Selective Exposure to Political Advertisement via Physiological Response

• In Event: Election Campaigns: Advertising, Propaganda, and Media

Abstract

Abstract. This study investigates voters’ selective exposure to political advertisement that is issue-focused and sponsored by varied political entities (2x3 design). Both survey and physiological measures (eye-tracking and facial expression analysis) were utilized to gauge viewers’ responses to advertisements. Results show that voters’ attention to political advertisements is influenced more by partisan congruence than by issue congruence. Viewers’ emotions after selective exposure are significantly less negative but fail to be elevated. People’s self-reported interests in issues and their eye-tracking records do not match; neither do their stated discrete emotions and automatically coded facial expressions. The canonical variates extracted from self-report emotions and physiological measures correlate only moderately.

Voters’ attention and emotional responses to political messages are pivotal to ascertaining campaign effects. Politicians are vitally keen on capturing the attention and eliciting the emotion of the electorate to their favor. There has been an immense body of literature on how political campaigns can elevate candidate electability. The topical level on which assessment of campaigns is based often range from awareness, affective effect, to eventually behavior. Yet, one of the major methodological commonalities is that research tends to rely on self reports of surveyed respondents. Researchers usually need to take respondents’ words as their truthful answers albeit barriers that may decrease or hinder the fidelity.
Respondents’ physiological responses to political messages can be an alternative for researchers to gauge campaign effects. This route of gathering data may circumvent or curtail shortcomings of self-reports and provide additional evidence to either verify the traditional self-reported measures or shed new light on the measurement of effect and processing of political message. With this rationale in mind, the experiment employs both routes of measurement to investigate people’s responses to political advertisement with survey and eye-tracking and facial expression detections, which are intended to capture viewers’ cognitive and emotional reactions.
Results show that partisan congruence with advertisement’s sponsor predicts voters’ attention to the screen and negative emotions generated from viewing political commercials. Issue congruence, however, does not result in an increase in people’s attention to issue-centered messages and its relationship with viewers’ ensuing sentiment is only significant for some specific emotions. Overall, the study generates more confident physiological evidence supporting the impact of partisan selective exposure than that of issue-based selective exposure. The influence of selective exposure on people’s emotions appears to be negativity averse. On the other, people’s viewing of congruent messages does not result in positive emotions – which indicates a ceiling effect of political campaigns that aim to resonate. The specific impact of selective exposure on voters’ discrete emotions should be dissected further across political parties and issues.
The study also verified whether and to what extent self-reported and physiologically based measures of attention and emotional reactions correspond. The fact that the two distinct measures of cognition and affect based on self-reports and physiological assessments do not necessarily and consistently match with one another is very intriguing and definitely merits future investigation. Such findings raise concerns for solely relying on self-reported data in light of their discrepancies with the corresponding expressed emotions that one may argue occur at a subconscious level, potentially making triangulating measures an even more effective approach when researchers aim to capture more nuanced audience responses. Findings also showed that at the attention level, the issues that concern participants more do not lead to greater time they actually spent watching the ads that discuss them. With regards to emotion, the overall correspondence between the self-reported emotions and facially manifested counterparts is merely modestly positive – as the canonical correlation test indicates.
This study shines light on the inconsistent congruence between self reports and physiological measures. On the one hand, it is unclear whether the two methods are in effect measuring identical objects. There could be a number of potential dynamics for future researchers to consider. For one thing, people’s self-reported interests in any issue can be mitigated by confounding factors such as perceived social expectation, personal relevance, or test sensitization; likewise, their facial expressions made during political commercial viewing can be a result of their habitual reaction toward political advertisements in general, that particular election or candidate, or indeed triggered by the discussed issue. Given these, it is too early to jump to the conclusion that these two distinct methods do not match.

Authors

• Denis Wu, Boston University

• Mina Tsay-Vogel, Boston University