In 1954, a freelance writer named Darrell Huff published a slim, funny little book called How to Lie with Statistics. It wasn't written by a statistician — Huff was a journalist who'd noticed how often numbers got twisted in newspapers and advertising without anyone technically lying. The book went on to sell over a million and a half copies and became a standard classroom text for decades, because the tricks it described never really went out of style.
One example from the book: a 1950s survey of Cornell graduates found that a much higher percentage of middle-aged male alumni were married than female alumni. A magazine writer at the time jumped straight to the conclusion that going to college hurt a woman's chances of marriage. The statistic was real. The conclusion didn't follow from it — there were plenty of other explanations (career-focused women marrying later, for one) that fit the same numbers just as well. The number was true. The story built on top of it wasn't.
Here are five versions of that same basic move, still everywhere today:
1. Correlation dressed up as causation. Two things moving together doesn't mean one caused the other. Ice cream sales and drowning deaths both rise in summer — one doesn't cause the other; heat causes both. The Cornell example above is this same trick with better packaging.
2. A relative number without its absolute baseline. "This doubles your risk!" sounds alarming — until you learn the risk went from 1 in a million to 2 in a million. Both the relative claim (100% increase) and the absolute numbers are true. Only one of them tells you anything useful about whether you should actually worry.
3. A cherry-picked timeframe. Any chart can be made to show growth, decline, stability, or crisis, purely by choosing where it starts and ends. The data isn't fabricated — the window around it is doing all the persuading.
4. An average that hides the real story. "Average income rose" can be true even if most people's income didn't move at all, because a handful of very large gains at the top can drag the average up. A median (the true middle value) often tells a very different, and more representative, story than a mean.
5. A graph with a chopped axis. Start a bar chart's Y-axis at 90 instead of 0, and a 2-point difference looks like a landslide. Nothing on the chart is false. The scale is just doing work the numbers alone wouldn't do.
None of this means "don't trust statistics." It means a number can be completely accurate and still be arranged to make you feel a certain way about it — the same lesson as spotting framing in a headline, just with charts instead of words. The fix is the same instinct either way: before reacting to a statistic, ask what it's being compared to, over what timeframe, and what got left off the chart.
One footnote worth knowing: Huff, the author who taught a generation to spot these tricks, later used some of the same statistical skepticism professionally on behalf of the tobacco industry, testifying in Congress to cast doubt on the link between smoking and cancer — a link that was, in fact, real. It's a useful reminder that "just asking questions about the statistics" isn't automatically virtuous. The same tools that protect you from being fooled can also be pointed the other way, by someone with a reason to muddy a fact that was never actually in doubt.
Solitaire FactStack drops one media literacy fact into every game of Klondike solitaire you play — free, ad-supported, and built to make you a little sharper one hand at a time. Play free now →