-hint | ["text to review or claim to calibrate"] |
Scientific Style Guide
This guide covers claim calibration, citation integration, and scientific hedging.
When to Use This Skill
- Calibrating claim strength to match evidence
- Integrating citations smoothly
- Adding appropriate hedging (or removing excessive hedging)
- Matching tense to context
- Avoiding overclaiming or underclaiming
Relationship to human-writing: The human-writing skill flags excessive hedging as an AI pattern. Scientific writing is an exception where measured hedging IS appropriate. This skill explains when and how to hedge in scientific contexts.
Hedging in Scientific Writing
Why Hedge?
Scientific hedging serves legitimate purposes:
- Acknowledges uncertainty inherent in empirical findings
- Distinguishes between data and interpretation
- Protects against overgeneralization
- Signals epistemic humility
- Follows disciplinary conventions
The Calibration Problem
| Too little hedging | Appropriate hedging | Too much hedging |
|---|
| "X causes Y" (when correlational) | "X is associated with Y" | "X may possibly be related to Y in some cases" |
| "This proves..." | "These results suggest..." | "These results might perhaps indicate..." |
| Overclaims evidence | Matches claim to evidence | Undermines your own work |
Claim Strength Ladder
From strongest to weakest:
| Level | Language | Use When |
|---|
| Certain | "X is...", "This demonstrates..." | Established facts, direct observations |
| Strong | "X indicates...", "These results show..." | Strong statistical evidence, replicated findings |
| Moderate | "X suggests...", "This supports..." | Good evidence, reasonable inference |
| Tentative | "X may...", "This could indicate..." | Preliminary findings, indirect evidence |
| Speculative | "It is possible that...", "One might speculate..." | Limited evidence, theoretical extension |
See HEDGING_AND_CLAIMS.md for detailed guidance.
Citation Integration
Two Basic Patterns
Integral citation (author as subject):
- "Smith (2023) demonstrated that..."
- "According to Johnson et al. (2024),..."
Non-integral citation (information emphasized):
- "Gene expression varies with temperature (Smith 2023)."
- "Previous work supports this model (Johnson et al. 2024; Lee 2023)."
When to Use Each
| Use Integral When | Use Non-Integral When |
|---|
| Author's contribution is noteworthy | Information is the focus |
| Discussing specific methodology | Supporting general claim |
| Comparing different authors' views | Citing multiple sources |
| Attributing a controversial claim | Stating widely accepted facts |
Reporting Verbs
| Verb | Connotation | Example |
|---|
| found | Neutral, empirical | "Smith (2023) found that X..." |
| showed | Strong evidence | "Johnson (2024) showed that Y..." |
| demonstrated | Strong, definitive | "Lee et al. demonstrated the effect..." |
| suggested | Tentative | "Prior work suggested a relationship..." |
| argued | Position/claim | "Wang (2023) argued that..." |
| reported | Neutral, descriptive | "Three studies reported increases..." |
| observed | Empirical, direct | "We observed a significant effect..." |
| noted | Passing mention | "As noted by Smith (2022),..." |
| claimed | Slight skepticism | "Authors claimed efficacy..." |
| proposed | Theoretical | "Chen (2024) proposed a model..." |
See CITATION_PATTERNS.md for more patterns.
Tense Usage
By Section
| Section | Primary Tense | Rationale |
|---|
| Introduction | Present | General truths, current state of knowledge |
| Methods | Past | Describes completed actions |
| Results | Past | Reports what was found |
| Discussion | Present + Past | Interprets (present), refers to results (past) |
By Content Type
| Content | Tense | Example |
|---|
| General fact | Present | "Temperature affects enzyme activity." |
| Previous finding | Past | "Smith (2023) found that..." |
| Your methods | Past | "We collected samples from..." |
| Your results | Past | "Gene expression increased..." |
| Your interpretation | Present | "These findings suggest that..." |
| Future work | Future/Conditional | "Future studies should..." |
Common Calibration Issues
Overclaiming
Problem: Claims exceed what evidence supports.
| Overclaim | Calibrated |
|---|
| "This proves X causes Y" | "These results suggest X is associated with Y" |
| "We discovered that..." | "We found that..." |
| "This novel finding..." | "We observed that..." (let novelty speak for itself) |
| "Clearly, X leads to Y" | "X appears to lead to Y" |
Underclaiming
Problem: Excessive hedging weakens legitimate findings.
| Underclaim | Calibrated |
|---|
| "It may be possible that..." | "This suggests..." |
| "Perhaps X might..." | "X may..." |
| "We tentatively propose..." | "We propose..." |
| Stacking hedges: "might possibly suggest" | Choose one: "suggests" |
Correlation vs. Causation
| Causal (use only with experimental evidence) | Associational (use with observational data) |
|---|
| causes, leads to, results in | is associated with, correlates with |
| produces, induces, triggers | is related to, is linked to |
| affects, influences | co-occurs with, predicts |
Section-Specific Guidelines
Introduction
- Present established facts confidently
- Use hedging when citing preliminary or contested findings
- State your objectives directly ("We tested..." not "We attempted to test...")
Methods
- Be direct; minimal hedging needed
- State what you did, not what you tried to do
Results
- Report findings directly
- Save interpretation (and hedging) for Discussion
- "Expression increased" not "Expression appeared to increase"
Discussion
- Match hedging to evidence strength
- Distinguish your findings (past tense) from interpretation (present tense)
- Acknowledge limitations without undermining your conclusions
Quality Checks
Before finalizing:
Related Files
