Evaluating Scientific Claims

How to read research, understand statistics, and spot misinformation.

Why This Matters

Being scientifically literate helps you:

  • Make informed health decisions
  • Avoid being misled by false claims
  • Understand news about science
  • Evaluate product claims
  • Discuss science-based topics intelligently
  • Protect yourself and family from misinformation

Understanding Research

Types of Studies

Study TypeDescriptionStrength
Case reportSingle patient observationWeakest; suggests hypotheses
Case seriesMultiple similar casesStill weak; patterns only
Cross-sectionalSnapshot of population at one timeShows association, not causation
Case-controlCompare people with/without conditionModerate; retrospective
CohortFollow groups over timeStrong; prospective
Randomized controlled trial (RCT)Random assignment to treatment/controlGold standard
Systematic reviewCombines multiple studiesVery strong
Meta-analysisStatistical combination of studiesStrongest evidence

Hierarchy of Evidence

LevelEvidence Type
HighestSystematic reviews and meta-analyses
HighRandomized controlled trials
ModerateCohort studies
LowerCase-control studies
LowCase series, case reports
LowestExpert opinion, anecdotes

Study Quality Indicators

Good SignWhat It Means
Peer reviewedExperts checked the work
Large sample sizeMore reliable results
Control groupComparison baseline
RandomizationReduces bias
BlindingPrevents expectations from affecting results
ReplicationOthers got same results
Pre-registrationMethods declared before study

Statistics Basics

Key Concepts

TermMeaningExample
MeanAverageSum of values divided by count
MedianMiddle valueHalf above, half below
Standard deviationSpread of dataHow much values vary
Sample size (n)Number of subjectsLarger = more reliable
Confidence intervalRange of likely true values"True value is between X and Y"

P-values and Significance

TermMeaning
p-valueProbability results happened by chance
p < 0.05Less than 5% chance of random occurrence
Statistically significantp-value below threshold (usually 0.05)
Clinically significantEffect large enough to matter

Important: Statistical significance does not mean the effect is large or important.

Understanding Risk

TermDefinitionExample
Absolute riskActual probability2 in 100 people affected
Relative riskComparison between groups"Doubles your risk"
Number needed to treat (NNT)How many need treatment for one to benefitNNT of 100 = treat 100 for 1 benefit
Number needed to harm (NNH)How many treated before one harmedRisk-benefit ratio

Misleading Statistics

TacticThe ProblemExample
Relative risk without contextSounds dramatic"Doubles risk" (from 1 to 2 in 1000)
Cherry-picking time framesChoose data that supports claimStart graph at convenient point
Confusing correlation/causationAssociation isn't causeIce cream and drowning
Small sample sizesRandom variation3 out of 4 dentists
Missing confidence intervalsUncertainty hiddenSingle number sounds precise

Reading Science News

Translation Problems

What Headlines SayWhat It Often Means
"Scientists say"May be one small study
"Breakthrough"Incremental progress
"Linked to"Correlation, not causation
"May cause"Possible but unproven
"Experts warn"Some experts, maybe fringe
"Study proves"Studies suggest, rarely prove

Questions to Ask

QuestionWhy It Matters
Who did the study?Conflicts of interest?
Where was it published?Reputable journal?
How big was the sample?Small = less reliable
Was it in humans?Mouse studies often don't translate
Was it replicated?Single studies are preliminary
What do other experts say?Consensus matters
What's the effect size?Is the difference meaningful?

Source Quality

More ReliableLess Reliable
Peer-reviewed journalsPress releases
Meta-analysesSingle studies
Expert consensusIndividual expert
Science reporters (specialists)General news headlines
Primary sourcesSocial media summaries

Red Flags for Misinformation

Warning Signs

Red FlagWhy Suspicious
"Miracle cure"Medicine doesn't work that way
"Scientists don't want you to know"Real science is published
"Natural therefore safe"Many natural things are dangerous
"Ancient wisdom"Old doesn't mean true
Testimonials as evidenceAnecdotes aren't data
Appeal to celebrityFame isn't expertise
"Just asking questions"Often hides agenda
"Do your own research"Implies conspiracy
Extreme certaintyReal science acknowledges uncertainty
"Suppressed" informationUsually means rejected for good reason

Logical Fallacies

FallacyDescriptionExample
Ad hominemAttack the person, not the argument"He's funded by pharma"
Appeal to natureNatural = good"Chemicals are bad"
Appeal to authorityFamous person says soCelebrity endorsement
Straw manMisrepresent opponent's position"Scientists think we came from monkeys"
False dilemmaOnly two options presented"Believe me or trust corrupt science"
Anecdotal evidencePersonal story as proof"My aunt was cured by this"
Moving goalpostsChanging standards when metAlways demanding more proof

Pseudoscience Markers

PseudoscienceReal Science
Unfalsifiable claimsTestable predictions
Immune to criticismWelcomes scrutiny
Relies on testimonialsRelies on data
Cherry-picks evidenceConsiders all evidence
Unchanging beliefsUpdates with new data
Conspiracy theoriesOpen publication
Personal attacks on criticsAddresses criticisms

Specific Topics

Health Claims

Claim TypeWhat to Look For
Supplement claimsFDA doesn't verify; "may support" is vague
Diet claimsOften based on weak evidence
"Detox" productsYour liver and kidneys already detox
"Boosts immunity"Usually meaningless
Cancer curesBe extremely skeptical

Evaluating Health Information

Good SourcePoor Source
Medical journalsHealth blogs
Major health organizationsSocial media posts
Your doctorCelebrities
Cochrane reviewsProduct websites

Climate Science

ClaimEvaluation
"Climate has changed before"True, but current change is different
"Scientists disagree"97%+ consensus on human causation
"It's the sun"Solar activity doesn't explain recent warming
"Models are wrong"Models have been remarkably accurate

Vaccines

Common ClaimReality
Cause autismLarge studies show no connection
"Too many, too soon"Immune system handles far more
Natural immunity betterRisks of disease far outweigh vaccine risks
Contain dangerous ingredientsIngredients are safe at given doses

Practical Application

Before Sharing

StepAction
1Check the source - is it reputable?
2Read beyond the headline
3Check the date - is it current?
4Look for the original study
5See what experts say
6Check fact-checking sites

Fact-Checking Resources

ResourceWhat It Covers
SnopesGeneral claims
PolitiFactPolitical claims
FactCheck.orgPolitical claims
Science-Based MedicineHealth claims
Climate FeedbackClimate claims
Full FactUK-focused

Talking to Others

ApproachWhy It Works
Ask questionsLess confrontational
Acknowledge valid concernsShows respect
Focus on shared valuesFind common ground
Provide contextHelp them evaluate claims
Be patientBeliefs change slowly

Updating Your Own Beliefs

PracticeBenefit
Seek disconfirming evidenceCounter confirmation bias
Notice when you're defensiveSignal to examine more closely
Change your mind publiclyModels intellectual honesty
Say "I don't know"Honest about uncertainty
Consider the oppositeChallenge your assumptions

The Scientific Consensus

What Consensus Means

AspectExplanation
Expert agreementMajority of qualified experts agree
Based on evidenceConclusions follow from data
Can changeUpdates with new evidence
Not a voteWeight of evidence matters
Doesn't mean certaintyBut reflects best current understanding

When to Trust Consensus

Trust Consensus WhenBe More Skeptical When
Field has been studied extensivelyTopic is new or controversial
Experts have relevant credentials"Experts" are outside their field
Multiple lines of evidence convergeBased on limited data
No major competing theoriesLegitimate scientific debate exists
Predictions have been validatedPredictions haven't been tested

Key Takeaways

  1. Not all studies are equal - Hierarchy of evidence matters; meta-analyses beat anecdotes

  2. Correlation is not causation - Just because two things happen together doesn't mean one causes the other

  3. Sample size matters - Larger samples are more reliable

  4. Relative risk can mislead - Always ask for absolute numbers

  5. Headlines exaggerate - Read the actual study or expert analysis

  6. Extraordinary claims need extraordinary evidence - Be proportionally skeptical

  7. Check multiple sources - No single source is always right

  8. Consensus exists for good reasons - Widespread expert agreement is meaningful

  9. Real scientists express uncertainty - Overconfidence is a red flag

  10. You can update your beliefs - Changing your mind with new evidence is strength, not weakness