Association-to-Causation Evidence Check
Match confidence to study design before turning a finding into advice.
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 5
- Confidence
- 98%
When Bartlett asks whether exercise caused younger-looking telomeres, Sinclair says the cited association cannot establish causation. Exercisers may also eat differently, sleep better, or share other characteristics that influence the result. His decision rule is to identify the design, name likely confounders, and then look for prospective or controlled comparisons in which behavior differs by study assignment. Even stronger designs must be interpreted according to what they actually measured: a biomarker, function, disease event, or mortality. This framework turns a conversational caveat into a reusable evidence check. It does not label every observational study useless; it limits the conclusion to association until stronger evidence supports a causal claim.
Origin
Sinclair applies this evidence check on The Diary of a CEO when Bartlett challenges a reported link between regular exercise and younger-looking telomeres.
Core principles
- 01An association does not show that the measured behavior caused the outcome.
- 02People who adopt one healthy behavior may differ in many other ways.
- 03Prospective and controlled evidence can reduce, but not eliminate, uncertainty.
- 04Conclusions should remain proportional to the study design and endpoint.
How to run it
- 1
Name the design
Determine whether the finding comes from an observational association, a prospective study, or a controlled intervention. Do not infer the design from a headline.
Pro tip Find what participants actually did and who decided the exposure.
Watch out Words such as linked or associated do not establish causation.
- 2
Map confounders
List other differences that could influence both the behavior and the result. Consider diet, sleep, income, baseline health, medication, and self-selection where relevant.
Pro tip Ask what else tends to travel with the behavior being studied.
- 3
Seek stronger comparisons
Look for studies that assign or prospectively track the intervention and use an appropriate comparison group. Check whether findings replicate across designs.
Pro tip Prefer converging evidence over one dramatic result.
Watch out A controlled study can still be small, short, biased, or poorly matched to the question.
- 4
Inspect the endpoint
Distinguish a surrogate marker such as telomere length from symptoms, function, disease, or lifespan. Match the claim to the outcome that was directly measured.
Pro tip Rewrite the conclusion using the study's exact population and endpoint.
Watch out A changed biomarker does not automatically mean a person will live longer.
- 5
Calibrate the conclusion
State what the evidence supports, what it does not, and what would reduce uncertainty. Keep personal recommendations within the evidence and relevant clinical guidance.
Pro tip Use may, is associated with, or caused according to what the design can justify.
In the wild
Bartlett cites a study associating regular jogging with telomeres that looked younger than those of sedentary people. Sinclair says that study alone cannot show exercise caused the difference because exercisers may also eat and sleep differently. He then points to intervention studies comparing people instructed to exercise with those instructed to sit as better causal evidence.
→ The claim moves from a broad causal headline to a design-calibrated conclusion.
Common mistakes
Promoting correlation to causation
A difference between existing groups may reflect many behaviors and baseline characteristics besides the one highlighted.
Treating every endpoint alike
A molecular marker, reported symptom, clinical event, and lifespan are not interchangeable outcomes.
Calling one trial definitive
Controlled design improves inference but does not remove limits involving sample, duration, execution, and replication.
Is it for you?
Best for
Reviewing health headlines, podcast claims, and studies before changing behavior or communicating a conclusion.
Not ideal for
Replacing specialist appraisal when treatment decisions, complex statistics, or high-risk medical choices are involved.
From the transcript
“we don't know for sure, speaking like a scientist”
“you have to be careful interpreting these association studies always”
From the episode
David Sinclair: Can Aging Be Reversed? After 8 Weeks, Cells Appeared 75% Younger In Tests!