Automatic Machine Coding of Verbal and Nonverbal Charismatic Signaling
Can charismatic signaling be automatically coded by machine? If yes, can we use them to predict social influence? Previous research already provided evidence that charismatic signaling used by informal distant leaders has a positive effect on their social influence. However, they relied on human coding of charismatic signaling which might be biased. Alternatively, using machine coding of verbal and nonverbal charismatic signaling reduces the risk of human biases and endogeneity issues. Accordingly, in our study, we used machine-based coding–instead of human coding–of verbal and nonverbal charismatic signaling as well as talk-based material to control for static attractiveness, on a larger dataset (769 TED talk videos). Our results show that human codings–and not machine codings–of nonverbal charismatic signaling are influenced by attractiveness and gender and that verbal charismatic signaling has a positive effect on social influence and perception of leadership, above and beyond nonverbal charismatic signaling and static attractiveness. We discuss the added value of using machine coding of charismatic signaling as a beneficial alternative to human coding.