Claims and evidence
Tell a checkable claim from an opinion, and rank evidence from a friend's story up to many studies agreeing.
THREE KINDS OF CLAIM
Fact, opinion, prediction
A claim is any statement that something is so. A factual claim can be checked, at least in principle. An opinion or value judgement is about taste or what matters. A prediction is about the future, so nobody can check it yet. Many posts mix all three in one sentence, and the first skill is to pull out the checkable part.
"Mount Damavand is over 5,000 metres high" is factual (it's about 5,600 m). "Damavand is the most beautiful mountain" is an opinion. "Climbing permits will double in price next year" is a prediction.
Check yourself
A forwarded message about a new herbal drink. Sort each piece.
- The drink went on sale in Tehran in March
- It tastes much better than green tea
- Soon every pharmacy will stock it
- It contains ginger and turmeric
- Its bottle is the prettiest on the shelf
- Its price will double by winter
Show the answer
Checkable fact: The drink went on sale in Tehran in March, It contains ginger and turmeric
Opinion: It tastes much better than green tea, Its bottle is the prettiest on the shelf
Prediction: Soon every pharmacy will stock it, Its price will double by winter
Vague words hide the claim
Vague
"Experts say this drink may help, and it's linked to a lower risk of colds, up to 80%."
Which experts? Help with what? Linked how? Up to 80% could mean 2%.
Precise
"In one study of 200 adults, people who drank it daily for three months had fewer colds than people who didn't."
Now you know who, what, how many and compared to what. Now you can check it.
Claim: "Screens are ruining kids' eyes, doctors warn."
Who? which doctors, where?
What? which eye problem, exactly?
How many? how many kids, how much change?
Compared to? kids with less screen time?
Since when? over what period?Rewritten: "Children who spend more than X hours a day
on screens are more likely to be short-sighted than
children who spend less, according to study Y."You don't need the answers yet. Just asking the five questions turns a scary slogan into a sentence you can look up.
Check yourself
Three comments under a phone review. Which one is a checkable factual claim?
- "This phone is ugly."
- "This phone was released in 2023."
- "This phone will be a huge hit."
- "This is the best phone for anyone who loves photos."
Show the answer
"This phone was released in 2023."
Right. A release date can be looked up. It might even turn out wrong, but it's the kind of thing that can be settled.
WHO HAS TO SHOW IT
The burden of proof
The person making a claim is the one who has to bring the evidence. "Prove it's false" is not evidence that it's true: you can't disprove every invented cure, and that doesn't make any of them work. And the bigger the surprise, the stronger the evidence needs to be. Carl Sagan popularised it as extraordinary claims require extraordinary evidence.
"This café opened last week" needs little: a photo, a map listing. "This tea cures diabetes" would overturn a lot of medicine, so it needs strong, repeated trials.
Check yourself
A post says: "Nobody has proved this cure doesn't work, so it probably does." That's a fair argument, because the doubters haven't brought any evidence.
Show the answer
False
False. The burden of proof is on whoever claims the cure works. "No one has disproved it" is true of countless invented remedies. Until someone shows it works, the honest position is "not shown", not "probably true".
A rough ladder of evidence (for health-type claims)
- A personal story or testimonial
"It worked for me." Real, and still one case with nothing to compare against.
- An expert's opinion without data
Better informed than a stranger's story, but an opinion is not a measurement.
- One observational study
Researchers watch what people already do and compare groups. Useful, but the groups may differ in other ways.
- One randomised controlled trial
People are split by chance into a group that gets the treatment and a group that doesn't, so the comparison is fairer.
- Several studies agreeing, or a systematic review
Different teams, different people, the same answer. This is where you can start to feel fairly sure.
Check yourself
Put these in order from weakest to strongest evidence: several trials that agree, a customer's testimonial, one randomised trial.
- One randomised trial, a testimonial, several trials that agree
- A testimonial, several trials that agree, one randomised trial
- A testimonial, one randomised trial, several trials that agree
- Several trials that agree, one randomised trial, a testimonial
Show the answer
A testimonial, one randomised trial, several trials that agree
Yes. One story is the bottom rung, one fair comparison is better, and several fair comparisons agreeing is best.
ONE CASE
Anecdote versus data
An anecdote is a single story. It can be completely true and still tell you almost nothing about a pattern, because one case can't show how often something happens or what would have happened otherwise. Data looks at many people and compares groups.
"My uncle smoked all his life and lived to 95" may be true. It says nothing about the risk, which only shows up when you compare thousands of smokers with thousands of non-smokers.
Check yourself
- A forwarded post: five people say a herbal paste cleared their back pain within a week. Each story comes with a smiling photo.
- A trial: 300 people with back pain are split by chance. Half use the paste, half use a plain cream. After a week, both groups report about the same improvement.
Why do the two tell such different stories?
- The five people in the post must be lying
- The post has no comparison group, and back pain often eases on its own within a week
- 300 people is too many for a fair test
- Plain cream secretly contains the same herbs
Show the answer
The post has no comparison group, and back pain often eases on its own within a week
Exactly. The testimonials can be sincere. Without a group that didn't use the paste, you can't tell the paste from time passing. The trial shows the improvement happened either way.
Three common mistakes
- Treating "a study says" as settled. One study is a starting point.
- Treating an opinion as a lie. Opinions aren't true or false the way facts are.
- Treating "no evidence yet" as "evidence it's false". Sometimes nobody has checked.
Check yourself
In the family group, Aunt Mina writes: "My grandmother drank this tea every morning and never got sick." What is this, as evidence?
- Proof that the tea prevents illness
- An anecdote: one case, with no comparison
- An expert opinion, because she saw it first-hand
- A small observational study
Show the answer
An anecdote: one case, with no comparison
Right. It may be a lovely, true memory. It's still one person, with no way to know how she'd have done without the tea.
Lesson recap
- A claim can be a checkable fact, an opinion or a prediction; pull out the checkable part first.
- Checkable doesn't mean true. It means you could, in principle, find out.
- Rewrite vague claims: who, what, how many, compared to what, since when.
- The person making the claim brings the evidence, and surprising claims need stronger evidence.
- Evidence climbs from one story to many studies agreeing; one case can't show a pattern.