This study combined an instrumental variable approach with a quasi-exp
erimental design to isolate a naturally occurring, behavioral media ef
fect. The data included measures of real-world earthquake severity, me
dia coverage and subsequent relief donations for 22 foreign earthquake
s. Levels of relief were compared across three sources: private U.S. c
itizens, the U.S. government, and international agencies. An instrumen
t based on the number of earthquake-related deaths was developed to pr
ovide an exogenous measure of media coverage. Findings suggest that mo
re U.S. network news coverage increased aggregate donations by private
U.S. citizens. By contrast, the amount of U.S. network news coverage
was not associated with U.S. government assistance or international ai
d, although these forms of aid were highly correlated with other (nond
eath) measures of severity.