Post by Chris Brunsdon
I’ve been involved recently in looking at ways in which geographical data (and particularly Human Geographical data) has been represented on the web. I think it is important to look at any data analysis or visualisation critically – are there any factors which have been overlooked in the collection of the data used, or in the way it has been represented graphically, or modelled statistically, that might lead to misinterpretation, for example. In the UK and a number of other countries, there have been recent moves towards the provision of open data – where data collected by governments is made freely available over the internet – here is an example. Also, some international organisations provide this data. Many providers of web sites use this data, linked to maps and other visualisations, to provide easy to access overviews of patterns in this data. Here, I will show two such sites – these are all interesting, although there are reasons generally why the critical approach I advocated earlier is valuable…
The first is the police.uk web site – showing details of recent crime in an area. I think it is a great advance in making data publicly available, but one major problem is the way data is represented if you ‘zoom in’ on the map. Here is a map showing crimes near to Liverpool Geography. If you zoom in to the area very close to the marker balloon, it shows the exact location in streets where offences have been committed – for example 3 in the middle of Back Mulberry Street, if you are looking at the map for August 2012. But does it really show this?
Quite likely not. A small caption underneath the map reads “Crime and outcomes are mapped to an anonymous point on or near the street or location where they occurred.” In fact, to preserve some degree of anonymity small groups of houses or other buildings are sometimes ‘clumped together’ and a point chosen in the middle of the clump. This is quite likely what happened in the example above. This can make it appear as though a mini crime wave is occurring in front of a particular house, or on a street corner, when this may well not be the case. This isn’t the first time the issue has been raised. I’m not sure that a small disclaimer underneath the map entirely compensates for the striking but ultimately misleading visual impact of the map.
The second example is less problematic – see here. This web site gives population pyramids for countries across the world at regular time intervals from 1950 to 2100. It is quite illuminating to compare, say, current pyramids from the UK and Nigeria. If you are not familiar with population pyramids as a model of population structure, the site provides a link explaining this, and suggesting interpretations for common shapes. The site also shows total population growth from 1950 to 2100 in graph form, as well as letting you select pyramids from different points in time. It is also interesting, for example, to compare the predicted population structures of Nigeria and the UK in 2050. Perhaps my one issue with this site is the ease from which you move from actual past figures to predicted future figures – since the latter are based on models they will be subject to a fair amount of uncertainty. Perhaps the graphics could give some kind of visual clue about this.
In both cases, I think it is excellent that the information has been released – and the organisations responsible for this deserve praise. However, I would always advocate looking at any graphic with a critical eye.