Numbers Don’t Have Faces
The paradoxical power of anecdotes.
by Abel Gustafson
August 2026
When I was a grad student trotting around campus with all the wisdom of a fawn in spring, I heard one of my fellow quant-y friends quip “the plural of anecdote is not data.” This sounded pretty clever, so I laughed along smugly, and made sure to save this bullet in my rhetorical bandoleer for future use.
But after a decade working on improving communication strategy across many contentious topics, I’ve seen enough data to convince me of the value of anecdotes.
Here’s a frustrating fact about making communication persuasive and memorable: sometimes better evidence is worse and worse evidence is better.
What I mean is that large, representative datasets are usually the best way to figure out whether something is true. If you want to know whether a product claim holds up, whether a policy works, or whether a social trend is real, you are going to need to gather evidence in the aggregate. You want a representative sample, consistent valid measures, and a clean design so that the results generalize beyond the cases you were able to observe. This is the whole point of doing research instead of guessing.
But once you’ve figured out what’s true and what’s important to tell people, then you have to find a way to tell them. And that’s where a strange inversion happens. The aggregate data that made you confident isn’t all that convincing. But a single vivid anecdote, which you as a decent scientist would never dare generalize from, is what actually resonates.
So, you end up with a paradox. Strong evidence is weak evidence and weak evidence is strong evidence. Anyone whose job is communication has to 1) make peace with this, then 2) become fluent in both languages, and then 3) consider yourself a translator between the two.
Here’s a few examples. First, notice how this headline reporting strong scientific evidence is handicapped by jargon, abstractness, and physical distance. Is 400 PPM a lot? Is CO2 bad? Does the Antarctic atmosphere really matter to me?
Even selecting the wrong imagery can create distance that makes the problem feel abstract, impersonal, and intractable. Check out the cover of TIME in red that says “Special Report: Global Warming…. Be Worried. Be Very Worried.”
And then the image is a polar bear. Perhaps it elicits some compassion. But I’m skeptical that it is helping the audience be—as the title demands—truly “worried.”
But some advocacy campaigns have figured it out. A great example of centering anecdotes is the work by a Pittsburgh environmental advocacy organization called Alleghany County Clean Air Now. Their entire comms strategy is to build vivid profiles of local citizens who are already experiencing health effects from living next to industrial polluters.
Basically, they are looking at their audience’s shoestring attention budget and deciding to spend $0.00 of it on scientific data. Instead, they go all-in on anecdotes.
Even in the profiles themselves, the visual emphasis is on making a human connection first. The focal point is large high-res photos popping out of the foreground, with direct eye contact and a name. Then it’s accompanied by some details about their personal experience (health concerns and a quote).
More is Less
This all feels kind of backwards, right? By the standards we would normally apply, a single story is terrible evidence. You can find proof of nearly anything if your evidence threshold is a sample size of 1. Just because a few people in Pittsburgh have asthma doesn’t mean there’s an industrial pollution problem! Further, a popular anecdote is probably popular precisely because it was the most striking instance available, which is another way of saying it was selected for its non-representativeness.
And yet the story wins. This is because human attention was not built to prioritize spreadsheets and trendlines and heatmaps and forecasting models. Instead, human attention is built to stay tuned into what’s been happening to us and our tribe. That instinct kept our ancestors alive. In the modern world, we’re still operating with mostly that same cognitive hardware and software.
A clear demonstration of this comes from the work of the psychologist Paul Slovic. In one experiment, he and some collaborators approached people who had just finished participating in a different study and asked them if they would donate part of their earnings to Save the Children. Unbeknownst to the participants, they had been randomly assigned to three different groups. One read about Rokia, a seven-year-old girl from Mali facing hunger, shown with her photo. Another group read statistics about how millions of children were at risk across several African countries. A third group got Rokia’s story plus the statistics too.
People donated the most, by far, when they read about Rokia alone (Group 1). As you might guess by now, Group 2 (statistics only) donated the least. But the most interesting group to me is Group 3. Even though they had seen Rokia’s story just like Group 1, the mere presence of the statistics actually reduced the average donation amount about 38% from about $2.25 to roughly $1.40. In this study, telling people the problem was bigger made them donate less.
Slovic called this “psychic numbing.” There’s a famous quote (probably falsely?) attributed to Stalin: “One death is a tragedy. One million is a statistic.” It’s a chilling line because 1) you sort of immediately know deep down that it’s accurate and 2) you know that if Stalin did say it, it was because he had special first-hand knowledge.
If you need a pick-me-up after that, there’s also a positive version, as said by Mother Theresa: “If I look at the mass I will never act. If I look at the one, I will.”
Through either frame, one child named Rokia is a person you can imagine and want to help. But a million children is just a number, and numbers don’t have faces.
What’s Going on Under the Hood?
You might think this is purely because stories are interesting while numbers are boring, and that’s that. But the effect runs deeper than that. Stories actually change how the brain handles the incoming information.
Researchers call it “narrative transportation.” When someone gets absorbed in a story—like when they’re following a character and picturing the scene—they sort of detach from the world around them. You’ve probably felt this when the lights come on in the theater after a movie and you realize you’ve been completely unaware of your surroundings for the past 120 minutes. Or maybe while you listen to a good friend tell a good story, you might not realize that the tea kettle is whistling.
When a person is immersed in this “transported” state, they tend to process information differently. One important difference is that people are less likely to counter-argue (i.e., look for flaws in reasoning, or think of counterexamples) because they’re lost in the story and are busy imagining the events. On top of that, a well-told story with a relatable protagonist activates emotions like fear, hope, and self-confidence—all strong precursors of taking action.
It can also become a vicarious experience where the listener starts to feel personally invested in the stakes and the outcomes. This kind of vicarious experience gets weighted pretty heavily by our ancient brain. More so, at least, than a scatterplot. These pathways—reducing opposition, sparking emotions, and increasing personal salience—are how storytelling gets its persuasive boost.
Recently, there’s been a wave of solid data supporting the value of storytelling for communicating about abstract, complex issues. For example, in a pair of studies, my colleagues and I tested a 90-second radio story: a North Carolina sportsman in his 70s describing, in plain words, how a warming climate had changed the places where he had been hunting and fishing his whole life. We played it for political conservatives and moderates, who tend to be reluctant to budge on this topic. But the experiment showed the radio story was pretty effective in shifting their beliefs and their risk perceptions. These effects held up across data collected in two separate samples. Our analyses also showed some evidence that people who changed their beliefs most were those who felt more emotional connection with the story in the form of worry and compassion.
My takeaway from this project was that the storytelling avoided the debate entirely and focused on just painting a vivid picture of a relatable character who the audience would care about.
Becoming Bilingual: Some Practical Advice
Earlier I said strong evidence is weak and weak evidence is strong. But I have to confess: I was doing something a little shaky with the semantics there. Part of the reason it sounds like a fascinating paradox is because I’m sort of conflating two different things with the words “strong” and “weak.” Clearing this up will help move us toward some practical advice.
On one hand, the strength/weakness of evidence can mean how good it is at telling us whether a claim is true (e.g., “there is strong evidence that smoking causes cancer”). This is about questions like: What is really happening overall? What’s the measured impact? Does this finding replicate and generalize? Scientific evidence is strong for that, while anecdotal evidence tends to be weaker.
On the other hand, the strength/weakness of evidence can mean how much of an impact it has on people when they hear it (e.g., “the witness’ strong testimony persuaded the jury”). Anecdotal evidence is often strong in this sense. It often has a greater impact on audiences’ beliefs and behaviors than scientific evidence does. So the personal story told by that North Carolina outdoorsman was bad data (for proving scientifically that climate change is happening) but is a great anecdote (for convincing an audience that climate change is already here).
Really, these are just two different languages. A good communicator is fluent in both languages and ideally can code-switch or translate between them.
So how do you know which to use? After a decade of fumbling around and making mistakes in both directions, here’s a rule of thumb I’ve landed on. You should try to figure out which of two modes your audience is more likely to be operating in: 1) Trying to honestly evaluate, in good faith, “Is this true?” or 2) Disengaged, distracted, skeptical, or unsure of “Why should I care?” There are, of course, unlimited exceptions and caveats to this (as well as audiences that fit both or neither), but despite all that it always surprises me how helpful this simple framework is as a default starting point.
When they are honestly evaluating “is this true?”
Here you should speak from the data. Many audiences are listening to you with the goal of verifying something or deciding something, and often they might be held accountable for their choices later.
Maybe it’s a senior executive, a regulator, campaign strategist, a school board, or a journal reviewer. They likely aren’t where the buck stops and their decisions are probably going to be have consequences later, so their radar is tuned to focus on information that can be a solid evidence base for their decisions. Saying “My neighbor’s barber told him that…” isn’t really what they need. Give them the results, the methods, the applications, and the caveats.
Data is also a good mode to operate in when the audience is professionally trained to distrust anecdotes. For these folks, an anecdote offered in place of evidence is sort of revealing that you don’t have evidence.
There’s less direct evidence on this, but another situation where it might help to focus on data is when the claim is contested and the other side has better anecdotes than you. Sometimes your opposition is wielding some scary stories about vaccines causing autism, or fascinating miracles of how someone survived a car accident by not wearing a seatbelt. If your audience already has those stories already installed in culture’s shared mental models, your competing story is just starting a “he-said-she-said” story fight. These story fights are decided by vividness, rather than by accuracy. A good move here can be to re-center the debate back to the data—even admitting that odd outliers do happen but the trend remains.
And while you’re at it, use the data when you are still deciding yourself! Before you go persuade anyone else, you are your own first audience, and you should be data-hungry. If you start with the story, you just might fall in love with it and then go looking for evidence to justify it.
2. When they are disengaged and DO NOT CARE
Here is where you can leverage the power of the anecdote.
A common case is when the audience already agrees on an intellectual level, but just isn’t motivated to act. In Slovic’s study, Save the Children didn’t need to convince people that child hunger exists and is bad. The real problem was that there was a disconnect between their knowledge (“hunger exists and is bad”) and their motivation to act (“I care enough to do something about it at my own personal expense”). If your audience’s beliefs are already generally where you want them, adding more data about the existence of the issue is wasting valuable bits of attention.
Another case is when the audience feels far away from the issue. Far away in distance, far away in time, or unlikely for anyone dear to them to be affected. The Antarctic atmosphere is far. A polar bear is far. An endearing outdoorsman in North Carolina is much closer, at least to conservatives and moderates who also live in the Southeastern U.S. (those were our participants in that study). Stories can shrink that distance and give the issue a relatable face.
A third case is when the audience is primed to argue with you. If you’re trying to make an impact on an audience that will counterargue or dismiss every data point you hand them, then offering a story will be more difficult to parry. If someone tells you their lived experience, it’s hard to dismiss that without looking like a jerk (and most people would rather not look like a jerk).
And finally, reach for an anecdote when you only have 10 seconds. This could be a headline, an ad, a social media post, the first slide of your deck, an elevator pitch, a sales opening, etc. In those slots you should try to present a face first (literally or figuratively). It’s rare that people will stop scrolling because they saw a point estimate with a confidence interval.
Imagine you’ve been trying for a long time to get your father to go see a doctor about a thing he keeps ignoring. He’s already heard the statistics and probably even mostly believes them. He does not need more of that. Really what he needs is to hear a personal experience from the guy in his bowling league who either a) put it off too long and now has bad consequences or b) took action and found it way more painless and beneficial than he expected.
In Closing
Looking back, I’ve become skeptical of that zinger that I laughed along with in grad school. The anecdotes, stories, and observations we soak up in the world around us are very valuable! They are often the Newton’s apple that spark our best research ideas. And then they turn into powerful tools for inspiring others to care.
And on top of that, in many ways, the plural of anecdote actually is data!
If you’re gathering a big pile of anecdotes (plural), are those not the seeds from which strong data can grow?
Cite this
Gustafson, A. (2026). Numbers Don’t Have Faces: The Paradoxical Power of Storytelling. XandY. New Haven, CT. Retrieved from: https://www.xandyanalytics.com/numbers-dont-have-faces