Abstract
Since the publication of The Broken Windows Theory and further theoretical developments there has been the widespread idea that a vandalised neighbourhood covered with graffiti, litter, and broken windows, increases physical and social disorder, fear, unsafe feelings and crime. Most research efforts have attempted to elucidate the implications of physical and social incivilities in order to find straight and relatively easy ways of creating safer environments. In this thesis such a standpoint is challenged. Here it is argued that if we are to explain people's perceptions of safety about victimisation we cannot just rely on it being a function of perceptions of physical and social disorder but more importantly of risk perception. Thus, a less deterministic and simplistic approach incorporating psychological variables is needed. This thesis critically analyses the theoretical frameworks traditionally used to explain fear of crime and perceptions of safety and puts forward a new conceptual model suggesting that risk perception partially mediates the relationship between perceptions of disorder and safety. It incorporates personal, psychological, social, and contextual factors that are said to influence perceptions of safety about victimisation in residential neighbourhoods. In addition, the notion of disorder is revised. Three empirical studies combining qualitative and quantitative methodologies were carried out to build each component of the model and test for the relationships proposed. Findings demonstrated that it is not the presence of incivilities what matters the most rather it is people in places what make us feel unsafe, and the physical environment provides cues that enable individuals to infer social attributes of places and people associated with those places. Research results also confirmed that risk perception partially mediates the relationship between perceptions of disorder and safety. In short, the more disordered a place is perceived the more a person relies on the perception of risk to estimate how safe she or he might be. In this sense, the relationships established in the model were confirmed, except for the influence that socio-demographic characteristics of the perceiver have in perceptions of safety since neither gender nor age or victimisation experience predicted perceptions of safety. It is concluded that cues to danger and safety are neither self-evident nor in the environment but in people's minds. It appears that what has guided designers, policy makers, police action plans, and even several researchers, for so many years, is now questioned.