Numbers that bend

A number arrives in a headline and it feels like the end of the argument. Words can be slanted, framed, spun — but a number is just a number, cold and final. That’s exactly the trap. A single true figure can be dressed to alarm you or to calm you, and both outfits are honest. Nobody has to lie. They just choose which true face of the number you see. Learn the handful of moves and you stop being startled by your own arithmetic.

Percentage points are not percent

Start with the move that trips almost everyone. Imagine unemployment rises from 5.0% to 6.0%. How big is that jump? There are two true answers.

It rose one percentage point. That’s the plain subtraction: 6.0 minus 5.0 equals 1.0. Calm, small, almost dull.

Or: unemployment leapt 20%. That’s also true. The rise of 1.0 is being measured against the old level of 5.0, and 1.0 divided by 5.0 is 0.20 — twenty percent. The same single fact, and the second version sounds like a crisis.

Neither is a lie. They measure different things. “Percentage points” is the gap between two percentages. “Percent” is how big a change is relative to where it started. A writer who wants alarm reaches for the bigger-sounding “20%.” A writer who wants calm reaches for “one point.” Watch which one they pick, and you’ve caught the thumb on the scale.

A giant percent can hide a tiny thing

Here’s the second move. A rare side-effect of a medicine went from 1 case to 3 cases across a whole year.

You could write: “Cases up 200%!” True. Going from 1 to 3 is a rise of 2, and 2 is two hundred percent of 1. The number is real and it sounds terrifying.

Or you could write: “the side-effect rose from 1 person to 3, in a city of millions.” Also true — and now it sounds like almost nothing, because it almost is.

The percent is doing a magic trick: it hides the base, the size of the thing you started with. A huge percentage of a tiny number is still a tiny number. “Doubled” sounds enormous and can mean two people became four. Whenever a percent shocks you, ask the quiet question the percent was designed to stop you asking: two out of how many?

Pick the window, pick the story

The third move is about time. Take a city’s crime index. Ten years ago it sat at 100. It fell, year after year, down to 55 last year — its lowest point ever. This year it ticked up slightly to 60.

Now watch one true dataset tell two opposite stories, just by choosing where to start the clock.

Start the clock last year: crime went from 55 to 60. That’s a rise of 5, and 5 divided by 55 is 0.0909 — about 9 percent. Headline: “Crime up 9%.” A city getting more dangerous.

Start the clock ten years ago: crime went from 100 to 60. That’s a drop of 40, and 40 divided by 100 is 0.40 — forty percent. Headline: “Crime down 40% over the decade.” A city getting steadily safer.

Both numbers are correct. Both describe the same city, the same crime, the same year. The only difference is the starting line — and whoever chooses it chooses the story. A short window from a low point manufactures an alarming rise. A long window swallows the same blip whole. When a trend headline surprises you, find out which window it cut from, and ask what the line looks like just outside the frame.

The number with no bottom

The last move is the simplest and the most common. A report says: “500 incidents on the rail network this year.” Five hundred. It lands like a warning.

But 500 of what? That’s the missing piece: the denominator, the total you’d divide by to know if 500 is a lot. Suppose the network carried 50 million journeys that year. Now do the division: 500 divided by 50,000,000 is 0.00001 — that’s one incident for every hundred thousand journeys. The same 500, against its real backdrop, becomes a sign of a remarkably safe network.

A raw count with no denominator is a number floating in the dark. “Deaths rose,” “complaints doubled,” “thousands affected” — they all sound like something until you ask out of how many, over what period. The count alarms; the rate informs. The two can point in opposite directions from the very same event.

A number is a number plus a choice

Step back and the four moves are one move. Percentage points versus percent, a fat percent on a thin base, a hand-picked window, a missing denominator — each takes a single honest figure and chooses the frame that makes it shout or whisper. This is the same craft lesson 4 showed with words: the facts stay true while the shaping does the persuading. Numbers just feel more immune, so the shaping slips past more readers.

So a number is never only a number. It is a number plus a decision about how to show it — what to compare it to, where to start counting, what to leave out of the bottom. None of that lives in the figure itself; all of it lives in the person who placed it. You’re not being lied to. You’re being pointed — and the figure that pointed you was chosen by someone with a story in mind, who is themselves usually working from a number someone handed them. Knowing the moves won’t tell you what’s true. But it lets you ask the figure the questions it was dressed to dodge — and that is the whole difference between reading a number and being moved by one. Next we follow the shaping to its quietest form of all: not how a fact is bent, but the facts that never appear at all.