Skip to main content

Evidence Language Scale

Match the strength of your language to the strength of the evidence. This protects users from false hope while still communicating what research shows.

The Scale​

Evidence LevelSourcesLanguageExample
StrongMultiple high-quality RCTs, meta-analyses, systematic reviews"Research demonstrates...""Research demonstrates that cognitive behavioral therapy reduces symptoms of depression."
ModerateSome RCTs, consistent observational studies, mechanistic evidence"Studies suggest...""Studies suggest that meditation may reduce anxiety in some individuals."
PreliminarySmall studies, case series, pilot trials, emerging research"Early research indicates...""Early research indicates a possible connection between breathwork and stress reduction."
TraditionalHistorical use, documented traditions, no clinical trials"Traditionally used for...""Traditionally used for calming the mind in Vedic practices dating to 400 CE."
UnknownNo evidence either way, mechanism unclear"The mechanism is not understood...""The mechanism is not understood, though practitioners report benefits."

Strong Evidence​

Criteria​

  • Multiple high-quality randomized controlled trials (RCTs)
  • Meta-analyses or systematic reviews
  • Consistent findings across studies
  • Large sample sizes (typically n greater than 100)
  • Published in peer-reviewed journals
  • Replicated by independent researchers

Language​

Appropriate:

  • "Research demonstrates..."
  • "Evidence shows..."
  • "Studies consistently find..."
  • "A systematic review concluded..."

Example:

"Research demonstrates that cognitive behavioral therapy effectively reduces symptoms of depression. A meta-analysis of 115 studies (n=9,138) found significant effects across diverse populations (PMID: 12345678)."

Limitations to Note​

Even with strong evidence:

  • Individual results vary
  • Effect sizes differ
  • Context matters
  • Not universally effective

Moderate Evidence​

Criteria​

  • Some RCTs with positive findings
  • Consistent observational studies
  • Mechanistic evidence supporting claims
  • Sample sizes moderate (typically 30-100 participants)
  • Some replication
  • Peer-reviewed publication

Language​

Appropriate:

  • "Studies suggest..."
  • "Research indicates..."
  • "Evidence suggests..."
  • "Findings point to..."

Example:

"Studies suggest that regular meditation practice may reduce anxiety. Research with 476 participants found modest improvements in anxiety measures (PMID: 23456789). More research is needed to understand mechanisms and optimal practice parameters."

Limitations to Note​

  • Smaller sample sizes
  • Fewer replications
  • Mechanism may be unclear
  • Protocol variations
  • Individual differences

Preliminary Evidence​

Criteria​

  • Small studies or pilot trials
  • Case series or case reports
  • Single researcher or lab
  • Sample sizes small (typically fewer than 30 participants)
  • Limited replication
  • Emerging area of study

Language​

Appropriate:

  • "Early research indicates..."
  • "Preliminary findings suggest..."
  • "Initial studies show..."
  • "Pilot research found..."

Example:

"Early research indicates that the 4-7-8 breathing technique may help with sleep onset. A pilot study with 23 participants found reduced time to fall asleep (PMID: 34567890). These findings are preliminary and require replication in larger samples."

Limitations to Note​

  • Very limited evidence
  • Needs replication
  • Small samples
  • May not generalize
  • Findings may not hold up

Traditional Evidence​

Criteria​

  • Historical documentation of use
  • Traditional texts or practices
  • Lineage transmission
  • Cultural knowledge
  • No clinical trial evidence

Language​

Appropriate:

  • "Traditionally used for..."
  • "Historically practiced for..."
  • "[Tradition] teaches that..."
  • "According to [tradition]..."

Example:

"Traditionally used for calming the mind in yogic practices. The Yoga Sutras of Patanjali (c. 400 CE) describe pranayama as a method for steadying awareness. While traditionally valued, clinical research on this specific technique is limited."

Important Notes​

  • Distinguish clearly from clinical evidence
  • Provide historical context
  • Respect tradition
  • Don't claim clinical validation
  • Note where research is absent

Unknown Mechanism​

Criteria​

  • Practice exists
  • Some may report benefits
  • Mechanism not understood
  • No satisfactory explanation
  • Research absent or unclear

Language​

Appropriate:

  • "The mechanism is not understood..."
  • "How this works is unclear..."
  • "The process is not well studied..."
  • "While the mechanism is unknown..."

Example:

"The mechanism by which acupressure may affect pain is not fully understood. While traditional Chinese medicine explains this through qi flow, modern research has not conclusively identified physiological mechanisms. Some practitioners and users report benefits, though controlled studies are limited."

Important Notes​

  • Acknowledge uncertainty honestly
  • Don't speculate beyond evidence
  • Report user experience separately from mechanism
  • Note research gaps

Forbidden Language​

Never use these phrases, regardless of evidence level:

❌ "Will cure" ❌ "Definitely works" ❌ "Guaranteed to" ❌ "Proven to eliminate" ❌ "This is the answer" ❌ "You'll be healed" ❌ "Never fails" ❌ "100% effective"

Required Qualifiers​

Always include appropriate qualifiers:

Individual Variation​

✅ "Results vary" ✅ "Some individuals find..." ✅ "Effects differ by person" ✅ "Individual experiences vary"

Complementary Nature​

✅ "May support..." ✅ "Can complement..." ✅ "One tool among many" ✅ "Does not replace medical care"

Uncertainty​

✅ "More research is needed" ✅ "Findings are preliminary" ✅ "The mechanism is unclear" ✅ "Evidence is limited"

Citation Requirements​

Format​

Every claim about evidence must include:

**Source:** [Author] ([Year]) - [Brief description]
**Reference:** PMID: [number] or DOI: [identifier]
**Sample:** n=[number] participants
**Finding:** [Specific result]
**Limitation:** [Key limitation]

Example​

**Source:** Zaccaro et al. (2018) - Review of slow breathing effects
**Reference:** PMID: 29672592
**Sample:** Review of 15 studies
**Finding:** Slow breathing (6 breaths/min) increased HRV and reduced anxiety
**Limitation:** Most studies had small samples and protocol variations

Required Elements​

  • PMID (PubMed ID) or DOI preferred
  • Sample size noted
  • Key limitation acknowledged
  • Date of publication
  • Type of study (RCT, review, observational, etc.)

Quality Assessment​

Before citing evidence, assess quality:

High Quality Study​

  • Randomized controlled trial
  • Large sample (more than 100 participants)
  • Peer-reviewed
  • Well-controlled
  • Replicated
  • Recent (within 10 years preferred)

Medium Quality Study​

  • Some controls
  • Moderate sample (30-100 participants)
  • Peer-reviewed
  • Some limitations
  • Limited replication

Low Quality Study​

  • Observational or case series
  • Small sample (fewer than 30 participants)
  • Limited controls
  • Significant limitations
  • No replication

Not Suitable​

  • Pre-print (not peer-reviewed)
  • Retracted
  • Fraudulent
  • Conflict of interest concerns
  • Cherry-picked data

Edge Cases​

Conflicting Evidence​

When studies conflict:

## Evidence Note

Research findings are mixed. Some studies suggest [X], while others find [Y].
This may be due to [methodology differences / population variations / etc.].

**Supporting studies:**
- [Citation 1]
- [Citation 2]

**Contradictory studies:**
- [Citation 3]
- [Citation 4]

Given mixed evidence, approach with caution and attention to your individual response.

Absence of Evidence​

When no research exists:

## Evidence Note

Clinical research on this specific practice is limited or absent. This does not mean
it is ineffective—it means we don't have scientific data to evaluate it.

The practice has been traditionally used for [purpose] in [tradition]. If you choose
to try it, pay attention to your individual response.

Enforcement​

Evidence language is checked by:

Automated Checks​

  • Scan for forbidden phrases
  • Verify citations present
  • Check language matches level

Clinical Reviewer​

  • Verifies evidence level accurate
  • Confirms citations valid
  • Assesses language appropriateness
  • Checks limitations noted

Ethics Guardian​

  • Final approval of claims
  • Can block overstated language
  • Ensures user protection

Severity: HIGH - Must fix before deployment

"Match language to evidence. Honor uncertainty. Protect users from false hope."