Year of walking outdoors - A multilevel Ecologic analysis of crime and disorder

Background: Although a number of studies have tested ecologic models that postulate relationships among social networks, the built environment, and active living, few neighborhood-based studies have considered the role of crime and violence. This study investigates the degree to which individual-level demographic characteristics and neighborhood-level physical and social characteristics are associated with increased fear of crime. Methods: Data were analyzed in 2007 from a 2005 survey of 901 randomly selected individuals living in 55 neighborhoods in Washington DC. Multilevel ordered logit regression was used to examine associations between individual-level and neighborhood-level characteristics and how often fear of crime prevents a respondent from walking outdoors. Results: Age and female gender were associated with an increase in fear; the percentage of a resident's life spent in the same neighborhood was associated with a decrease in fear. Results of cross-level interactions showed that at the neighborhood level, women were more fearful than men in neighborhoods without violence, but that the difference in fear between men and women shrinks as neighborhood violence increases. Collective efficacy was found to increase fear among black respondents and had no effect on fear among nonblack respondents. Conclusions: If the study of neighborhoods and active living is to progress and contribute to both etiologic understanding and policy formation, it is essential that theoretical and empirical models consider the impact of violence and fear on walking. Efforts to increase active living in urban neighborhoods that do not account for the impact of crime and fear may fall short of their intended outcomes.