| Abstract: | I seek to determine the degree and nature of bias in the military spending estimates of the two main sources of such data—the U.S. Arms Control and Disarmament Agency (ACDA) and the Stockholm International Peace Research Institute (SIPRI). I examine how ACDA and SIPRI revise their estimates of foreign military spending. I do so by evaluating ACDA and SIPRI growth rate estimates from successive statistical volumes for a full sample of countries and years as well as for seven identified regions (Africa, East Asia, Latin America, the Middle East, South Asia, NATO Europe, and the Warsaw Pact), to determine whether, and by how much, early and late estimates of each organization depart from its own or the other's final estimates. The findings reveal that systematic (or patterned) error for some regions is high and that unpatterned error for some regions is extremely high. The latter finding is important because unpatterned error can become systematic with changes in sample size. |
