Politicians make me queasy. So do self-help books, Twitter wars, linguistics professors, even TED talks. It’s not that they lack value—Twitter excepted, of course—it’s that they rely too heavily on examples and stories to persuade.

Why? It’s effective. As Chip and Dan Heath put it in Made to Stick (one of those self-help books), anecdotes—simple, concrete, and relatable—”have the amazing dual power to simulate and to inspire.” But such a quote isn’t as persuasive as the real examples they offer in the book, like how Steve Denning managed to convince the World Bank senior leadership to restructure the organization by sharing the story of a single Zambian health worker searching for malaria treatment info. Stories are more convincing than statistics.

But with billions of people and anecdotes in the world, you can cherry-pick one for almost any argument you’d like. So for every story Elizabeth Warren tells about an immigrant who lost everything to Trumpian cruelty, I’m sure Donald Trump could find a loyal storyteller who lost everything to immigrants. It breaks my heart that stories outpersuade statistics: in other words, that people are human.

The issue, to me, is that people (a) don’t remember numbers, and (b) distrust statistics. Point (a) is probably an irreversible pillar of human nature that will always sway the persuasion pendulum in favor of stories. For point (b), anecdotes often seem more trustworthy because they happened to real, imaginable people, while we generally don’t know where most of the stats we consume come from. The Heath brothers put it better than I can: “tinkering with statistics provides lucrative employment for untold numbers of issue advocates. Ethically challenged people with lots of analytical smarts can, with enough contortions, make almost any case from a given set of statistics.” But the same thing happens with stories! It seems that the problem with statistics is their deceptive pretense to be fact, while apparently stories don’t start off with this claim in the first place.

Of course, it would be impossible and non-productive to qualify every argument with accurate statistics. I didn’t begin this post with, “Seventy-two percent of politicians make me queasy (standard error of 8.2 percentage-points).” But education in statistics and statistical ethics can make us better listeners and communicators. Our statistics should be prospective and inclusive, and we must have the flexibility to accept and share statistics that don’t support our arguments.

If you’re trying to convince someone, stories are your best bet. But to understand an issue, statistics can be skyscrapers built of story-bricks. For some final advice, I’ll turn to the Heath brothers one last time: “when it comes to statistics, our best advice is to use them as input, not output. Use them to make up your mind on an issue. Don’t make up your mind and then go looking for the numbers to support yourself—that’s asking for temptation and trouble.” But don’t do that with stories, either.

---

Quotes are cherry-picked from the popular book by Chip and Dan Heath, Made to Stick: Why some ideas take hold and others come unstuck (Penguin Random House, 2007), pages 237 and 147.