Fifty Years of Violent War Deaths From Vietnam to Bosnia (Obermeyer, Murray and Gakidou, 2008)

A peer reviewed study in the BMJ using sibling survival histories from the World Health Survey Programme's household surveys across thirteen countries to derive war death estimates by a statistical method the authors argue is less prone to undercounting than earlier approaches. For Vietnam it reports a substantially higher toll than earlier estimates, 3,812,000 dead between 1955 and 2002, with 3,091,000 of that total falling in the 1955 to 1975 window. Its strength is a named, reproducible cross national survey method. Its limit, which the authors themselves discuss, is that sibling survival methods are sensitive to survivor and recall bias, and the finding was contested by other demographers on those grounds.