A Quantitative Approach to Evaluate and Develop Theories on (Fear of) Crime in Urban Environments

Well established work in criminological, architectural and urban studies suggests that there is a strong correlation between crime, perceived safety, the fear of crime, and the presence of different demographics, the people dynamics, in an urban environment. These studies have been conducted primarily using qualitative evaluation methods, and are typically limited in terms of the geographical area they cover, the number of respondents they reach out to, and the temporal frequency with which they can be repeated. As cities are rapidly growing and evolving complex entities, complementary approaches that afford social and urban scientists the ability to evaluate urban crime and fear of crime theories at scale are required. In this thesis, I propose a combination of methodologies following a data mining and crowdsourcing approach to quantitatively validate these theories at scale, and to support the exploration of new ones. To relate people dynamics to crime quantitatively, I first analyse footfall counts as recorded by telecommunication data, and extract metrics that act as proxies of urban crime theories. Using correlation and regression analysis between such proxies and crime activity derived from open crime data records, the method can help to understand to what extent different theories of urban crime hold, and where. To relate people dynamics to fear of crime quantitatively, I then built two image– based online crowdsourcing platforms to investigate to what extent online crowdsourcing can be used to gather safety perceptions about urban places, defined by the combination of built environment and the people inhabiting it. As existing theories suggest that knowing who the respondents are is crucial for understanding safety perceptions, I also gathered their demographic background information to discuss their perceptions accordingly. I applied analysis of variance (ANOVA) and covariance (ANCOVA) to these data. The method can help to understand what visual properties based on peopl