The production of accurate models of crime risk rely on a number of different factors, two of which will be focused upon here. First, there is the issue of the specification of meaningful boundaries and secondly, there is the measurement of relevant environmental factors and land uses.
The size and shape of boundaries should always be considered in the production of crime models. Empirical examples will be used to demonstrate how the selection of boundaries affects patterns of risks identified. This has implications for both criminal understanding and crime policy. Problems associated with the relativity and modification of boundaries will be described.
Greater computational power and access to richer geographical data sets have enabled crime modelling to be undertaken at ever smaller units of analysis. This raises important issues concerning the analysis of background factors such as housing distribution, road layout and land use. Manipulation of these data sets at the small area will be demonstrated and issues relating to appropriate levels of aggregation addressed.
Finally, future challenges will be outlined including production of crime risk models that are multi-level, that consider links between areas and that identify natural boundaries.