In the 1960s, the economist Thomas Schelling performed research demonstrating that people are more likely to be moved by single victims than by statistics. In 2005, the psychologists Deborah A. Small, George Loewenstein, and Paul Slovic found the limits of human compassion to be even more irrational and constrained. In their study, students at a university in Pennsylvania were paid five dollars to complete questionnaires on technology. Enclosed with the questionnaire was a seemingly unrelated letter soliciting donations to a hunger relief organization in Africa.
The study's first conclusion was what the researchers had expected: people are more compassionate when they are told about a specific victim. When respondents were asked to donate money to help feed a seven-year-old African girl named Rokia, they contributed more than twice what they did when just confronted with general statistics on hunger.
But then things got surprising. When Rokia was presented with the statistics, the donations fell by nearly half. Worse still, when the authors asked one set of subjects to perform mathematical calculations and the other set of subjects to describe their feelings when they heard the word "baby," the subjects who'd done math gave only about half as much to Rokia as the ones who'd thought about babies. Apparently, just thinking analytically makes us stingier. The authors of the study concluded that "calculative thought lessens the appeal of an identifiable victim."
I find the opposite effect, but I am eccentric.