Does reciprocal gratefulness in twitter predict neighborhood safety?: Comparing 911 calls where users reside or use social media

Abstract

Is there a relationship between urban neighborhood safety and helpful or supportive user networks on Twitter? An interdisciplinary, community-partnered team analyzed one year (2013-2014) of geo-tagged tweets from a 16-county region to generate a network of users who expressed gratefulness to one another. Call counts to 911 (2013-2015) around locations in the urban center indicate safety-oriented activity in residential areas. We compared frequencies of 911-related police activity across geographic pockets (200m radius or 200×200 meter2 areas) where mutually helpful or thanked users lived or frequently tweeted, to pockets randomly selected in proximity. The 13 naturally helpful users with predicted home locations in this city lived where fewer 911 calls for police service were initiated over time, on average. Our results show that close neighborhood locations where 79 naturally helpful users had strong local geographic ties in this urban area, as evidenced by Twitter use patterns, functioned as safe zones needing fewer policing services on average than surrounding areas. We discuss the implications of predicting real-world needs for police services based on supportive qualities of one's social media neighborhoods.