War expansion and war outcome

Most wars do not expand beyond the initial two participants. Why is this so? We argue that wars remain small because initiators select as targets states that they believe will not receive third-party help and that they can defeat without such help. Drawing on the idea of selection effect, a model of this choice is presented and a hypothesis is derived in which initiators (1) will win most often in wars of one against one and (2) will win least often when the target receives any help. This hypothesis is tested against war outcomes for initiators and targets in the period 1816-1975 using probit regression. The expectation is supported. The authors conclude that initiators act as predators and are likely to attack target states they know they can defeat if these targets are not joined by coalition partners. This selection pattern tends to make small wars likely.