The Power of Television Images in a Social Media Age: Linking Biobehavioral and Computational Approaches via the Second Screen

There is considerable controversy surrounding the study of presidential debates, particularly efforts to connect their content and impact. Research has long debated whether the citizenry reacts to what candidates say, how they say it, or simply how they appear. This study uses detailed coding of the first 2012 debate between Barack Obama and Mitt Romney to test the relative influence of the candidates’ verbal persuasiveness and nonverbal features on viewers’ “second screen” behavior—their use of computers, tablets, and mobile phones to enhance or extend the televised viewing experience. To examine these relationships, we merged two datasets: (1) a shot-by-shot content analysis coded for functional, tonal, and visual elements of both candidates’ communication behavior during the debate; and (2) corresponding real-time measures, synched and lagged, of the volume and sentiment of Twitter expression about Obama and Romney. We find the candidates’ facial expressions and physical gestures to be more consistent and robust predictors of the volume and valence of Twitter expression than candidates’ persuasive strategies, verbal utterances, and voice tone during the debate.