Isolation as a Predictor of Homicide in Canadian Communities

INTRODUCTION:  Most violent crimes in the developed world are committed as part of an existing relationship, whether familial or gang-related or romantic.  Protective services often focus on enabling the termination of these relationships, but the efficacy of such services is difficult to prove because of self-selection. Those who choose to stay in a violent situation differ nonrandomly from those who choose to leave. 

 However, leaving is sometimes impossible. In the roadless communities of northern Canada, there is one ship a year and often no plane for weeks. Thus, if the ability to leave a situation predicts victimization, these communities should have elevated rates of violence. Using homicide as a well-reported and perfectly-defined proxy for all violence, I investigate whether the ability to leave a situation independently predicts the risk of violence. 

METHODS:  

Homicide data was mapped onto the 2006 Canadian census, allowing analysis of predictors of homicide rate at the unit of police district (n=898 police districts, of which 890 had complete data).  Predictors from the census included age structure, sex ratio, ethnic makeup, family structure, and education.

RESULTS:  

Communities without year-round road access (n=53, 46 complete) have higher rates of homicide 
than comparable communities with it (rate ratio 1.4, p<0.05) and the effect is stronger in Native-majority communities (n=133, rate ratio=1.6, p<0.02.) Homicide rate is predicted sometimes, but not always, by characteristics of individuals within that community: for example, sex ratio is predictive but a large young-adult cohort is not.

CONCLUSIONS:  

Communities that are difficult to leave, have an elevated rate of homicide. This effect gains strength when using ethnically matched controls, implying it is not a result of residual confounding by Native identity. This supports the hypothesis that revictimization can be prevented by easing and enforcing physical separation, such as with protective orders and minimum sentences.