sationalized stories about high-profile murders and mass shootings (i.e., the “mean world syndrome” [Gerbner et al., 1980, 1986]). Some evidence exists that these types of news stories are associated with unrealistic perceptions of low community safety (Chiricos et al., 2000; Ditton et al., 2004; O’Keefe, 1984) as well as, in some cases, secondhand traumarelated fear, depression, feelings of vulnerability, and PTSD (Ahern et al., 2002; Bernstein et al., 2007; Comer et al., 2008; Fremont et al., 2005; Otto et al., 2007; Saylor et al., 2003). The extent to which high exposure to such stories leads to changes in proneness to violence for the exposed individuals, though, has not been the subject of systematic research.
A number of longitudinal studies document long-term associations between violent media exposure in childhood and the later occurrence of real-life aggression or violence (Anderson et al., 2010; Boxer et al., 2009; Browne and Hamilton-Giachritsis, 2005; Eron and Huesmann, 1980; Eron et al., 1972; Huesmann, 1996; 2007; Huesmann and Taylor, 2006; Huesmann et al., 1984; Krahé and Möller, 2010; Savage, 2004; Savage and Yancey, 2008; Slater et al., 2003). Some studies have shown that children who favor violent television, movies, or video games or who are heavily exposed to these types of media have elevated rates of later aggression and violence, such as bullying, physical fights, spousal abuse, responding to insults with violence, committing and being convicted of crimes, violent delinquency, and committing moving traffic violations (Anderson et al., 2008; Hopf et al., 2008; Huesmann et al., 2003; Olson et al., 2009). However, the fact that these studies are nonexperimental introduces uncertainties in interpreting the associations they document because of the possibility that unmeasured common causes could account for the associations. Advocates of a causal interpretation of these associations have argued that a causal link is indirectly supported by evidence of dose–response relationships between the magnitude of exposure and subsequent violence (Anderson and Dill, 2000; Anderson et al., 2008; Huesmann et al., 2003) and by the fact that associations persist after introducing statistical controls for plausible confounders (Anderson et al., 2008, 2010; Huesmann et al., 2003; Olson et al., 2009). However, the adequacy of these controls has been disputed by critics (Ferguson, 2011; Ferguson et al., 2008, 2012; Savage, 2004).