Project Title: What Makes Citizens Click? Testing the Impact of Emotional Discourse with Digital-Age Field Experiments
Faculty Sponsor and PI: Ted Brader
Ph.D. Student: Timothy J. Ryan
How do citizens learn about politics? In an age of abundant information and highly differentiated media, the issue is not so much whether citizens can become informed, but rather which sources they consult and how much they choose to consume. Roy Pierce, like many scholars, took a strong interest in the vigilance of democratic citizens and the roots of representation. Along the way, he delved deeply into how partisanship and ideology shape perceptions and choices. Prior research reveals that citizens sometimes engage in selective exposure—seeking out information to confirm their own partisan point of view—but other times do not. What explains these fluctuations in selective exposure? And if citizens are more responsive to some types of appeals than others, how does this influence the incentives elites have to engage in different types of discourse?
We wish to explain why citizens seek out political information and what sorts of information they seek. Recent research suggests that emotions motivate information-seeking as well as other forms of political action. This line of work sheds light on how specific emotions—e.g., fear, enthusiasm, anger—whet or dull the appetite for information generally. We posit that emotions evolved through natural selection to elicit behavior directed at specific goals. As a result, a politician trying to lure voters to her website have to consider what voters find emotionally arousing, and whether the information available helps resolve the aroused tension: Does available information serve the goals activated by a person’s emotional state? We hypothesize that citizens are more likely to consume information when it satisfies emotion-consistent goals.1 We plan to test this proposition with randomized field experiments over the Internet. While scholars have learned much about emotions and behavior from lab experiments and surveys, field experiments deployed unobtrusively into citizens’ “natural environment” extend tests to an increasingly important context while avoiding pitfalls that can arise when research subjects are aware their behavior is being observed.
Researchers have shown that emotions powerfully predict opinion and behavior, and political messages can elicit emotions that alter the very process by which decisions are made. Affective Intelligence Theory (AIT) represents the best-known effort to construct an over-arching model of emotion and politics. Borrowing from affective neuroscience, AIT posits two emotional systems that monitor the environment to allocate thinking and behavior efficiently in accordance with situational needs. The disposition system modulates enthusiasm based on feedback from current pursuits; the surveillance system modulates anxiety in response to threats.
Noting several empirical inconsistencies in past work, we believe the insights of AIT must be revised and expanded to take fuller notice of the function and goal-orientation of multiple emotions. For this purpose, we draw on appraisal theories of emotion from social psychology. This class of theories calls our attention to subtle differences in the relationship between an individual and his situation and are instructive on how arousal and context intersect to produce behavior. Inconsistencies likely arise not because emotions work differently across situations, but rather because we have been too broad in defining their behavioral implications.We propose to test a broader view of arousal in which emotions activate specific goals, rather than simply more or less information seeking. Evolutionary psychology suggests that emotions serve a directive function, having evolved to orchestrate the activity of specialized mental subsystems and, are relevance detectors that allocate attention, physiological resources, etc., as circumstances demand. In short, emotions are adaptive responses to the environment that define behavioral goals and elicit behavior in keeping with those goals.
If the perspective above is correct, however, most current studies, which seek a direct link between emotion and behavior, will systematically miss an important contextual factor: the relationship between emotional arousal and the way in which various behaviors might be useful. It may well be the case that anxiety does not elicit information seeking inherently, but only information seeking useful for protection against a looming threat. Similarly, anger might not dampen information seeking per se. Rather, it might encourage information seeking that is useful for a more narrow goal, such as punishing others. If true, such a reassessment has the potential to explain a number of previously inconsistent findings. It also has the potential to explain why politicians – at least in some forums – find certain kinds of discourse so appealing. For instance, perhaps political blogs gravitate toward disdainful, overheated rhetoric because the anger these styles elicit effectively produces the desired effect among readers.
Although political scientists have begun to use field experiments to test voter mobilization efforts and other interventions, we are unaware of any field experiments designed to test hypotheses about emotions or selective exposure. We plan to test the propositions with regard to specific emotions and goals using a field experiment conducted online. In the course of routine browsing on social networking or commercial websites, real web users will be exposed to advertisements inviting them to “click for more information” on a political topic. Two properties of the ads will be manipulated at random. First, the ads will contain imagery designed to elicit a specific emotion (e.g., anger, anxiety, no emotion). Second, the ads will contain a brief message suggesting what goal might be served by clicking. For instance, because evolutionary perspectives link anxiety with protection from physical danger, an ad about health care might suggest that clicking will reveal how “to protect your health.” In some conditions, emotional arousal will match the goal. In others, there will be a mismatch. We hypothesize that both anxiety and anger can elicit information seeking as long as there is a match between the goal activated by arousal and the goal served by behavior. A pilot study along these lines, designed and conducted by Ryan (currently under review), testifies to the soundness of this innovative approach.
This summer we plan to design a full slate of treatments, conduct brief manipulation checks, and, finally, run the experiments. Turnaround time is faster than in conventional data collection, allowing us to field and learn from successive trials in short order. We also will design destination websites carefully constructed to allow collection of a richer set of dependent measures, once people have clicked on an ad. For example, the websites will allow us to track the extent to which people consume information that is 1) confirmatory or non- confirmatory; 2) defensive or punitive; and 3) threat oriented or non-threat oriented. Each of these pairs captures an element of behavioral goals, and their addition would allow for a more nuanced examination of behavior once initial searching has begun. This design can be adapted to study the effects of several emotions which have received very little attention to date. We are especially enthusiastic about the field experiments because they offer a test unmatched in realism and well attuned to study the explosion of “new” social media.
Our proposal is not merely an extension of one or the other’s research agenda, but rather a genuinely collaborative project that grew out of related interests we had developed separately. In this project, Ryan builds on expertise in field experimentation begun in college and his growing interest in the political implications of evolutionary psychology. These interests led Ryan to formulate hypotheses and test ideas for field experiments on emotions. Quite independently, Brader has been contemplating for years remarkably similar “next steps” in his long-running research on the behavioral effects of political emotions and has had a desire to pursue some of this research through field experimentation for the first time. We are excited to discover this convergence in our research agendas and decided to reap the benefits of full collaboration. Ryan is currently testing theoretical and methodological approaches to play a role in his dissertation. Because it incorporates his longstanding interest in emotions, evolutionary psychology, and field experimentation, the project very much advances that effort. Moreover, Ryan hopes to develop more experience on large-scale original data collection, on professional presentations of findings (a poster related to this vein of research has been accepted at APSA), and writing up results for publication.