Escaping the pushpin paradigm in geographic information science: (re)presenting national crime data
Alex Singleton; Chris Brunsdon (2014). Area, 46(3), 294-304. DOI: 10.1111/area.12116
Abstract
In 2011 the Home Office released the police.uk website, which provided a high‐resolution map of recent crime data for the national extents of England, Wales and Northern Ireland. Through this free service, crimes were represented as points plotted on top of a Google map, visible down to a street level of resolution. However, in order to maintain confidentiality and to comply with data disclosure legislation, individual‐level crimes were aggregated into points that represented clusters of events that were located over a series of streets. However, with aggregation the representation of crimes as points becomes problematic, engendering spurious precision over where crimes occurred. Given obvious public sensitivity to such information, there are social imperatives for appropriate representation of crime data, and as such, in this paper we present a method of translating the ‘point’ crime events into a new representational form that is tied to street network geography; presenting these results in an alternate national crime mapping portal http://www.policestreets.co.uk.
Extended Summary
This research critiques the cartographic representation of crime data on the UK government’s police.uk website and proposes an alternative visualisation method that better reflects the spatial accuracy of the underlying data. The study addresses how the official crime mapping portal creates misleading impressions of precision by representing aggregated crime clusters as individual points on street-level maps. To comply with data disclosure legislation, individual crimes are grouped into points representing clusters of events across multiple streets, yet these appear as precise locations to users. This creates what the research terms ‘spurious precision’ - giving the false impression that crimes occurred at exact locations when they actually represent broader neighbourhood areas. The methodology involved reverse-engineering the aggregation strategy used by police.uk, which employs Thiessen polygons around street centroids to create disclosure zones. Where zones contained fewer than eight addresses, they were merged with adjacent areas to prevent individual identification. The research reconstructed this zonal geography using crime centroid locations and developed an alternative representation using street network topology. Rather than showing crimes as points, the alternative system displays crime rates along actual street segments using either line width scaling or colour intensity to indicate crime density. This approach was implemented in a demonstration website (policestreets.co.uk) built using open-source mapping technologies including PostGIS, Mapnik, and OpenLayers. The alternative visualisation presents two cartographic options: streets displayed with varying line widths or colour intensities to represent crime rates per kilometre. The research argues this street-based representation is more appropriate because it matches the geographic resolution of the source data whilst avoiding false precision about crime locations. The broader significance extends beyond technical cartographic concerns to public policy implications. Crime mapping influences public fear of crime and community perceptions of safety, making accurate representation crucial for addressing the ‘reassurance gap’ between actual crime risk and public anxiety. The work demonstrates how open-source geospatial technologies can create more transparent and appropriate visualisations of sensitive government data. The research contributes to geographic information science by highlighting the importance of matching cartographic representation to data quality and spatial accuracy, whilst addressing wider debates about transparency in government data dissemination and the democratisation of mapping through neogeography.
Key Findings
- The police.uk website creates spurious precision by representing aggregated crime clusters as individual points on street-level maps
- Crime data aggregation uses Thiessen polygons around street centroids with zones merged when containing fewer than eight addresses
- Street network visualisation using line width or colour intensity provides more appropriate representation than point mapping
- Alternative cartographic approaches can better match data accuracy whilst maintaining required disclosure controls for sensitive information
- Open-source mapping technologies enable creation of more transparent crime mapping portals that avoid misleading spatial precision
Citation
@article{singleton2014escaping,
author = {Alex Singleton; Chris Brunsdon},
title = {Escaping the pushpin paradigm in geographic information science: (re)presenting national crime data},
journal = {Area},
year = {2014},
volume = {46(3)},
pages = {294-304},
doi = {10.1111/area.12116}
}