There is a sentence about death being a tragedy once and a statistic a million times that everyone attributes to Stalin, and everyone is wrong. The idea behind it is older than the misquote and it has a name: the identifiable victim effect. The short version is that we help a specific person far more readily than we help a vaguely defined crowd with exactly the same need, and that this is not a small inconvenience in human generosity but a structural feature of how the mind decides who deserves what. It is the reason charity appeals almost never show you a crowd. It is also, as it turns out, one of the most confidently repeated findings in psychology that recent large-scale replication has had serious trouble standing up.
🧠 The Girl Who Outraised a Famine
In a 2007 paper in Organizational Behavior and Human Decision Processes, Deborah Small, George Loewenstein and Paul Slovic gave people fundraising materials in three versions. One described Rokia, a seven-year-old girl in Mali, with her name, her photograph, and her particular situation. One described the same food crisis as a statistical event — millions of people at risk, nobody in particular. The third gave Rokia’s story and then appended the statistics.
The version with the one named girl raised the most. The version with only numbers raised the least. That much is intuitive enough for anyone who has ever written a fundraising letter.
The third condition is the one that stung. Adding the statistics to Rokia’s story made people give less than they gave for Rokia alone. The information was true, relevant, and about the very crisis they had just been moved by — and it worked like a bucket of cold water.
The intellectual ancestor of the finding is the economist Thomas Schelling, writing in 1968: harm to a particular person summons “anxiety and sentiment, guilt and awe, responsibility and religion,” but most of that “awesomeness disappears when we deal with statistical death.” The line usually credited to Mother Teresa makes the same point from the other side: if I look at the mass I will never act; if I look at the one, I will.
⚙️ How It Works
Helpers do not allocate. They soothe. The affect heuristic says we decide by consulting a feeling and then reasoning backward to a justification. So the question a donation appeal actually has to answer is not “is this an efficient use of my money” but “does this make me feel something I want to stop feeling.”
A single person generates an image, and the image is what carries the feeling. A name, an age, a face — Rokia is no longer a category, she is a specific child you can picture, and picturing her produces sympathy and distress. Kogut and Ritov found in 2005 that once a critically ill child’s identity was supplied, people reported more distress at her plight and donated more, which suggests the giving is acting as emotional relief rather than resource allocation.
Statistics generate no image, so no feeling fires, and the mind switches into accounting mode. Three million people does not register as three million of anything; it registers as a sentence. Now the operative questions become practical ones — will my twenty dollars matter, what is my relation to this — and a small donation looks like a rounding error against a number that size, so the hand stays in the pocket.
The operative word is “single,” not “vivid.” This is the part that gets misremembered. In the same 2005 research, participants facing a group of eight critically ill children were given every child’s name and photograph, and the donations did not move. Naming the whole group changed nothing. Paul Slovic calls this the singularity effect, and his team found people would give less to help two starving children than to help one. Compassion does not scale with the number of sufferers; past a very small count it collapses, a phenomenon usually filed under psychic numbing.
So the one beats the many — and then the numbers arrive and break the spell. Reading about Rokia, you are in feeling mode. Adding the famine statistics re-tasks you with evaluating an enormous abstraction, and the evaluation you produce is that your contribution is negligible. That is the third condition of the 2007 study: not a lack of information, but too much of the wrong kind at the wrong moment.
One boundary condition matters. The effect holds most reliably when the victim is plainly not responsible for the situation. When a victim can be read as having caused their own trouble, identifying them can increase blame and reduce help — the same machinery that makes sheer innocence moving makes blame feel like justice.
🤔 The Part Where It Gets Awkward
The identifiable victim effect has been received wisdom for two decades. It shapes charity campaigns, humanitarian messaging, and a great deal of advice about storytelling. And when researchers went back with large samples and preregistered analyses, the thing mostly was not there.
Lee and Feeley’s 2016 meta-analysis of 41 studies found an overall effect that was significant but tiny: r = .05. It showed up mainly in narrow conditions — a single identified or single unidentified victim, a photographed child in poverty, monetary requests, an obviously blameless sufferer. Then in 2024, Maier, Wong and Feldman published a preregistered replication of two of the original Small, Loewenstein and Slovic studies and found no identifiable victim effect at all, and therefore nothing to debias. Their reframing is sharper than a null result: perhaps people were never giving more to the one. Perhaps they are simply insensitive to the number — the crowd from one to a million moves the needle almost not at all, which makes the single case look generous by comparison. A separate 2024 replication of the Kogut and Ritov experiment, with over two thousand participants, also failed to reproduce the single-identified-victim effect and instead found something older and less flattering: an ingroup bias, people helping their own.
The 2026 preprint is where the case gets closed for now. A meta-analysis of 266 effect sizes covering 344,962 people found an overall effect that did not reach significance, r = .04, p = .057. Only one slice survived — child victims — at r = .07, which is very small.
Two things can be true at once here, and both should be held. Real cases have moved real money: when 18-month-old Jessica McClure fell down a well in Texas in 1987, the public sent hundreds of thousands of dollars over the 58 hours of her rescue; when the photograph of Alan Kurdi, a three-year-old Syrian refugee, was published in 2015, one migrant rescue charity saw donations rise fifteen-fold in twenty-four hours. And none of that is evidence of a general law — the Wikipedia article on the effect is careful to label those cases illustrations, not proof. What the lab work suggests is that scope insensitivity is the sturdier finding, and that the tidy story of “one named victim wins” was a story we liked enough to keep telling after the evidence got thinner.
The honest summary for anyone who writes for a living: telling one person’s story is still the strongest tool available, and the mechanism may be less “the one is more than the many” and more “the many have never registered at all.” Which is a stranger claim, and a better one.
🔗 Why It Should Matter to You
If you build anything where a person has to be persuaded — a fundraiser, a product page, a case study — the practical takeaway survives the replication mess: show one named individual with a face and a situation, because there is no evidence that the alternative works. What changes is the confidence label. This is a technique with real-world demonstrations and a shaky experimental foundation, so it belongs in the “works, and here is why I am not sure” column rather than the “proven law of psychology” column.
Writers get the same lesson from the opposite direction. Anyone who has tried to make readers care about a war, a famine, or a generation of people will have noticed that the moment the prose zooms out to “thousands suffered,” the temperature drops to room level. The fix is not more adjectives about the thousands. The fix is one person in one room, and the crowd stays off the page. A single protagonist holding an entire novel is not a convention; it is the only setting on this machine that reliably produces feeling.
🎲 Bonus
The effect has a mirror image, and it runs on the same fuel. Researchers have found an identifiable perpetrator effect: people are willing to pay real costs to punish a specific named wrongdoer, even when the punishment is expensive to them personally, while anger at an abstract “system” or “company” dissipates into nothing. One named victim pulls help out of us. One named villain pulls punishment out. Everything in between is arithmetic.