Presidential approval remains one of the most discussed aspects of American politics among citizens, pundits, and political scientists. For all the research that has been conducted in this area, however, most conclusions are generated from aggregate data and limited to such phenomena as macroeconomic forces or political shocks. The research presented below provides an experimental take on the presidential approval puzzle. We concentrate on the sources of information that may influence approval of the president's handling of specific issues. We do this by employing an experiment that manipulates information individuals receive about President George W. Bush's handling of the situation in Iraq and Social Security. We attempt to shed some light on how information, varying in tone and substance, about issues, varying in saliency, affects perceptions of the president. We find that, under certain circumstances, information can move people's opinion of the president, controlling for such factors as partisanship and affect. Furthermore, we find that the results of public opinion surveys are as influential as newspaper opinion content among the types of information that affect people's perceptions of the president.
The article proceeds in the following manner. In the next section, we discuss the literature on presidential approval, information, and issue saliency. Next, we explain the experiment underlying this research and discuss the context in which it was conducted. Then, we present the results of the statistical models. Finally, we conclude with a discussion of the research's implications.
Presidential Approval, Information, and Issue Saliency
The empirical study of the determinants of presidential approval has occupied scholars for decades. Since Neustadt (1960) first argued that public approval affects a president's prestige, which in turn influences his ability to be successful, research has focused on the factors associated with the president's fluctuating standing with the electorate.
The focus of much of this research has centered on objective economic indicators and major events, such as the immediate aftermaths of military interventions or domestic crises, as the primary predictors of public approval of the president. Past research has variously found that national unemployment levels (Hibbs 1982; MacKuen 1983), the inflation rate (Kenski 1977; Kernell 1978; Hibbs 1982; MacKuen 1983; Lanoue 1988), and various measures of personal income levels (Tufte 1975, 1978; Hibbs 1982, 1987; Lanoue 1987, 1988; Erikson 1989) all influence significantly the percentage of the public that approves of or votes for the president. In addition to its popularity as a dependent variable, scholars have used approval as an independent variable with substantial explanatory power predicting the president's electoral success, his party's, particularly at midterm (Tufte 1975; Kernell 1977, 1978; Abramowitz 1985; Abramowitz, Cover, and Norpoth 1986), his legislative program (Rivers and Rose 1985; Bond, Fleisher, and Northrup 1988; Canes-Wrone and de Marchi 2002), and his status among other Washington elites (Kernell 1997).
Thus, a great deal of the research seeking to both explain presidential approval and use it as an explanatory factor has been conducted at the aggregate level. More recently, scholars have begun investigating the individual-level consequences of these aggregate-level relationships. The research presented below moves in this direction as well.
The larger literature regarding how information influences attitudes has flourished in recent years. Numerous studies demonstrate opinion change based on such factors as attention paid to or knowledge of politics and the tone or direction of arguments made for or against particular issue positions. Moreover, scholars have paid more attention to the type of information received, as well as the predispositions of the individual receiving such information. …