Most misleading numbers aren't made up. They're real figures, framed to push you to a conclusion the data doesn't actually support. Once you've seen the handful of standard moves, you can spot them everywhere — in ads, headlines, and posts — and they stop working on you.
See it for yourself: the cut-off axis
The most common chart trick is starting the scale somewhere other than zero. Here's the same two numbers — 102 and 105 — drawn both ways. Press the button and watch a three-percent difference turn into a landslide.
More tricks worth knowing
The scary percentage with no base rate
"Doubles your risk!" sounds alarming — but doubling a 1-in-a-million chance is still 2 in a million. A relative change (doubled, +50%) means little without the absolute number it started from. Always ask: from what to what?
Correlation dressed up as cause
Two things rising together doesn't mean one caused the other — both may be driven by a third, or it may be coincidence. "People who do X live longer" often means healthier people do X already. Ask what else could explain it before believing the arrow.
The cherry-picked sample
Start the graph at the one good year; survey only your happiest customers; show the five winners and not the thousand who lost. A true number from an unrepresentative slice tells you nothing about the whole. Ask: who or what got left out?
The average that hides the spread
"Average" can mean the mean, which a few huge values yank upward — put a billionaire in a room and the average wealth says everyone's rich. The median (the middle person) is often the honest figure. Ask which average, and how spread out the data is.
Percent vs. percentage points
A rate going from 4% to 5% is a one percentage-point rise — but you can also call it a "25% increase." Both are true; one is chosen to sound bigger. Watch which is being used.
The big number with no denominator
"40,000 incidents!" means nothing until you know: out of how many, over what period, compared to when? A huge raw count can be a tiny rate. Always look for what it's divided by.
Questions to ask any statistic
Run a suspicious number or chart through these. Tick the ones you can't answer — each is a reason to doubt it.
The close
You don't need to be "good at maths" to stop being fooled by numbers — almost none of this is arithmetic. It's the habit of asking what's missing: the baseline, the denominator, the comparison, the people left out. The figure on the screen wants you to leap to a conclusion. Slow down, ask for the part it didn't show you, and the leap usually disappears.