Misleading numbers
30 cases in this category
Easy
What should you think about this 87% figure?
An ad for an energy gel announces: “87% of users felt more energized!” At the bottom of the page, in tiny print: “Survey of 45 people who had already purchased the product.”
- It's already suspect: only 45 people, and they're all buyers — not a genuine test audience
- It's reliable: 87% is a clear majority
- The math checks out, so the number must be accurate
See the answer and the explanation
The right answer: It's already suspect: only 45 people, and they're all buyers — not a genuine test audience
Those 45 people don't represent anyone: they're people who already paid for the product and want to feel good about their purchase. No control group, nobody else tested it for comparison. An impressive percentage built on a tiny, biased sample is worth nothing. The move: who actually answered? How were they chosen?
Related reflex(es) : 1. Who is speaking? · 9. Does the number have context?
Is the headline accurate?
A regional newspaper's headline: “ER admissions up 300% in a single month!” The article clarifies: “From 8 visits to 24 visits in January compared to the previous December.”
- No — mathematically correct but misleading: this is a tiny absolute increase in raw numbers, and this ER normally handles 800-1,000 cases a month
- No, the real increase is 16 visits, not 300
- Yes, going from 8 to 24 is a 300% increase, so it's correct
See the answer and the explanation
The right answer: No — mathematically correct but misleading: this is a tiny absolute increase in raw numbers, and this ER normally handles 800-1,000 cases a month
Tripling a tiny number is still a tiny number. Going from 8 to 24 visits at a center that handles 1,000 cases a month is just normal day-to-day variation. The percentage change hides the real scale. The move: find the starting point in absolute numbers, then judge whether the change is genuinely unusual or just noise.
Related reflex(es) : 9. Does the number have context? · 6. Too good to be true?
Why is this number misleading on its own?
A beverage brand announces: “Our customers buy from us an average of 4 times a month.” What that hides: within that same group, some customers buy 20 times a month, others just once. The average is papering over two completely different worlds.
- Because you'd need to know the total number of customers
- Because 4 times a month is way too much
- Because the average hides the spread: some customers are super-buyers, others are rare buyers. The genuinely interesting number stays invisible.
See the answer and the explanation
The right answer: Because the average hides the spread: some customers are super-buyers, others are rare buyers. The genuinely interesting number stays invisible.
An average of 4 could represent 1,000 customers who buy 4 times AND 10 customers who buy 100 times. The middle number hides the extremes. Brands love the average because it sounds reasonable. The move: ask for the median (the number that splits the group exactly in half) or the gap between the biggest and smallest buyers. That brings you back to reality.
Related reflex(es) : 9. Does the number have context?
What's missing to judge this number?
You read: “Sales for a watch brand grew 150% in three years.” No date, no starting number given. You have no idea whether that means going from 2 watches to 5, or from a million to 2.5 million.
- The currency exchange rate
- The absolute figures at the start AND the end: without them, 150% is just noise
- The name of the brand's founder
See the answer and the explanation
The right answer: The absolute figures at the start AND the end: without them, 150% is just noise
A percentage with no context is an empty shell. 150% of an unknown base tells you nothing. The move: always look for the raw numbers. What was there before? What's there now? Only with both figures can you judge whether the increase is real or theatrical.
Related reflex(es) : 9. Does the number have context? · 3. What's the date?
What's the visual trick here?
A chart shows temperature changes over a month. The Y-axis starts at 18°C (even though the lowest reading is 19.5°C) and tops out at 25°C. The day-to-day swings look enormous, with dramatic peaks and dips.
- The chart shows real, genuinely impressive swings
- The Y-axis doesn't start at zero: that amplifies small variations and makes them look dramatic
- There's no problem, the chart is accurate as it is
See the answer and the explanation
The right answer: The Y-axis doesn't start at zero: that amplifies small variations and makes them look dramatic
By starting the axis at 18°C instead of 0°C, a 2-3 degree change looks visually enormous. It's a classic trick for turning something mundane into something dramatic. The reader sees a roller coaster where there's really just normal variation. The move: always check where the axis starts and what the real numeric range is, not just how big it looks.
Related reflex(es) : 4. Does the image show what they claim? · 9. Does the number have context?
What should you think of the headline number?
An article about accidents claims: “12,000 scooter accidents this month!” Reading further: “That's the combined total for the last three months, nationwide.”
- 12,000 is the real figure, the headline is accurate
- The headline is accurate but the rest of the article is useless
- The headline hides the unit: it mixes a three-month total with a headline that talks about a single month
See the answer and the explanation
The right answer: The headline hides the unit: it mixes a three-month total with a headline that talks about a single month
Saying “this month” when you actually mean a three-month total is misleading. The real figure is closer to 4,000 a month, but the headline creates the impression of a single-month emergency. It's a deliberate mix-up between a running total and a single snapshot. The move: check whether the number being announced is yearly, monthly, weekly, or a cumulative total. That detail changes everything.
Related reflex(es) : 3. What's the date? · 9. Does the number have context?
Is this 20% figure usable?
A news outlet announces: “Crime is up 20% in the south-central region this year!” It turns out the data lumps together burglaries, minor offenses, and unpaid parking tickets — all mixed into a single number.
- No: it mixes completely different phenomena together. A rise in parking tickets isn't the same as a rise in assaults. The number confuses more than it clarifies.
- Yes, it's an official figure
- No, but only because it's too old to matter
See the answer and the explanation
The right answer: No: it mixes completely different phenomena together. A rise in parking tickets isn't the same as a rise in assaults. The number confuses more than it clarifies.
“Crime” is vague. Lumping together unpaid parking tickets, burglaries, and violent offenses, then averaging it all into one number, mixes categories that don't belong together. The real question: what, exactly, went up? The move: ask for the breakdown. Which type of offense? What's actually driving the increase? Otherwise the overall figure says nothing at all.
Related reflex(es) : 9. Does the number have context? · 2. What's the source?
Is this really a “historic record”?
A startup announces: “We hit a record: 500,000 downloads of our app in three months!” This is only its third year in existence. Across those three years, the first two quarters had 50,000 and 100,000 downloads respectively.
- Technically yes, but the record only spans three years — that's a short window to call it “historic” when it's really just a young company's first growth streak
- No, 500,000 is less than the true record, which is a million
- Yes, 500,000 is the biggest number this app has ever hit
See the answer and the explanation
The right answer: Technically yes, but the record only spans three years — that's a short window to call it “historic” when it's really just a young company's first growth streak
A “record” over just three years is lightweight marketing. If the app had existed for ten years with a flat trend before this spike, that would be a real record. Over three years, it's just normal growth. The word “historic” gives weight to something that hasn't earned it. The move: check how much time the record actually spans. Is one strong month over three years really “historic”? No.
Related reflex(es) : 3. What's the date? · 9. Does the number have context?
Why is this raw number hard to judge on its own?
An NGO report states: “2.3 billion people don't have access to clean drinking water. That's enormous!” No context about the world's population, no comparison offered anywhere.
- Because an NGO can't really count that high
- Because 2.3 billion is such a huge number that it must be a serious problem
- Because without knowing the world's total population (around 8 billion), you can't tell if that's 30% or 90% — the real scale stays fuzzy
See the answer and the explanation
The right answer: Because without knowing the world's total population (around 8 billion), you can't tell if that's 30% or 90% — the real scale stays fuzzy
A raw number with no scale for comparison just floats in a vacuum. 2.3 billion out of roughly 8 billion works out to about 29%: that means most of the world does have access. It's serious, but not universal. The number alone says “a lot” while hiding the actual proportion. The move: always look for the denominator — the total. What percentage does this represent? Compared to what?
Related reflex(es) : 9. Does the number have context? · 6. Too good to be true?
How can two papers say the opposite thing?
One newspaper writes: “Median pay dropped 3% this year.” Another, at the very same time: “Pay rose by an average of 2% this year.” Neither one is lying.
- Because the year isn't over yet for one of the papers
- One of the two must be wrong
- Because “average” pay can rise while “median” pay (the middle figure) falls — very high salaries can jump a lot, pulling the average up, while most people are actually earning less
See the answer and the explanation
The right answer: Because “average” pay can rise while “median” pay (the middle figure) falls — very high salaries can jump a lot, pulling the average up, while most people are actually earning less
The average and the median don't tell the same story. If executives get a 50% raise while 90% of workers earn 5% less, the average climbs (dragged up by the top earners) while the median falls (because most people are earning less). That's why you need both. The move: notice whether a figure is an “average” or a “median” — it changes everything. If it's an average, ask for the median too, and vice versa.
Related reflex(es) : 9. Does the number have context?
Medium
What's off about the causal story here?
A news article announces: “Study: drinking a glass of orange juice a day reduces wrinkles by 45%!” Further down: “A team of dermatologists tracked the habits of 1,200 women over two years and measured changes in their skin.”
- The study ran for too long
- The correlation observed — drinking orange juice and having fewer wrinkles — isn't necessarily causation: women who eat well and drink juice probably also drink plenty of water, use skincare, or simply have favorable genetics. The factors are all tangled together.
- The sample is too small to support such a big conclusion
See the answer and the explanation
The right answer: The correlation observed — drinking orange juice and having fewer wrinkles — isn't necessarily causation: women who eat well and drink juice probably also drink plenty of water, use skincare, or simply have favorable genetics. The factors are all tangled together.
Seeing two things together — drinking juice and having fewer wrinkles — doesn't prove that one causes the other. Women who choose to drink juice may also be the ones who take care of their skin, sleep well, and live with less stress. Juice is just one detail in what's really a healthier lifestyle overall. This is the classic correlation-versus-causation mix-up. The move: look for the hidden variables. What else could explain the link?
Related reflex(es) : 5. Who else is saying it? · 2. What's the source?
What's the methodological trap here?
A researcher publishes: “67% of social media users feel lonelier after an hour of use.” He asked participants every evening whether they felt lonely, right after having them scroll a short-video app for an hour.
- 67% isn't a high enough number to generalize from
- People who agree to take part in this kind of study are also agreeing to scroll on command — that's not normal use. And asking “do you feel lonely?” right after a loneliness-focused test shapes the answer. Measuring the feeling changes the feeling.
- The study only lasted one evening
See the answer and the explanation
The right answer: People who agree to take part in this kind of study are also agreeing to scroll on command — that's not normal use. And asking “do you feel lonely?” right after a loneliness-focused test shapes the answer. Measuring the feeling changes the feeling.
A survey that asks “do you feel lonely?” right after making people believe social media causes loneliness shapes the response. This is what's called an expectation bias. On top of that, someone who volunteers for a study about loneliness isn't a typical user to begin with. The study is built to confirm itself. The move: ask what normal life looked like before the study. Was there a baseline? How do we really know 67% are lonely?
Related reflex(es) : 3. What's the date? · 1. Who is speaking?
Why doesn't this 28% drop mean anything?
A city council announces: “Crime is down 28% this year compared to last year!” Digging further, you find that last year saw a major crackdown campaign with heavy reporting. This year, police were told to issue fewer citations and more warnings instead.
- Because 28% isn't a big enough number to be reliable
- Because one year isn't enough time to draw conclusions
- Because the way things are counted has changed: fewer cases are being logged, not fewer cases happening. The situation on the ground may have gotten worse even while the statistics drop by pure definition.
See the answer and the explanation
The right answer: Because the way things are counted has changed: fewer cases are being logged, not fewer cases happening. The situation on the ground may have gotten worse even while the statistics drop by pure definition.
Official numbers are shaped by how they're gathered. If the instruction becomes “stop citing minor offenses,” those offenses vanish from the count, but not from the streets. This council simply started counting less. It's a classic move in policing and health statistics: change how you measure without saying so. The move: find out what changed in the method. Who's logging the cases? What were they told to do? Did the real cases change, or just the numbers?
Related reflex(es) : 2. What's the source? · 9. Does the number have context?
How does this comparison rig the results?
An ad for a phone battery claims: “Our battery lasts 40% longer!” Reading the fine print: “Compared to an older model from a competing brand, measured with the screen off and network disabled.”
- The test conditions are completely unrealistic: nobody uses a phone with the screen off. And comparing against an old model from a rival brand is straight-up cherry-picking. The real question: how does it compare to my current phone, screen on, normal network use?
- It's fair: a battery that lasts longer is simply a better battery
- The battery can't really last 40% longer than claimed
See the answer and the explanation
The right answer: The test conditions are completely unrealistic: nobody uses a phone with the screen off. And comparing against an old model from a rival brand is straight-up cherry-picking. The real question: how does it compare to my current phone, screen on, normal network use?
A manufacturer wants its number to look as dramatic as possible. It picks the weakest opponent (an old rival model), the most favorable condition (screen off), and its best possible showing. This is called cherry-picking: handpicking the results that flatter you. An honest test would compare it to the same brand's previous model, screen on, normal use. The move: read the test conditions in the fine print. If they're strange or clearly tilted toward the seller, the number means nothing.
Related reflex(es) : 7. What is it selling me? · 2. What's the source?
What should you conclude from this 200% figure?
A report states: “Food-related accidents in the city have jumped 200% in a year.” Checking further: the city recently opened a reporting center for food poisoning that made filing a report much easier. Before that, people mostly didn't bother reporting at all.
- It mostly reflects a change in how things are counted: reporting used to be hard, now it's easy. Real accidents may have ticked up slightly, but mainly we're just seeing them now. The number doesn't say what people think it says.
- You'd need 300% for this to be a genuine problem
- There really are three times as many accidents now
See the answer and the explanation
The right answer: It mostly reflects a change in how things are counted: reporting used to be hard, now it's easy. Real accidents may have ticked up slightly, but mainly we're just seeing them now. The number doesn't say what people think it says.
What's rising is visibility, not necessarily real danger. When you make it easy to report something, reports go up — that's a pure counting effect. This is called a statistical artifact. The year before, people just stayed quiet or never filed a report at all. The move: whenever a number jumps sharply, check what changed in how it's measured before declaring a crisis. Was there a new rule? A new tool? An awareness campaign? That often explains 80% of the increase.
Related reflex(es) : 2. What's the source? · 3. What's the date?
Does this 15% figure really reflect discrimination for doing the same job?
A magazine's headline: “Women earn 15% less than men!” The number comes from median income reported across every category combined: employees, freelancers, retirees, and more. Women make up proportionally more of the retiree group (lower income) and fewer of the very top earners.
- No, it's simply because there are more women than men overall
- Yes, this is direct proof of discrimination
- No, not at all: the figure blends completely different populations together. A median retirement income isn't comparable to a salary; retirees earn less in general. And within salaried work alone, the gap would be much smaller. The move: make sure you're comparing apples to apples.
See the answer and the explanation
The right answer: No, not at all: the figure blends completely different populations together. A median retirement income isn't comparable to a salary; retirees earn less in general. And within salaried work alone, the gap would be much smaller. The move: make sure you're comparing apples to apples.
A pay gap exists and is well documented, but this particular number inflates it artificially by mixing categories that don't belong together. If you compared female managers to male managers, or nurses to their male counterparts in the same role, the gap would still be there, just far less dramatic. The raw headline number caricatures the issue. This is a case where a better number is available: the real gap exists but isn't 15% by this crude method. The move: ask for a like-for-like comparison. Same sector? Same experience level? Same hours?
Related reflex(es) : 9. Does the number have context? · 2. What's the source?
What should you think of this 6% figure?
A newspaper claims: “Average incomes rose 6% over the past decade.” It turns out the figure includes heavy inflation: real purchasing power barely moved, and it actually dropped for lower incomes.
- 6% is very little for ten years
- It's technically true but misleading: if inflation ate up 5 points of that increase, the real gain in purchasing power is only about 1%. The nominal figure (before inflation) flatters; the real figure (after inflation) is what actually counts.
- It's good news: 6% over ten years is genuine progress
See the answer and the explanation
The right answer: It's technically true but misleading: if inflation ate up 5 points of that increase, the real gain in purchasing power is only about 1%. The nominal figure (before inflation) flatters; the real figure (after inflation) is what actually counts.
Nominal statistics (in today's currency) can rise while real statistics (purchasing power) stagnate or fall. This is a fashionable trick: show numbers that shine on paper but weigh less in people's pockets. Meanwhile, people feel poorer — and in real terms, they're right. The move: always ask for the real, inflation-adjusted figure. And if that adjustment is never mentioned, it's often a sign the nominal number would be embarrassing.
Related reflex(es) : 9. Does the number have context? · 7. What is it selling me?
How does this 10x figure lie?
A startup announces: “We've grown 10x in three years!” No other data given. It turns out that in year 1 it had 2 customers, in year 2 it had 15, and in year 3 it had 20 — almost no growth in that final year. The “10x” comes from comparing year 1 against year 3, which inflates both ends artificially.
- It doesn't lie, it's mathematically accurate
- It's technically true in absolute terms, but it hides the real story: all the growth happened between year 1 and year 2, and year 3 was essentially flat. Saying “10x in three years” dramatizes a curve that's actually leveling off. The move: look at the year-by-year curve, not just the start-to-finish ratio.
- 10x is too much to be believable
See the answer and the explanation
The right answer: It's technically true in absolute terms, but it hides the real story: all the growth happened between year 1 and year 2, and year 3 was essentially flat. Saying “10x in three years” dramatizes a curve that's actually leveling off. The move: look at the year-by-year curve, not just the start-to-finish ratio.
A start-to-finish ratio inflates things: starting at 2 and ending at 20 sounds spectacular. But if the real story is 2, then 15, then 20, that means growth is running out of steam. A company slowing down in year 3 after a sprint in year 1 doesn't look like a company growing steadily. The move: demand the year-by-year breakdown, not just the start-to-finish ratio.
Related reflex(es) : 9. Does the number have context? · 3. What's the date?
Can both claims be true at the same time?
A health article claims: “Vaccinated people catch the virus at a rate 20% lower than the unvaccinated — this proves the skeptics wrong!” Another source states: “Among older adults, the vaccine is 95% protective, though that rate declines with age.”
- No, one of them has to be wrong
- 20% and 95% can't possibly come out of the same vaccine
- Yes: if 80% of vaccinated people are young and 70% of unvaccinated people are older, the average 20% gap mostly reflects the age difference between the two groups, not the vaccine's effectiveness within the same age bracket. It's a case of mixing different populations together.
See the answer and the explanation
The right answer: Yes: if 80% of vaccinated people are young and 70% of unvaccinated people are older, the average 20% gap mostly reflects the age difference between the two groups, not the vaccine's effectiveness within the same age bracket. It's a case of mixing different populations together.
The worst trap in any comparison: the two populations aren't identical. Vaccinated people include far more young people (naturally lower risk), unvaccinated people include more older people (naturally higher risk). Mix them together, and the 20% statistic becomes an artifact. The real figure — the vaccine's effect within the same age group — is far higher. This is called confounding bias. The move: in every comparison, check for gaps in age, prior health, or exposure. Otherwise the number is poisoned.
Related reflex(es) : 2. What's the source? · 9. Does the number have context?
What should you conclude from this 87% figure?
A source claims: “A researcher found that natural substance X fights cancer with 87% effectiveness!” You read the original paper: the study covers 34 lab mice, not humans.
- That's good news, there's real hope here
- 87% seems suspiciously high
- It's a lab result that could be a promising start, but going from a mouse to a human is a huge leap. The number only describes mice; extrapolating it to people is an enormous jump. The move: understand the study's stage. Animals or humans? Lab or real life?
See the answer and the explanation
The right answer: It's a lab result that could be a promising start, but going from a mouse to a human is a huge leap. The number only describes mice; extrapolating it to people is an enormous jump. The move: understand the study's stage. Animals or humans? Lab or real life?
Lab results in animals matter for exploring possibilities, but they almost never carry over exactly to humans. Our bodies are far more complex machines; something that works in a test tube can be ineffective, or even toxic, in a person. That's exactly why clinical trials exist: to check what actually holds up. The move: find out what stage the research is at. Lab? Animals? Humans? The meaning of a number changes completely depending on the stage.
Related reflex(es) : 2. What's the source? · 8. Is the expert really an expert?
Tough
Is this really an explosion?
A polling institute announces: “Reports of online abuse have exploded: up 89% in three years!” You look for the raw numbers: the site received 4 reports the first year, 6 the second, 7 the third. Nobody else is reporting any such explosion.
- No, because the survey was conducted online
- No: going from 4 to 7 cases is a tiny shift, barely visible at all. On a real base of thousands, nobody would even mention it. The impressive percentage hides the fact that the base was so small that adding three cases is enough to nearly double it.
- Yes, 89% is massive
See the answer and the explanation
The right answer: No: going from 4 to 7 cases is a tiny shift, barely visible at all. On a real base of thousands, nobody would even mention it. The impressive percentage hides the fact that the base was so small that adding three cases is enough to nearly double it.
A dramatic percentage on a tiny base is worth very little. 89% of 4 cases equals 3.5 extra cases. Going from 4 to 7 isn't an explosion, it's statistical noise. On a base of 4,000 growing to 7,560, that would be a real increase. The move: always look for the raw numbers first. If the base is tiny, the percentage tells you nothing.
Related reflex(es) : 9. Does the number have context?
Why is this average misleading?
A school announces: “The average age of our job-placement students is 31 — perfect for our continuing-education and work-study programs!” In reality, the group has 200 students: 80 are between 16 and 24, and 120 are between 45 and 60.
- Because the real students are younger than they used to be
- Because not a single student is actually 31: the average masks two radically different populations — young people starting out, and older adults retraining. This is a real case of two separate peaks hiding behind one number.
- Because 31 is too old for higher education
See the answer and the explanation
The right answer: Because not a single student is actually 31: the average masks two radically different populations — young people starting out, and older adults retraining. This is a real case of two separate peaks hiding behind one number.
An average drawn from two far-apart groups invents a phantom person in the middle. The school is mostly made up of very young students and near-retirees, but the average claims “middle-aged group.” This is a classic trap in HR and demographic marketing. The move: ask for the median (the number that splits the group exactly in half) and the breakdown by age bracket. That tells the real story.
Related reflex(es) : 9. Does the number have context?
Can you conclude that studying late CAUSES better grades?
A newspaper headline: “Studying at night means better grades. The more students study after 8pm, the better they do!” The figure comes from a real observed correlation: night-owls average 2 points higher.
- Yes, the correlation is measured and real
- No, because grades never change
- No: the correlation is real, but it's a causation trap. Strong students study more overall, including at night. The late hour is just a trace of how hard-working they already are. Cutting into sleep to study late would likely backfire.
See the answer and the explanation
The right answer: No: the correlation is real, but it's a causation trap. Strong students study more overall, including at night. The late hour is just a trace of how hard-working they already are. Cutting into sleep to study late would likely backfire.
Seeing two things together — studying late and good grades — doesn't prove one causes the other. Hard-working, high-achieving students study more, including late at night, and including other, healthier hours too. The late hour is a symptom of the effort, not the cause of the success. This is the classic mistake: confusing correlation with causation. The move: always look for the hidden variable (here: work ethic, discipline). What else could explain the link?
Related reflex(es) : 5. Who else is saying it? · 9. Does the number have context?
Is the growth really exponential?
A chart showing audience growth displays a curve that shoots up spectacularly. The data points are: 10 (year 1), 32 (year 2), 1,000 (year 3). The Y-axis is logarithmic — which is never labeled anywhere on the chart.
- Impossible to say without reading the axis: if it's logarithmic, you're actually looking at a straight line — steady percentage growth, not an exponential curve. If it's a normal scale, it is exponential. Reading the axis is essential, and here, it's missing.
- Yes, the curve looks spectacular
- No, 1,000 isn't that big a number
See the answer and the explanation
The right answer: Impossible to say without reading the axis: if it's logarithmic, you're actually looking at a straight line — steady percentage growth, not an exponential curve. If it's a normal scale, it is exponential. Reading the axis is essential, and here, it's missing.
A logarithmic scale flattens out large numbers. The exact same curve looks smooth on a log scale and explosive on a linear one. This is a presentation trick: making the mundane look dramatic, or the reverse. This particular chart might be perfectly honest, but it's hiding the key to reading it. The move: always check whether the Y-axis is linear or logarithmic. And if whoever's presenting it forgets to say, that's often not an accident.
Related reflex(es) : 4. Does the image show what they claim? · 9. Does the number have context?
Should this be alarming?
A health channel announces: “The risk of a rare post-vaccination complication has DOUBLED: it went from 0.001% to 0.002%.” Headline: “Risk multiplied by two — watch out!”
- No: this is a relative-risk-versus-absolute-risk trap. Doubling a vanishingly small risk (0.002% is 1 person in 50,000) is still vanishingly small. The headline dramatizes the ratio while burying the real scale — it should really say “one extra case per 50,000 doses.”
- No, because no risk at all should ever be considered acceptable
- Yes, a risk that doubles is serious
See the answer and the explanation
The right answer: No: this is a relative-risk-versus-absolute-risk trap. Doubling a vanishingly small risk (0.002% is 1 person in 50,000) is still vanishingly small. The headline dramatizes the ratio while burying the real scale — it should really say “one extra case per 50,000 doses.”
Saying “the risk doubled” sounds dramatic. Saying “it went from 1 in 100,000 to 2 in 100,000” makes it clear. Both are true, but they land very differently. This is a classic fear-marketing move: amplify the ratio while hiding the real scale. It's mathematically honest and deeply misleading at the same time. The move: always demand the risk in absolute numbers, not as a ratio. How many people are genuinely affected, out of how many?
Related reflex(es) : 6. Too good to be true? · 9. Does the number have context?
How do you explain this gap?
A market survey announces: “73% of consumers say they'd be willing to buy an eco-friendly brand even if it costs 15% more.” The survey itself is methodologically sound. And yet, in stores, eco-labeled products of the same type capture only 8% of sales.
- The 73% must be shopping somewhere else
- The survey is lying — people really only care about price
- The survey measures an intention (“I'd be willing to”) but the actual behavior is different: at the checkout, people go for the cheaper option. It's the classic gap between what people say they want and what they actually do.
See the answer and the explanation
The right answer: The survey measures an intention (“I'd be willing to”) but the actual behavior is different: at the checkout, people go for the cheaper option. It's the classic gap between what people say they want and what they actually do.
Saying “I'd pay more for the greater good” is easy. Actually paying more when it hits the monthly budget is another matter entirely. This is the classic weakness of surveys: they measure stated intentions, not real behavior. Life is tight; values often bend under the weight of the wallet. The move: be wary of “would you” surveys. Look for actual behavior: what do people really buy? What price do they really pay?
Related reflex(es) : 1. Who is speaking? · 6. Too good to be true?
Is this comparison valid?
A research group compares two regions: “Region A has 45% of its population with a college degree, Region B only 28%. That's a massive education gap!” Later, you find out that A is a large urban center and B is rural, and that the definition of “resident” includes or excludes students depending on the year measured.
- No, because averages never change
- No: the two areas are radically different in scope — urban versus rural, a shifting population versus a stable one. Comparing them without matching the method means comparing two different kinds of numbers. You'd need the same boundaries, the same definition of “resident,” the same age brackets.
- Yes, 45% versus 28% is a real, meaningful gap
See the answer and the explanation
The right answer: No: the two areas are radically different in scope — urban versus rural, a shifting population versus a stable one. Comparing them without matching the method means comparing two different kinds of numbers. You'd need the same boundaries, the same definition of “resident,” the same age brackets.
Two regions with opposite geographies, economies, and demographics don't compare cleanly on a raw ratio. It's noise. You'd need to control for the same age bracket, the same definition, the same measurement year, maybe even weight by industry. Region A attracts young, educated urban residents; Region B has an older population. That's normal, but it isn't an apples-to-apples gap. The move: before any “A versus B” comparison, check that you're measuring exactly the same thing. If the scope differs, the comparative number is worth nothing.
Related reflex(es) : 9. Does the number have context? · 2. What's the source?
What's the headline's real sin?
A headline announces: “17,000 cases of this syndrome since the start of the decade!” Reading further: that's a nine-year stretch, and the cases are spread out evenly across it. In reality: roughly 1,900 cases a year, or about 5 a day nationwide.
- It's a made-up number, 17,000 is too high to be real
- The headline hides the unit: it cumulates nine years into one alarming sentence. 17,000 sounds like an emergency. “1,900 a year” or “5 a day” tells a very different story — routine, predictable, under control.
- 17,000 over nine years is actually pretty low, so it's not a big deal
See the answer and the explanation
The right answer: The headline hides the unit: it cumulates nine years into one alarming sentence. 17,000 sounds like an emergency. “1,900 a year” or “5 a day” tells a very different story — routine, predictable, under control.
Adding up years of data creates an impressive-sounding number. 17,000 over nine years works out to about 1,900 a year. Presenting it as a lump sum without saying so sells urgency where there's actually a steady, routine pattern. This is a classic trap in health, crime, and accident reporting: piling numbers up without dividing them creates artificial fear. The move: always check what time period a big number covers. Is it one year? Ten years? One month? Divide by the number of periods to see the real underlying rate.
Related reflex(es) : 9. Does the number have context? · 3. What's the date?
Why does “historic record” overstate things here?
An economist publishes a note: “This year: a historic growth record for region X!” You check the available data: it only goes back three years. The values are: year 1: 3.1%, year 2: 2.8%, year 3: 3.4%. It genuinely is the highest of the three, but that's all there is to compare it against.
- Because 3.4% is a small number
- Because growth is bound to slow down again afterward
- Because a “record” over just three years of data isn't a real record. If the data went back thirty years and this year were the highest in three decades, that would be a record. Here, it's just the peak of a series too short to call “historic.”
See the answer and the explanation
The right answer: Because a “record” over just three years of data isn't a real record. If the data went back thirty years and this year were the highest in three decades, that would be a record. Here, it's just the peak of a series too short to call “historic.”
The word “historic” demands depth over time. A record over three years is just a fluctuation. Over thirty years, it's a genuine record. This is a classic move in finance and economics: people call something a record based only on the period they happen to have data for. The move: always look for the real historical record. Does it go back ten years? A hundred? Just three? The shorter the series, the less weight the word “record” carries.
Related reflex(es) : 3. What's the date? · 9. Does the number have context?
Is this figure reliable enough to draw conclusions from?
A government agency publishes: “According to the annual survey of 8,500 households (margin of error ±1.2%), the relative poverty rate this year is 13.7%, up 0.8 points from last year. Methodology: national statistics agency survey, scope: mainland territory.”
- No, because it comes from the government — it must be biased
- No, because 13.7% sounds like a suspiciously round number
- Yes, largely: a solid sample, a reputable source, a stated margin of error, transparent methodology, a clear scope, and a valid year-over-year comparison. This is a well-built number, properly framed.
See the answer and the explanation
The right answer: Yes, largely: a solid sample, a reputable source, a stated margin of error, transparent methodology, a clear scope, and a valid year-over-year comparison. This is a well-built number, properly framed.
This case is the opposite of the previous ones: a genuinely solid number. Traceable source, open methodology, an honestly stated limitation (margin of error), a clear definition (relative poverty, mainland territory). You can trust this and use it for real decisions. Not every number is a trap; this one is built to be used. The move: learn to recognize well-presented numbers. Reputable source plus visible methodology plus stated limits equals fine to use.
Related reflex(es) : 2. What's the source? · 9. Does the number have context?