Spatial and temporal patterns in civil violence: Guatemala, 1977–1986

Abstract

Civil violence is a complex and often horrific phenomenon whose characteristics have varied by era, setting, and circumstance. Its objective analysis has rarely been feasible at spatial and temporal scales great enough and resolutions fine enough to reveal patterns useful in prevention, intervention, or adjudication. An extraordinary data set simultaneously meeting scale and resolution criteria was collected during conflict in Guatemala from 1977 through 1986. Reported here is its spatial-temporal analysis; reported as well is a putatively novel method for estimating power-law exponents from aggregate data. Analysis showed that the relationship between ethnic mix and killing was smooth yet highly nonlinear, that the temporal texture of killings was rough, and that the distribution of killing-event sizes was dichotomous, with nongenocidal and genocidal conflict periods displaying Zipf and non-Zipf distributions, respectively. These results add statistical support to claims that the Guatemalan military operated under at least two directives with respect to killing and that one of these effected a genocidal campaign against an indigenous people, the Mayans. Implications for group-behavioral modeling, conflict prevention, peace-keeping intervention, human-rights monitoring, and transitional justice are noted.