Applying Natural Language Processing to Evaluate News Media Coverage of Bullying and Cyberbullying

Bullying events have frequently been the focus of coverage by news media, including news stories about teens whose death from suicide was attributed to cyberbullying. Previous work has shown that news media coverage is influential to readers in areas such as suicide, infectious disease outbreaks, and tobacco use. News media may be an untapped resource to promote bullying prevention messages, though current news media approaches to describing bullying and cyberbullying remain unexplored. The purpose of this study was to evaluate the current state of news media coverage of bullying and cyberbullying. A sample of newspaper articles covering bullying or cyberbullying across regional and national US newspapers from 6 recent years was identified. A content analysis using natural language processing was conducted with the Linguistic Inquiry and Word Count (LIWC) software program for key variables including affective, social, and cognitive processes. Evaluation included the percentage of words that represented Fear-based reporting such as alarmist words (e.g., epidemic, tragic), as well as words that represent Public Health-oriented messages such as prevention. A total of 463 newspaper articles met inclusion criteria, including 140 cyberbullying articles and 323 bullying articles. Findings indicated that cyberbullying articles scored higher on affective processes such as measures of anxiety (Mdn = 0.34) compared to bullying articles (Mdn = 0.22). A greater number of cyberbullying articles were Fear-based (41.4%) than were bullying articles (19.5%). An equivalent number of cyberbullying articles (50.0%) and bullying articles (49.8%) were Public Health-oriented. Findings may be used to collaborate with journalists toward optimizing prevention-oriented reporting.