Visualising Deprivation Change
Problem
As part of the analysis of the Indices of Multiple Deprivation 2007 (IMD2007), the Research and Information wanted to explore how deprivation has changed in Leicestershire LAA Priority Neighbourhoods. The IMD data in itself is fairly complex, dealing with ranks rather than absolute values in addition to the number of data points involved:
- 19 priority neighbourhoods (made up of 66 LSOAs)
- Ten categories for:
- Overall Deprivation
- 7 Domains
- 2 Sub-Domains
- Average district ranks for all categories
- Change between 2004 and 2007 for all categories
As a result, we had to develop a graphic which would visualise 40 pieces of information for each of 66 LSOAs in a relatively small space in such a way that comparisons could be made within each priority neighbourhood.
In total, 2,640 pieces of information had to be visualised systematically.
Solution
The end product was a graphic for each priority neighbourhood LSOA displaying the IMD ranks for 2004 and 2007 (by decile) along with change between the two years and district average rank. The graphics were grouped according to priority neighbourhood:
The grey squares refer to individual county rank by decile, with the red arrow showing the decile change between 2004 and 2007 (the longer the arrow, the greater the change). The small circles represent the average rank for each category for all LSOAs in the district.
By stacking the categories and by placing the graphics side by side as they are in the report, patterns within LSOAs and across priority neighbourhoods can be better understood.
How we did it?
The difficulty in displaying the data was based around the fact that there is so much variation in the data, with some areas experiencing very small changes with other experiencing significant change between 2004 and 2007. In addition to this, the number of LSOAs in each priority neighbourhood varied considerably (Enderby Centre is made up of one LSOA whilst Loughborough East consists of nine). As a result, we decided not to aggregate the data in any way (either statistically or spatially), instead to display it by LSOA.
Through trial an error we decided that the best way to display the data itself was through deciles rather than the rank itself, as the scale would be too big (there are 396 LSOAs in Leicestershire. Given the space we had to work with, small changes on a scale like this would not show up. In addition, policy officers were only interested in significant change, which supported the idea of using deciles.
The graphics themselves were developed from horizontal bar charts, removing as much ‘chart junk’ as we could whilst still retaining the important information and in turn, layering additional information on top using contrasting symbols and colours. The result is a cross between a dot plot and a bar chart.
How you can do it?
The charts were produced in Excel and created by hand rather than using conditional formatting and functions. Whilst this was a very long-winded and painstaking way to create the graphics, at the time it was the only way given the skills available at the time. There is no reason why these charts couldn’t be created using the functions behind a program like Microcharts.
Evaluation
Whilst not perfect, the graphics attempt to visualise a considerable amount of information in a very small space in order to draw attention to key changes. Although initially complicated due to the counter-intuitive nature of the way IMD is ranked (1st is usually considered to be ‘good’ but in the case of the IMD 1st is ‘most deprived’ and therefore ‘bad’), the reader can make sense of the charts and begin to understand the data and patterns with simple explanation.
The way the data is organised and displayed allows for straightforward comparison between categories and across LSOAs. In this respect it has been very useful for those working in priority neighbourhoods in order to understand what is a complicated dataset. Whilst not the most striking visualisation, it fulfills its purpose in taking a dataset and making it understandable for a non-technical audience.
The final report contains the complete range of analysis as well as background information on Priority Neighbourhoods.