Teach People Why They Give, and They Give Less

A paragraph explaining why people give more to one named child made them give less to her — and no more to the statistics. $2.83 against $1.26, real money.

An experimenter walks up to a student sitting alone in the student center at a university in Pennsylvania. Short survey about technology products, five dollars for your time. The student fills it in and receives five one-dollar bills, a receipt, a blank envelope and a charity request letter from Save the Children.

Some letters describe Rokia: a seven-year-old girl from Mali. Others describe the same food crisis in Southern Africa and Ethiopia as numbers.

Students who read about Rokia put an average of $2.83 into the envelope. Predictable. The experiment was built to test what happens next.

Half of them had first been handed a short paragraph explaining the research to them. People react more strongly to specific individuals than to statistics, it said. When Baby Jessica fell into a well in Texas in 1989, people sent over $700,000 for her rescue, while the thousands of children who will almost surely die in automobile accidents this coming year seldom evoke such strong reactions.

The paragraph was meant to make people fairer. It made them stingier.

Students who got no lesson and read about Rokia gave $2.83. Across the other three groups — statistics with the lesson, statistics without it, and Rokia with the lesson — the average was $1.26. Explaining the bias did not lift giving to the statistical victims. It pulled the identifiable one down to their level and left it there.

Daniel Small, George Loewenstein and Paul Slovic published this in Organizational Behavior and Human Decision Processes in 2007. Study one: 121 people, real money going to a real charity.

No lesson required

Study three used 159 people and taught them nothing at all. The researchers simply printed the victim statistics next to Rokia's description, in the same letter, before the ask.

Rokia alone: $2.38. The statistics alone: $1.14. Rokia with the statistics attached: $1.43.

Accurate information about the scale of the problem cut giving nearly in half. The number of hungry children was not in dispute. It was the mention of them that did the work.

A fourth study, 165 people, replaced the lesson with a mood. One group got a "Calculations Questionnaire" and was told to work carefully and deliberatively: if an object travels at five feet per minute, how many feet will it travel in 360 seconds? The other got an "Impression Questionnaire" — when you hear the word "baby", what do you feel, one word.

Then the same envelope and the same five dollars. Priming arithmetic cut what people gave Rokia. Priming feeling did nothing for the statistics, which stayed where statistics always stay. Sympathy can be switched off from the outside. It cannot be switched on the same way.

The wobble

In 2023 a team led by Maier published a preregistered replication of studies one and three in Collabra: Psychology. 1,004 participants on Mechanical Turk. They found no support for the identifiable victim effect at all — an effect size of .000, confidence interval .000 to .003 — and so nothing left to debias.

Their own explanation is the one worth sitting with. The 2007 studies put five actual one-dollar bills in a person's hand and a sealed envelope between them and the decision. The replication asked people online what they would hypothetically give. The replicators say this themselves, and end by calling for a high-powered replication with real donations.

So the careful version is narrower than the popular one. With real cash on the table, a named seven-year-old moved people, and both explaining that fact and surrounding her with numbers took the effect away. Whether any of it survives when nothing is actually at stake is unsettled.

Why this sits on a map of good deeds

Good People Map is a map of individual acts by named people in named places. The temptation is always to summarise — a counter at the top, a total for the month, the case made in aggregate because aggregate looks serious.

That summary is the move the 2007 experiments punished. Not because numbers lie, but because they are read with a different part of the mind than the one that decides to help.

None of which argues for hiding the scale of a problem. It argues for the order in which a person meets it: one neighbour first, the map afterwards.

Sources. Small, D. A., Loewenstein, G. & Slovic, P., "Sympathy and callousness: The impact of deliberative thought on donations to identifiable and statistical victims", Organizational Behavior and Human Decision Processes 102 (2007), pp. 143–153. Replication: Maier et al., "Revisiting and Rethinking the Identifiable Victim Effect", Collabra: Psychology 9(1), 2023. All figures read from the papers themselves, 31 August 2026.