...according to the folk, anyway. In a fascinating paper, Nordgren and McDonnell found that subjects judged that wrongful actions (defrauding people and knowingly selling poisoned food) were judged as less severe when they harmed 30 people than when they harmed 3. Subjects also judged that the perpetrators should get shorter sentences in the 30 people cases than in the 3. Nordgren and McDonnell suggest that the effect is due to our feeling greater sympathy for identifiable individuals than for abstract individuals. Sure enough, they found that subjects asked to describe victims in the small number cases produced a longer list of traits. They also found that increasing identification in the large number cases, by giving subjects a photo of one of the victims, partially eliminated the effect, inasmuch as subjects then judged the actions as equally severe in both conditions (this may not count as elimination, since one would think that we ought not to judge the actions as equally severe; we ought to judge actions harming more people as worse than actions harming fewer).
Nordgren and McDonnell also looked at damages awarded in actual cases like those they gave their subjects, and found that the effect seems to be at work in the real world: juries punish defendants less harshly when their actions harm more people than when they harm fewer.
H/T Ben Goldacre.