Abstract: | Objectives. Our purpose was to develop and test several hypotheses concerning the impact of poll‐question wording on aggregate public support for war. We drew on general insights from framing theory and specific insights from various theories of public support for war. Methods. Our database consisted of two collections of aggregate poll results drawn from the prewar and major combat phases of the Gulf War (1990–1991) and the Second War with Iraq (2002–2003). For each data set, we used multivariate OLS regression to gauge the impact of specific question‐wording variations on the percentage of respondents expressing support for war, controlling for systematic time and pollster effects. Results. Most of the hypothesized wording effects were significant in the expected direction. Mentioning WMDs, terrorism, Saddam, hostages, and international support for war boosted aggregate war support in one or both wars; mentioning the president, oil or gasoline, international opposition to war, and U.S. or Iraqi casualties depressed support. Conclusions. Various theories emphasizing different “rational” aspects of public attitudes toward war are supported. However, the significance of mentioning Saddam by name in the Second War with Iraq, and mentioning the president in both wars, would seem to imply framing effects based more on emotion and/or symbolism. |