Radil, Flint & Tita (2010) take into consideration both the socio-relational and geographic components of gang violence to examine the distribution of rivalries and amounts of violence in an area of East Los Angeles referred to as Hollenbeck. The aim of their study was to explore whether social networks (in this case, rivalries between certain gangs) could be used in conjunction with the spatiality of gangs to partially explain their behaviour (in this case, gang violence).
Considering the situation purely from a geographic point of view, Tita (2006) found through use of a global Moran’s I test that there was only very weak positive spatial autocorrelation of gang violence in Hollenbeck. Thus Radil, Flint & Tita considered the social relations between gangs as a partial explanation of where gang violence occurred. To do this they used a network analysis technique called CONCOR (convergence of iterated correlations), which recursively divides census block groups based on both geographic embeddedness (spatiality of gangs and gang violence) and network positionality (rivalries between gangs).
Unsurprisingly, the first split resulted in a north-south division of the area which can be explained by landscape: they are on opposite sides of a major highway. The results become more interesting in the second split, which divides the northern gang turf into a center-periphery arrangement, that can only be explained by network positionality and amounts of gang violence. The southern division followed the same pattern as the first split and was divided seemingly geographically into another north-south orientation. The third split continues to suggest center-periphery arrangements of gang turf, in which turf in the central areas has both higher amounts of gangs rivalling over it, and thus greater amounts of gang violence, while turf in the periphery areas have lower amounts of rivalry and violence.
The third split reveals the existence of a spatiality referred to as geographical betweenness: areas composed of the turf of several different gangs are more similar to each other in the amounts of gang violence than to other areas. At the same time, the study shows that relational betweenness also leads to similarity between areas in the amounts of gang violence experienced. Some areas are composed of the turf of only one gang, but experience similar amounts of violence due to the gang’s relational position between rival gangs that also happen to be rivals of each other.
While one could guess that geographic and relational betweenness are important to think about when considering levels of gang-related violence, it is great that Radil, Flint & Tita were able to find a way to actually model these behaviours using network analysis. Hopefully this study will encourage future use of social network analysis in GIScience to investigate the embeddedness of social behaviour across space.
-yee
Tags: yee