Graduate School of Political Science, Waseda University, Building No.3 1-6-1 Nishiwaseda, Shinjuku-ku, Tokyo 169-8050, Japan.
Osaka School of International Public Policy, Osaka University, 1-31 Machikaneyama, Toyonaka, Osaka 560-0043, Japan.
Soc Sci Med. 2018 Dec;219:19-29. doi: 10.1016/j.socscimed.2018.10.004. Epub 2018 Oct 12.
Rises in suicide rates following media reports of the deaths by suicide of public figures are a well-documented phenomenon. However, it remains unclear why, or by what exact mechanism, celebrity suicides act to increase suicidal risk in the wider public due to the lack of data showing how the public understands and reacts to the suicide of well-known figures. This study used a supervised machine learning approach to investigate the emotional content of almost 1 million messages sent on Twitter related to the suicides of 18 prominent individuals in Japan between 2010 and 2014. The results revealed that different demographic characteristics of the deceased person (age, gender, and occupation) resulted in significant differences in emotional response; notably that the suicides of younger people, of women and of people in entertainment careers created more emotional responses (measured as a ratio of emotionally-coded tweets within the overall volume of tweets for each case) than for older people, men, and those in other careers. Moreover, certain types of emotional response were shown to correlate to subsequent rises in the national suicide counts, with "surprised" reactions being positively correlated with the suicide counts, while a high proportion of polite messages of condolence were negatively correlated. The study demonstrates the importance of, and describes a methodology for, considering the content of social media messages, not just their volume, in research into the mechanism by which these widely-reported deaths increase suicide risk in the broader public.
媒体报道公众人物自杀身亡后，自杀率上升是一个有据可查的现象。然而，由于缺乏数据表明公众如何理解和对知名人士的自杀做出反应，因此仍不清楚为什么名人自杀会导致更广泛的公众的自杀风险增加，以及确切的机制是什么。本研究使用有监督的机器学习方法，调查了 2010 年至 2014 年间日本 18 位知名人士自杀事件后，将近 100 万条在 Twitter 上发布的与自杀相关的消息的情绪内容。结果表明，死者的不同人口统计学特征（年龄、性别和职业）导致了情绪反应的显著差异；特别是年轻人、女性和娱乐业人士的自杀引起了更多的情绪反应（以每个案例中情感编码的推文数量与总推文数量的比例来衡量），而年龄较大的人、男性和其他职业的人则没有。此外，某些类型的情绪反应与随后的全国自杀人数上升有关，“惊讶”反应与自杀人数呈正相关，而大量礼貌的吊唁信息则呈负相关。该研究表明了考虑社交媒体消息内容的重要性，并描述了一种方法，不仅要考虑消息的数量，还要考虑这些广泛报道的死亡事件如何增加更广泛公众的自杀风险的机制的研究。
