Measuring the Mortality Consequences of Armed Conflict in Amritsar, India: A New Approach to the Indirect Sampling of Conflict-Related Mortality

Romesh Silva, University of California, Berkeley
Jeff Klingner, Human Rights Data Analysis Group, Benetech

The measurement of the mortality consequences of armed conflict poses several challenges. Perhaps most challenging is that traditional data systems and demographic estimation methods, which demographers normally rely on, break down during conflict, and thus are rarely viable platforms to analyze mortality consequences. Recently, the literature on network-based and adaptive sampling designs has developed substantially. This paper presents research on the mortality consequences of counterinsurgency violence in Amritsar, India between 1984 and 1996. We present a new network-based sampling design that makes use of social networks in rural India. We show how this sampling design yields a lower bound estimate of direct conflict mortality. We present sensitivity analyses to demonstrate the need for careful, broad-based selection of primary sampling points in the network sampling design. The paper concludes by outlining future research directions that seek to triangulate existing incomplete death registration systems and our network-based sample via multiple systems estimation.