Social bots versus real humans : the framing of 'Trump's Wall' on Twitter

Abstract

Around the globe, elites are using social media and computational propaganda to manipulate public opinion, (Bradshaw and Howard, 2018) increasingly degrading the traditional news media's gatekeeping function while building a symbiotic relationship with ideological media that forgo objectivity (Entman and Usher, 2018). Although a number of studies have examined framing of content in social media, including Twitter, no study known to the author has isolated bot-generated tweets to understand if they are capable of framing issues and how they frame them compared to real humans. The current research explores this issue by using a machine-learning (ML) software that detects whether a post came from a social bot account versus a real human with up to 100 percent accuracy for political bots (Yang et al., 2019). After an extensive manual data collection procedure, the current research goes through three steps: 1) identify whether a Twitter post originated from a social bot vs. real human, 2) determine the frame(s) and sentiment used in the post, 3) determine if the results fall in line with an asymmetrical cascading network activation model where the posting of right-leaning content is more automated than left-leaning content (Entman and Usher, 2018). To explore the existence of a new cascading network activation model that is asymmetrical, the content examined had to be polarizing. Thus, the context selected for the study is President Donald Trump's current Mexico border wall campaign. In addition to the data capture method described, a content analysis method is also utilized to make comparisons between frames used by social bots versus real humans and the right versus the left.