REVIEW: COMMUNITY FACTORS AND VIOLENCE
On the whole, correlations between community factors and violent crime have been relatively consistent. Ethnic heterogeneity and residential instability are both consistently positively related to measures of violence (Barnett & Mencken, 2002; Crutchfield et al., 1982; Hannon & DeFronzo, 1998; Hipp, Tita, & Boggess, 2009; Kposowa et al., 1995; Krivo & Peterson, 1996; Miethe, Hughs, & McDowall, 1991; Oh, 2005; Patterson, 1991; Peterson & Krivo, 2010; Sampson & Groves, 1989; Van Wilsem, Wittenbrood, & De Graaf, 2006; Warner & Pierce, 1993; Warner & Rountree, 1997; Wilcox et al., 2004), with periodic exceptions (e.g., Kennedy, Silverman, & Forde, 1991; Kubrin & Weitzer, 2003b; O’Shea, 2006; Patterson, 1991;
Steffensmeier & Haynie, 2000; Warner & Pierce, 1993; Worrall, 2005). Measures of social and/or physical disorder, such as abandoned buildings, litter, loitering, or public drunkenness, are also frequently positively related to violence (O’Shea, 2006; Sampson & Raudenbush, 1999; Skogan, 1990; Wilcox et al., 2004). While the percentage of community residents who own their own home, which could be seen as an indicator of residential stability, is so far consistently, negatively associated with violence (Hipp, 2007; Jobes, Barclay, Weinand, & Donnermeyer, 2004), other measures of residential stability, such as the average length of residence in the community, sometimes generate the expected negative relationship with violence (Rountree & Warner, 1999; Sampson et al., 1997; Warner & Fowler, 2003), sometimes generate an unexpected positive relationship (Hipp et al., 2009), or fail to generate a significant relationship (Bellair, 2000; Kaufman, 2005; Morenoff et al., 2001; Warner & Rountree, 1997).
Other community structural characteristics vary in the consistency with which they affect violence. Findings from studies using measures of family disruption are mixed, and this seems to be related to the measure used. The percentage of divorced community residents generates a high ratio of positive and significant associations (Hipp, 2007; Kennedy et al., 1991; Kposowa et al., 1995; Sampson, 1986) compared to nonsignificant associations (Brezina, Agnew, Cullen, & Wright, 2004). About half of the studies using measures such as the “percent dualparent," “single-parent," or “female-headed" households report significant positive correlations with violence (Hipp et al., 2009; Kelly, 2000; Rountree & Warner, 1999; Sampson, 1986; Sampson & Groves, 1989; Steffensmeier & Haynie, 2000 ) with the other half failing to demonstrate statistically significant associations (Oh, 2005; Patterson, 1991; Stewart & Simons, 2006; Worrall, 2005). The reader should note, however, that the proportion of associations between indicators of family disruption and violence exceed chance levels by a wide margin.
As with the studies that examine total crime or general offending, the behavior of residents within communities appears to exert more consistent effects on the occurrence of violence. Measures of the presence of neighborhood ties or the extent of interaction between neighbors frequently generate the expected negative relationship (Bellair, 1997, 2000; Lowenkamp, Cullen, & Pratt, 2003; McNulty & Bellair, 2003; Patterson, 1991, Rountree & Warner, 1999; Sampson & Groves, 1989; Veysey & Messner, 1999; Warner & Rountree, 1997; Wilcox et al., 2004), though these associations are sometimes not statistically significant (De Coster, Heimer, & Wittrock, 2006; Morenoff et al., 2001; Sampson & Groves, 1989; Wilcox et al., 2004). Studies examining residents’ willingness to intervene to solve collective problems find that this “collective efficacy" is consistently negatively related to violence (Browning, Feinberg, & Dietz, 2004; Kirk, 2009; Morenoff et al., 2001; Sampson & Raudenbush, 1999; Sampson, Raudenbush, & Earls, 1997). Results are more mixed in regard to the relationship between residents’ participation in local organizations and crime, with Veysey and Messner (1999) finding a negative relationship, but Lowenkamp and his colleagues (2003) finding no significant relationship. Finally, we observe that measures of the extent to which community residents hold violent values stand alongside measures of collective efficacy as being the most consistent predictors of community violence (Bernburg & Thorlindsson, 2005; Brezina et al., 2004; Felson, Liska, South, & McNulty, 1994; Heimer, 1997; Markowitz & Felson, 1998; McNulty & Bellair, 2003; Ousey & Wilcox, 2005; Stewart & Simons, 2006).