power or possession of a border province, to a more ambitious goal, like the complete conquest
of the enemy state or the destruction of its regime (Labs 1997). Conflicts in which war aims
begin at a high level or increase during the course of the conflict tend to be longer and costlier
than other wars because they inspire greater resistance in the potential victim, opposition that in
turn forces the attacker to employ greater levels of violence to obtain its increased aims (classic
cases occurred in the Franco-Prussian War and World War I). Moreover, escalations of war
aims may spark the intervention of a third party into the war—as in Korea in 1950—which can
increase the length and destructiveness of the conflict. Finally, heightened objectives obviously
make negotiated settlements harder to achieve because leaders cannot easily climb down from
such escalations once publicly committed to them. I use a dummy variable coded 1 if a
belligerent sought total conquest, unconditional surrender, or regime change from its opponent,
or if its war aims expanded during the course of the conflict.
Two other factors may increase the likelihood of civilian targeting indirectly by
increasing the probability that wars will become costly or protracted. First, the closer the
military balance between attacker and defender, the harder it is to achieve a quick and decisive
victory, and the more likely the war will result in a stalemate. Conversely, when the military
balance is highly skewed in favor of one side or the other, the war is likely to be over quickly:
the victor in such a conflict has little need for strategies that target civilians, and the loser usually
does not have the ability to adopt such a strategy (either because he is too weak, or the war is
being fought on his soil).
12
My measure of the military balance is constructed by computing each participant’s share of the total capabilities
of all the belligerents combined (population, urban population, energy consumption, iron and steel production,
military personnel, and military expenditure), producing a score between zero and one. Because countries at the
high and the low end of the scale should be unlikely to kill enemy civilians because wars between them should be
quick and decisive, using the raw figures would be misleading. Therefore, I subtracted 0.5 from every score—which
resulted in a scale between -0.5 and 0.5—and then took the absolute value of that number. The resulting variable
ranges from zero to 0.5; states close to zero are more evenly matched with their opponents, whereas those nearer to
0.5 are either heavily advantaged or disadvantaged. This variable should be negatively related to civilian
victimization.
19
Convention