Mapping the Boundaries of Elite Cues: How Elites Shape Mass Opinion across International Issues

Abstract

Despite intense interest in the nature and malleability of public opinion about foreign policy, there remains debate over when and how elite messages shape mass opinion on international issues, especially whether the informational or partisan components of elite cues dominate. The rise of survey experiments has offered conflicting insights. We argue that the single-issue nature of most survey experiments masks systematic variation in elite cue effects across international issues, and that these effects depend on the baseline distribution of mass opinion on the issues themselves. Two characteristics of underlying opinion are crucial: the share of those not aligned with expert opinion, and the degree of partisan polarization. Where polarization is limited, information effects should dominate, but where issues are polarized, information intake should be limited by partisan attribution. We test these hypotheses using nine survey experiments across a range of issues, including the rise of China, climate change, international institutions, and the use of force. At one extreme, all messages, even those endorsed by generic or opposition experts, can shift opinion; at the other, only partisan-attributed messages matter. The findings are important not only for understanding public opinion about international issues, but also for mobilizing opinion in a democratic setting. Mapping the Boundaries of Elite Cues: How Elites Shape Mass Opinion Across International Issues International relations scholarship has taken to heart the idea that domestic politics shapes how states make foreign policy choices across a wide range of issues. As part of the debate about exactly how domestic politics matters, scholars have focused extensively on public opinion, not only in terms of the nature of public attitudes themselves, but also the extent to which these attitudes are malleable. Thanks in part to the rise of the survey experiment, there has been a surge of research on the determinants of public attitudes about international issues, including trade (e.g., Herrmann, Tetlock, and Diascro 2001; Hiscox 2006), security (e.g., Berinsky 2009; Tomz and Weeks 2013; Kreps 2014), and transnational issues such as climate change and international organizations (Bechtel and Scheve 2013; Tingley and Tomz 2014). The degree to which different messages and messengers can shift opinion is an important concern for policymakers seeking to mobilize domestic coalitions around a policy, particularly on international issues which are by nature distant from most voters’ everyday concerns and thus especially ripe for cue-giving by elite actors. Yet as more and more evidence from survey experiments accumulates, an important concern is that survey experiments usually proceed issue-by-issue and are rarely attentive to variation in issue context. Some of this variation may be idiosyncratic, but there may also be systematic variation across issues at a given time. For example, some issues like climate change may be more politically polarized at the time of a survey, while others may simply have received less attention. Variation across issue context is potentially crucial for understanding the impact of elite cues on public attitudes. Despite wide agreement that elite cues matter (see, for example, Herrmann, Tetlock, and Diascro 2001; Hiscox 2006; Berinsky 2009; Trager and Vavreck 2011; Levendusky and Horowitz 2012), there remains debate about how messages and messengers shift mass opinion. In one view, cues can convey, distill, and contextualize information about policies or events for the benefit of the voter (e.g., Gilens 2001; Hiscox 2006). From another perspective, however, voters use the identity of cue-givers—most commonly, their partisanship—as a shortcut (e.g., Zaller 1992; Berinsky 2009). While IR scholarship has tended to emphasize the role of information, recent research has brought partisanship to the fore, particularly in a security context (see Berinsky 2009; Trager and Vavreck 2011; Levendusky and Horowitz 2012). Each view finds support in specific issue areas even as other issues have been