I awoke last Friday to discover, from this article in the Irish Times, that I live in a “crime black spot”. My first reaction was “great someone is finally carrying out some real spatial analysis on crime in Ireland”. My second reaction was “hey, I live here – crime is not so bad, what are they talking about?”.
Well of course the headline is totally meaningless, what was actually being reported was a numerical analysis of offences per 100,000 people. How can 100,000 people constitute a “black spot”? Surely the word “spot” implies a level of spatial precision? In fact, on further analysis the “spot” they are talking about is not terribly well defined, no surprise there. It’s variously referred to as “The Garda’s south central division in Dublin” and the “areas policed by Garda stations in Donnybrook, Harcourt Terrace, Irishtown, Kevin Street, Kilmainham and Pearse Street”.
Mmmm, quite a big spot then! I’ve tried in the past to find out the exact boundaries of the Garda Divisions without success, it seems to be a contentious issue. There is an ongoing debate about aligning Garda operational areas onto more commonly used statistical reporting areas.
However just to illustrate how big this “black spot” is I plotted the locations of the Garda Stations named in the report and computed their minimum bounding polygon of “convex hull”. I then added the 2006 ED’s on which the Census is reported and symbolised population density using a dot
Crime Blackspot?
density symbol where each dot represents 20 persons. Then I intersected the convex hull with the ED’s. Guess what? This “spot” includes a total of 27 ED’s with a Population at the 2006 Census of 89,138 persons.
It would be great if we could report crime statistics spatially like in other countries. We could use that information to target real crime backs spots, we could use it to identify repeat victimisation, anti-social behaviour, vulnerable people, we cold use it to decide where to open or close Garda Stations, we could use it to make decisions about where to live.
The sad fact is, as illustrated by this example, that we can’t do any of these things because we don’t have a meaningful, stable and widely adopted spatial reporting framework for official statistics. Talk of postcodes and Small Area’s offer some hope but only if they are actually adopted and embedded into the recording systems of the agencies that record crimes and other official statistics.
check out http//:www.irishpostcodes.ie for a beta test of a geo postcoding solution which will go live shortly.
The solution that the Government proposed as a traditional postcode was due on 1st Jan 2008 (2 years ago!!!) and according to the DCENR will only have a resolution equivalent to townland level in non urban areas (where 40% of the population lives). It will never be of any use for GIS.
However, the full version of PON Codes will be able to define individual properties to within +/-6m accuracy, can be considered as a point and can be tested against any polygon necessary for analysis. Plugin tools will be available for GIS systems.
Full release very shortly
PON codes are a great advancement, however many geo-statistics need to be aggregated up to overcome data protection issues. We still need a Small Area structure or a conventional postcode for this purpose.
All aggregations or polygons are an ordered collection of Points… PON Codes are the points – pick what ever aggregates or polygons you wish to order them into – most likely ASAC’s – but change to suit the statistical test
It’s a classic case of Cart and Horse – Points and Polygons or Coordinates and GIS!