Evolution of attitudes in the United States
Figure 5.1 illustrates the evolution of support for military action in the US. Before the war started, American public opinion was divided on the desirability of using air force against Serbia. The first available question was asked by Gallup in October 1998, at the height of an exchange of threats between NATO and Milosevic. At that time, the American public was strongly divided with 42 percent in favor of a war, 41 percent opposed and 17 percent with no opinion.9 Five months later, the situation was

Figure 5.1 Support for air strikes in war against Serbia over Kosovo, US, (1999) (in % support) less tense as negotiators were on their way at Rambouillet, but the (divided) attitudes were still remarkably similar to the earlier poll, with 43 percent in favor of the US being part of a military action against the Serbian forces in Kosovo, and 45 percent opposed to it while 12 percent did not answer.10 A similar question asked only one day before the attack showed 47 percent in favor of an air attack, 47 percent opposed and six percent uncertain.11
The launching of air strikes on March 24 had a positive effect on support for the NATO air operation, but it did not result in a clear majority in favor. Contrary to the Gulf War of 1991 (Mueller, 1994: 70-73), we do not find evidence in the case of Kosovo of a major 'rally around the flag' effect (Brody and Shapiro, 1989) among the American public and the same happened in Europe, for those countries for which data are available. This was probably a reflection of the remaining divisions existing among the (American and European) political elites on the appropriateness of the initiative to initiate the airstrikes and of the reluctance, in the US case, of the Clinton administration to be embroiled in the Serbian situation. No matter how the question was formulated, support for the war hovered around only 50 percent in the first days of the war, and it did not increase much after that.12
Support for the US participation in air strikes against Serbia started around 50 percent at the beginning of the war, then moved up slightly, to sixty percent by mid-April, after which it began to erode steadily and slowly to return to the level reached at the beginning of the war.13 The level of support obtained was different whether a reference to 'NATO allies' was added in the question or not.