This paper develops and applies a new approach for analyzing the spatial aspects of individual adoption of a technology that produces a mixed public-private good. The technology is an animal insecticide treatment called a “pouron” that individual households buy and apply to their animals. Private benefits accrue to households whose animals are treated, while the public benefits accrue to all those who own animals within an area of effective suppression. A model of household demand for pourons is presented. As a private good, household demand for the variable input depends upon output price, input cost, and household characteristics. Input costs for pouron treatments include both the market price of the pourons and the transaction costs that the household must incur to obtain the treatments. Demand also depends upon the way that each household expects its neighbors to respond to one’s own behavior. Free-riding is expected in communities with no tradition or formal organization to support collective action. Greater cooperation is expected in communities that have organizations that reward cooperative behavior and punish deviant behavior. Data for estimation of the model were collected for all of the 5,000 households that reside within the study area of 350 square kilometers in southwest Ethiopia. Geographic reference data were collected for every household using portable Geographic Positioning System units. GIS software was used to generate spatial variables. Variables for distance from the household to the nearest treatment center and number of cattle-owning neighbors within a 1-kilometer radius of the household were created. The density of cattle-owning neighbors was used as a measure of the potential benefits from cooperation; this variable was expected to have a positive effect on household pouron demand in communities able to support effective collective action and a negative effect in communities not able to support effective collective action. A set of community binary variables was interacted with the density variable to capture differences between communities. The results confirm the importance of the household-level variables. The results also indicate large differences in ability to cooperate between local administrative units. Everything else equal, the areas least able to cooperate were located farthest from the treatment center, were ethnically heterogenous, and had a different ethnic composition than areas around the treatment centers.