After noting the challenges and benefits of working with Twitter data and briefly reviewing studies that used Twitter data to explore community responses to mass violence, this study demonstrated the use of Twitter data to examine community responses to a specific event, i.e., the 2015 San Bernardino terrorist attack, in which 14 people were killed and 22 were wounded.
Studying the community impact of mass violence using a Big Data approach from social media data (e.g., Twitter) offers traumatic stress researchers an unprecedented opportunity to study and clarify theoretical assumptions using large-scale, observational, ecologically valid data. In a 6-week time frame around the attack addressed in the current study, researchers examined the time course of community-level negative emotion. The study downloaded 1.16 million tweets, representing 25,894 users from San Bernardino, CA, and a matched control community, Stockton, CA. All tweets were coded in R using the Linguistic Inquiry and Word Count (LIWC) negative emotion dictionary. A piecewise regression technique with a discontinuity analysis was used to evaluate pre- and post-event trajectories of negative emotion across the study window. Controlling for within-user variability, negative emotion increased by 6.2 percent, β = .182, SE = .014, p < .001, in San Bernardino on the day of the attack and remained elevated for 5 days; no elevation was observed in Stockton. This article discusses how data-driven text analytic techniques are useful for exploring Twitter content generated after collective traumas and describes challenges and opportunities accompanying analyses of social media data to understand the impact of mass violence on affected populations. (publisher abstract modified)