Crime is not spread evenly over space or time. This suggests that offenders favour certain areas and/or certain times. People base their daily activities on this notion and make decisions to avoid certain areas or feel the need to be more alert in some places rather than others. Even when making choices of where to stay, shop, and go to school, people take into account how safe they feel in those places. Crime in relation to space and time has been studied over several centuries; however, the era of the computer has brought new insight to this field. Indeed, computing technology and in particular geographic information systems (GIS) and crime mapping software, has increased the interest in explaining criminal activities. It is the ability to combine the type, time and spatial occurrences of crime events that makes the use of these computing technologies attractive to crime analysts. This current study predicts robbery crime events in the City of Tshwane Metropolitan Municipality. By combining GIS and statistical models, a proposed method was developed to predict future robbery hotspots. More specifically, a robbery probability model was developed for the City of Tshwane Metropolitan Municipality based on robbery events that occurred during 2006 and this model is evaluated using actual robbery events that occurred in the 2007. This novel model was based on the social disorganisation, routine activity, crime pattern and temporal constraint crime theories. The efficacy of the model was tested by comparing it to a traditional hotspot model. The robbery prediction model was developed using both built and social environmental features. Features in the built environment were divided into two main groups: facilities and commuter nodes. The facilities used in the current study included cadastre parks, clothing stores, convenience stores, education facilities, fast food outlets, filling stations, office parks and blocks, general stores, restaurants, shopping centres and supermarkets. The key commuter nodes consisted of highway nodes, main road nodes and railway stations. The social environment was built using demographics obtained from the 2001 census data. The selection of these features that may impact the occurrence of robbery was guided by spatial crime theories housed within the school of environmental criminology. Theories in this discipline argue that neighbourhoods experiencing social disorganisation are more prone to crime, while different facilities act as crime attractors or generators. Some theories also include a time element suggesting that criminals are constrained by time, leaving little time to explore areas far from commuting nodes. The current study combines these theories using GIS and statistics. A programmatic approach in R was used to create kernel density estimations (hotspots), select relevant features, compute regression models with the use of the caret and mlr packages and predict crime hotspots. R was further used for the majority of spatial queries and analyses. The outcome consisted of various hotspot raster layers predicting future robbery occurrences. The accuracy of the model was tested using 2007 robbery events. Therefore, this current study not only provides a novel statistical predictive model but also showcases R’s spatial capabilities. The current study found strong supporting evidence for the routine activity and crime pattern theory in that robberies tended to cluster around facilities within the city of Tshwane, South Africa. The findings also show a strong spatial association between robberies and neighbourhoods that experience high social disorganisation. Support was also found for the time constraint theory in that a large portion of robberies occur in the immediate vicinity of highway nodes, main road nodes and railway stations. When tested against the traditional hotspot model the robbery probability model was found slightly less effective in predicting future events. However, the current study showcases the effectiveness of the robbery probability model which can be improved upon and used in future studies to determine the effect that future urban development will have on crime. / Dissertation (MSc)--University of Pretoria, 2020. / Geography, Geoinformatics and Meteorology / MSc / Unrestricted

No description available.

Background and objectives: Childhood sexual abuse (CSA) is a substantial public health and human rights problem, as well as a growing concern in sub-Saharan Africa (SSA). It has both short and long term effects on girls: physical and psychological, including negative sexual outcomes. Up to one-third of adolescent girls report their first sexual experience as being forced. Despite growing evidence supporting a link between contextual factors and violence, no studies have investigated the connection between CSA and contextual factors. It is therefore important to identify the extent of CSA and understand factors associated with it in SSA in order to develop interventions aimed to address the scale of the problem. Aim: The overall aim of this thesis is to assess the individual and contextual factors associated with CSA. In addition, the thesis aims to quantify the magnitude of CSA and describe the factors associated with CSA among women from SSA (Study I). This thesis also examines the independent contribution of individual and community socio-economic status on CSA (Study II). Moreover, it scrutinises the effect of social disorganisation on CSA (Study III) and explores the relationship between CSA and sexual risk behaviours as well as potential mediators (Study IV). Methods: This thesis used the Demographic and Health Survey (DHS) datasets conducted between 2006 and 2008 from six SSA countries. The thesis used multiple logistic regression models to describe and explore factors associated with CSA among 69,977 women (Study I).  It used multivariable multilevel logistic regression analysis to explore the effect of contextual level variables (neighbourhood socio-economic status) on CSA among 6,351 girls (Study II). Neighbourhood socio-economic status was operationalized with a principal component analysis using the proportion of respondents who were unemployed, illiterates, living below poverty level and rural residents. Study III applied multivariable multilevel logistic regression analysis on 6,351 girls and considered five measures of social disorganisation at the community level: neighbourhood poverty, female-headed households, residential mobility, place of residence, population density, and ethnic diversity. In study IV, 12,800 women from the Nigerian DHS were used. Structural equation modelling was applied using a two-step approach. The first step used a confirmatory factor analysis to develop an acceptable measurement model while the second step involved modifying the measurement model to represent the postulated causal model framework. Results: In study I, the reported prevalence of CSA ranged from 0.3% in Liberia to 4.3% in Zambia when the prevalence was based on all respondents aged between 15 and 49 years and who were present during the survey. None of the socio-economic factors were associated with CSA. In study II, where the data was restricted to permanent residents aged between 15 and 18 years, the prevalence ranged between 1.04% in Liberia to 5.8% in Zambia. At the individual level, there was no significant association between CSA and wealth status while at the community level, there was no significant association between CSA and socio-economic position. However, 22% of the variation in CSA was attributed to the community level factors. In study III, there was significant variation in the odds of reporting CSA across the communities, with community level factors accounting for 18% of the variation. In addition, respondents from communities with a high family disruption rate were 57% more likely to have reported sexual abuse in childhood. Study IV showed that there was a significant association between CSA and sexual risk behaviours and the association was mediated by alcohol and cigarette use. Conclusions: The study provides evidence that adolescents in the same community were subjected to common contextual influences. It also highlighted the significance of mediators in the relationship between CSA and sexual risk behaviours. It is therefore important that effective preventive strategies are developed and implemented that will cut across all socio-economic spheres in a context that both permits and encourages disclosure as well as identifying predisposing circumstances for recurrence.

Mental health problems have increased among adolescents in Sweden and research suggests that contextual matters could be of importance over and beyond individual socio-demographic characteristics. One such social context is school, where both the student composition of the school and its support can influence student health. This study explored the distribution of psychosomatic health complaints (PHC) and perceived teacher support (PTS) as well as the association between PTS and PHC, across school segregated profiles. It also examined gender differences in these distributions and associations. The study design was cross-sectional, and data came from classroom-surveys within Stockholm municipality of ninth grade students in 2014 (n=4904). Linear regression analyse was applied. Results showed that average levels of PHC varied across school segregation profiles for girls, while PTS varied for both gender. PTS was negatively associated with PHC for all students, while the strength of association varied across school profiles to the benefit of students in the most privileged schools. Gender differences in these associations was also observed. Conclusions were that school context, based on the student composition of the school, and its provided support was linked to psychosomatic health complaints among students in Stockholm and that gender played a role in understanding pathways in these associations.

Although the theory of social disorganisation does emphasise the influences neighbourhood dynamics have on youth, it does not consider if the risk factors of mobility, delinquent friends, ethnicity, and low socioeconomic factors, affect boys and girls in similar ways. The purpose and aim of this study is to contribute with gender-based research to the field of criminology that can be used when developing interventions among juveniles with delinquent behaviour. Delinquent behaviour was in the study delimited to drug use. The study used bivariate analysis, logistic regression and hierarchy regression as the method of analyses. The results showed that there are significant similarities and differences between girls and boys. It was possible to see significant similarities between boys and girls in relation to delinquent friends and drug use. Mobility was an important variable for girls in relation to drug use, but not for boys. Mobility affected girls in the opposite direction of what was expected according to theory of social disorganisation: low amount of mobility increased the risk for drug use significantly among girls.