�����������������������������������������������������������������
as risk of developing depressive symptoms ( Yen & Kaplan , 1998 ; Yen & Kaplan , 1999 ). In another study in Alameda County , clusters of socio-demographic , socioeconomic , and environmental quality measured at the zip-code level were prospectively associated with risk of death . In that study , respondents who lived in zip codes with elevated mortality were more likely to have a wide variety of social , socioeconomic , behavioral , and socio-environmental risk factors such as low education and income , being unable to fill a prescription , obesity , social isolation , living in an unsafe neighborhood , being a crime victim , physical inactivity , and current smoking ( Kaplan , 1996 ).
More recent work connecting neighborhood conditions to health outcomes has focused on traffic , noise , crime , trash and litter , lighting , and public transportation ( Stark et al ., 2013 ), the food environment ( Stark et al ., 2013 ), segregation ( Acevedo-Garcia et al ., 2003 ), collective efficacy and social capital ( Sampson , Raudenbush , & Earls , 1997 ), walk ability ( Auchincloss et al ., 2013 ), deprivation level ( Cubbin , Hadden , & Winkleby , 2001 ), and many more neighborhood characteristics ( Kaplan , 1996 ). Much of this work takes the neighborhood environment as fixed and given , and does not interface with the large literature on the factors that contribute to neighborhood formation and change ( Sampson , 2013 ; Sharkey , 2013 ; Wilson , 1997 ). However , neighborhoods are the results of historical and contemporary interacting forces acting over time . Figure 3 portrays just a few of the forces acting on neighborhoods that shape the racial and economic composition of those neighborhoods , both strongly associated with health , and to that we could add the social and political forces that drive metropolitan fragmentation ( Orfield , 2002 ), also related to health outcomes and health disparities ( Andre Hutson , Kaplan , Ranjit ,
& Mujahid , 2012 ).
As in Figure 2 , Figure 3 leaves out many potentially important pathways , but even in this simplification we see a dense network of interacting forces , operating over time and over the life course , that determine the composition , resources , and risks associated with living in a particular place . It is this dynamic interplay of determinants , across numerous scales and levels , that determines the composition and character of places , the risk , resources , and opportunity structures in those places , and it is the complexity of these processes that is masked in much of the neighborhood-health literature . It is unlikely that simple or multi-level regression techniques are up to the task of capturing the complexity of this system , and identifying independent effects seems implausible . Thus , it should come as no surprise when the results of interventions at the neighborhood level do not necessarily lead to anticipated outcomes ( Kessler et al ., 2014 ).
There are several features worth noting about both socioeconomic and neighborhood effects on health , as represented in Figures 2 and 3 . First , both seem to require a multi-level perspective that includes individuals , families , communities , and institutions , plus all the more proximal biological pathways that led to health outcomes . They also represent what we can characterize as multi-scale systems , with multiple temporal and spatial scales . In addition , they are systems in which time and the dynamic processes over time , at many scales , are critical . These multi-level , multi-scale , and temporal processes , with considerable positive and negative feedback are the ����� ���� ��� of complex systems .
While the patterns shown , which are surely , an under-specification of the important pathways , are complicated , they are more than that , they are complex . Com-
92