Identifiable Victim Effect: One Named Person Beats a Statistic

A named person with a face produces a feeling, and the feeling is what most giving decisions run on. Statistics about millions produce an understanding, which is a different thing and a weaker motivator. Small, Loewenstein and Slovic put it as a difference in processing: reactions to an identified victim are affective and immediate, while reactions to statistical victims have to be assembled deliberately, and the second route generates far less sympathy per unit of suffering described.

Karen Jenni and George Loewenstein tested four candidate explanations in 1997: that identified victims are more vivid, that their harm is certain rather than probable, that the help arrives after the fact rather than before it, and that the identified person makes up the whole of the group being helped. The last one got the most support. If one named child is the reference group, a donation covers 100 percent of the people at risk; if three million children are the reference group, the same donation covers almost none of them, and that proportion is what people seem to respond to.

The uncomfortable follow-up is that correcting the gap does not raise giving. In the 2007 studies, teaching people about the discrepancy, or presenting the statistics alongside the named child, pulled donations to the named child down toward the statistical level instead of pulling the statistical donations up. Deliberation reduced the sympathy that was there without creating any where there was none.

How large the effect is in the first place is now contested, and this page would be overselling if it skipped that. Lee and Feeley pooled 41 studies in 2016 and found an overall effect of r = .05, which is small. Maier, Wong and Feldman reanalyzed that meta-analysis in 2023, found signs of publication bias, and ran a preregistered replication of two of the 2007 studies using hypothetical donations: they found no support for the effect, and suggested the pattern is better described as insensitivity to the number of victims than as a bonus for the named one. Their replication used imaginary money, as the demo above does, while the original handed people real dollar bills, and they called for high-powered replications with real donations. Expect the split between the two versions on this page to be small or absent for that reason.

Deborah Small, George Loewenstein and Paul Slovic published "Sympathy and callousness: The impact of deliberative thought on donations to identifiable and statistical victims" in Organizational Behavior and Human Decision Processes in 2007. The donations were real. An experimenter who did not know which version a participant received approached people sitting alone on a university campus in Pennsylvania and offered five dollars to complete a short survey about technology products. Afterward each participant was handed five one-dollar bills, a receipt, a blank envelope and a charity request letter, then left alone to decide privately.

Study 1 had 121 participants in a two-by-two design: the letter described either an identifiable victim, a girl pictured and described from the Save the Children website, or statistical victims, four lines of figures about food shortages in Malawi, Zambia, Angola and Ethiopia. Half of the participants also read a short lesson about research showing that people respond more strongly to specific people than to statistics. Those who faced the identifiable victim gave more, an average of 2.12 dollars against 1.21 dollars. The lesson cut giving rather than evening it up: the identifiable condition without the lesson averaged 2.83 dollars, while the other three cells together averaged 1.26 dollars.

Study 3, with 159 participants, is where the widely quoted figures come from. It had three conditions: the named child alone, the statistics alone, and the named child presented together with the statistics. Donations averaged 2.38 dollars for the named child, 1.14 dollars for the statistics, and 1.43 dollars when both appeared together. Adding the statistics to the named child's letter removed most of the difference, which is the result the authors called an implicit intervention.

The paper reports further studies in which priming people to calculate reduced giving to the identified victim. Two limits are worth keeping in view: all of this was run on and around one American campus, and the sums involved were five dollars, not the scale of decision a donor or an agency makes.

The pattern appears wherever one person can stand in for a number. The rows below are illustrations, not findings from the studies cited on this page.

The 2007 studies tested two corrections and both backfired, lowering giving to the named victim without raising giving to the statistical ones. So the goal is not to feel less for the named person; it is to stop the choice of recipient from being made by whichever story arrived most recently.

• Decide the allocation before the story. Setting a budget and a split in advance, then letting appeals compete inside it, keeps sympathy as the motivation and arithmetic as the allocator.
• Automate the unglamorous part. A standing monthly donation to coordinated relief does not need a face to compete with, because the decision was made once.
• Ask for cost per outcome. Independent evaluators publish it for many causes, and it is the number the photograph is standing in for.
• Notice the reference group. A donation covers all of one named child's need and a fraction of three million children's need; that proportion, which Jenni and Loewenstein found mattered most, is not a measure of how much good is done.
• Be wary of debiasing by lecture. Teaching people about this effect reduced total giving in Study 1, so if your aim is more help rather than tidier reasoning, change the rule rather than the feeling.
• If you run appeals, say both. A named person plus a number is a smaller appeal than the person alone, which the 2007 study showed directly, and that is a real trade-off to make deliberately rather than by accident.

Several biases describe the gap between a case and a rate, and they enter the judgment at different points.

• It is not the availability heuristic, which is judging how common something is by how easily examples come to mind. The identifiable victim effect is about how much a single case motivates action, not about estimates of frequency.
• It is not the anecdotal fallacy, in which one story is treated as evidence against data. Here the number is not disputed; it just moves nobody.
• It is not neglect of probability, the tendency to ignore how likely an outcome is. Both of the 2007 letters described certain, ongoing harm; what differed was a named child against people described statistically.
• It is not the empathy gap, the difficulty of predicting how you or others will feel in a different state. The identifiable victim effect is about which target of help feels worth acting on now.