Geography Reference
In-Depth Information
violence and murder) rather than ''discovering the unexpected'' (though the neg-
ative relationship of vehicular crime with violence and murder may be worth
further investigation; also the possibly linked positive relationship of vehicular
crime with lack of policing).
These vignettes are important in the communication of discovered information
to the public. It is a key stage in Thomas and Cook's (
2005
) framework for visual
analytics: to ''communicate assessment effectively for action''. This element is
made manifest in eXplorer as storyboards that would provide the explanation
accompanying a specific linked representation of data.
5 Conclusions
WikiCrimes is a Brazil-based Web 2.0 resource for volunteered crime information.
As such, it has a sizeable point-based resource that is ripe for exploration using
geovisual analytics tools, a dataset rich with attributes such as type of crime, that
simply would not have existed anywhere in the world ten years ago. eXplorer has
been used to good effect for this purpose on the WikiCrimes data aggregated to
Brazilian state level and to degree grid level, mostly identifying expected relation-
ships but also uncovering unexpected ones that may be worthy of further analysis.
The use of eXplorer in this way is sympathetic to WikiCrimes' aims and as such
is a valuable supporting tool. It could promote enhanced crime transparency and
generate added publicity for these all too hidden events. It does so by putting the
data in a more digestible form for public and law enforcement officers alike.
However, by aggregating from conventional pinmap and density hotspot repre-
sentations to a choropleth one, errors due to ecological fallacy and the MAUP
abound, though the effects of these well-known sources of error are mitigated
through the grid-based representation.
The crime reporting representations that WikiCrimes uses, straightforward
pointwise display (and density kernels) can give insight on distribution (it is the
intention that such information will have value in the hands of the public) but
misses much information of value. Nevertheless the grid-based implementation
retains most of the distribution information lost in the state-based implementation,
at an appropriate scale (degree grid) to Brazil's overall size.
The state-based implementation still has value in reporting data in special units
that have a history and resonance with the intended audience, though the mis-
leading over-aggregation of the data will outweigh this benefit. Furthermore,
representation using recognized political or organisational boundaries facilitates
future analysis with census data, which is reported via such aggregation units.
To facilitate these anticipated benefits, the tool will be in the public domain,
where it will hopefully become a focus for public collaboration. However, there
are issues that any user of the tool should be mindful of. The database builds
gradually, and while every effort has been made to anticipate crime types, settings
and reasons, inevitably a category that was created to fit a crime entry early on
