Emotions : Facial Expressions as a Measurement & Effects on Political Attitude

Abstract

Emotions guide human behavior in all facets of life. In politics, emotions impact for example attitudes towards policy issues or how one makes a voting decision. In this dissertation, I advance our knowledge on emotions in political science by studying a new measurement technique of emotions and I investigate further how emotions impact attitudes towards candidates.<br /><br />In the first study (Chapter 2), Tim Höfling and I show that off-the-shelf facial expression recognition systems produce valid measurements of emotional expressions in controlled laboratory settings for clear and prototypical emotional expressions. However, we further find significant performance problems on data that is more `messy', characterized by varying camera angles, imperfect lighting and more variability in facial expressions.<br /><br />In the second study (Chapter 3), I explore in a lab experiment whether emotional political speeches trigger emotions in the audience via emotion contagion and whether these emotions impact populist or extremist attitudes. I find no indication of emotion contagion from the speeches and thus no systematic differences in emotions between the treatment groups. At the same time, observational analyses show that angrier subjects report more populist attitudes and take less time to express them. These findings yield support for Affective Intelligence Theory and the hypothesis that anger is the driving emotion behind more populist and extremist attitudes.<br /><br />In study three (Chapter 4), I investigate further the effects of emotions elicited in political speeches on candidate evaluations. Such effects were also observed in Chapter 3 and other studies. I show that the effect of an emotion experienced by a person on candidate evaluation of a politician varies with prior political attitude of the person towards the politician. This calls earlier research into question, which (implicitly) assumes a constant effect of an experienced emotion on candidate evaluation. Furthermore, I show that the Appropriateness Heuristic provides a valuable extension to Affective Intelligence Theory to account for varying effects of emotions between politicians.<br /><br />In sum, I make two contributions to science with this dissertation. First, my work shows that caution is appropriate when applying off-the-shelf facial expression recognition tools as measurement strategy in the social sciences and other behavioral research areas, especially in less controlled environments. Second, I confirm that emotions towards a politician impact the evaluation of said politician. Additionally, I show that this effect varies with prior attitude towards the politician. Similarities and deviances of results shown in this dissertation with other research highlight the importance of studying effects of emotions in politics in different cultural and political circumstances.