Political Elites and Public Support for War

Download data and study materials from OSF

Principal investigators:

William Howell

University of Chicago

Email: whowell@uchicago.edu

Homepage: http://harrisschool.uchicago.edu/directory/faculty/william_howell

Douglas Kriner

Boston University

Email: dkriner@bu.edu

Homepage: http://people.bu.edu/dkriner/index.html

Sample size: 1617

Field period: 2/23/2006-3/1/2006


A growing body of research underscores the importance of political elites in shaping the public’s views about war. Building on this literature, we examine how the political identity of elites interacts with the content of appeals in order to influence (or not) the views of different constituents of the American public. Linking over 5,000 congressional speeches about the Iraq War with trends in public support for the War, and examining three experiments imbedded within a national public opinion survey, we find considerable evidence that elite appeals are most influential when they are deemed either trusted (because they come from a like-minded source) or costly (because they self-evidently conflict with a source’s private interests). We also find that public appeals on war can occasionally backfire, inducing members of the opposing party to respond in ways exactly opposite an elite’s intentions.


Hypothesis 1: The influence of elite cues on public support for war critically depends on three factors: the content of the signal; the identity of the sender; and the predispositions of various groups of citizens to believe or reject the sender’s claims.
Hypothesis 2: Two types of elite appeals should be particularly influential in shaping popular support for war: “trusted” signals, which are conveyed by elites with whom citizens share partisan or ideological ties; and “costly” signals, which self-evidently conflict with the sender’s ideological priors, partisan commitments, or vested interests.

Experimental Manipulations

With a nationally representative sample of 1,617 adults, we randomly assign individuals to one of nine (one baseline, eight treatment) conditions that present various vignettes about the foreign policy positions of the president and either congressional Republicans or Democrats, the United Nations, or international aid organizations. The president’s position is always stated first, while those of other political elites follow. In the first four treatment conditions, respondents are told that the elite group supports the president’s position; in the last four treatment conditions, respondents are informed that the elite group opposes the president’s position. Respondents then are asked whether they support the president’s preferred policy.


The three key dependent variables identify respondents’ support for: 1) setting a timetable to withdraw American troops from Iraq; 2) using military force to combat suspected terrorism in Eritrea; and 3) using military force to curb alleged human rights abuses in Liberia.

Summary of Results

Many, though not all, of the observed differences across baseline and treatment conditions can be attributed to either trusted or costly signals. Here we summarize those that relate to congressional signals. Consistent with expectations about trusted signals, Republican respondents take cues from Republican members of Congress. Though the differences are not always statistically significant, Republican respondents appear more likely to support the president when their party in Congress supports him, and less likely to do so when their party in Congress opposes him. In line with expectations about costly signals, respondents from both parties appear more likely to support the president when Democratic members of Congress support a presidential use of force; and respondents from both parties express lower levels of support when Republican members of Congress oppose the president. We also find evidence of backlash effects: when congressional Democrats oppose the president, Republican respondents express higher levels of support for the president.


This paper focuses on two types of elite communications—trusted and costly signals—that predictably influence public opinion on war. Knowing little about ground operations in Iraq and nothing about hypothetical conditions in Liberia and Eritrea, we have shown that citizens sort themselves in systematic ways in response to the kinds of arguments that different elites, manifesting unambiguous political characteristics, offer to the American public. When Republicans support Bush’s foreign policy and Democrats oppose it, citizens within each of their respective parties update their views accordingly. And when Republicans oppose the president and Democrats support him, all citizens take notice. In political life, arguments about war, as with arguments about all policies, do not circulate singly. Rather, elites advance the arguments. And knowing the political identity of these elites, citizens have a stronger basis upon which to evaluate their claims.


William Howell and Douglas Kriner. “Political Elites and Public Support for War.” Working Paper.
William Howell and Douglas Kriner. 2007. “Bending so as Not to Break: What the Bush Presidency Reveals about Unilateral Action.” In George Edwards and Desmond King (eds), The Polarized Presidency of George W. Bush. Oxford: Oxford University Press.