Banner showing England map

Exploring the index of deprivation that ranks all local authorities in England using multiple deprivation factors including income, employment, health, education, crime and living environment. 


For some practice with mapping statistics, I decided to plot the Index of Multiple Deprivation (IMD) data for England. The IMD has been produced five times since 2000 – 2000, 2004, 2007, 2010 and 2015. During that time, we hit a global recession that kicked off late 2007 and the UK has had two changes of government – from Labour to a Conservative-Liberal coalition in 2010 and then a Conservative majority in 2015.

A popular news headline is that the Conservatives invest in growing London at the expense of the rest of the country and that the North-South divide has grown due to the recession with the North not being the benefactor. Soooo….. does the IMD data support the opinion?

The Data

First up, an explanation of the data. Three measures are included here: rank of average score, rank of income scale and rank of employment scale. For income and employment, the scale is the number of people in a local authority who are income deprived or employment deprived. Therefore the ranking is going to be skewed against larger authorities. Ideally, the data should be normalised by population count in each local authority but that data isn’t provided. The average score is population weighted for larger areas but the method for weighting isn’t included. The score itself is a weighted combination of: income deprivation; employment deprivation; education, skills and training deprivation; health deprivation and disability; crime; barriers to housing and services; and living environment deprivation. The data for each year the IMD is published is likely to be approximately two years old. For example, the 2015 IMD is based on statistics for the tax year 2012/13.

It’s difficult to use this data to compare regions with each other without more supporting information such as population counts for each LA at each point in time when the IMD was calculated. However the ranks do provide a useful indication about changes to each region over time. The number of local authorities changed in 2007, reducing from 354 to 326, mostly a result of mergers.  The data for 2000 and 2004 were merged to match the current number of local authorities, the average score was calculated and used for the merged authority. Merging was determined by good ol’ Wikipedia

A tour of England

To help put the following visuals into context, the maps below show the nine regions of England and the approximate location of some cities and towns to help locate areas (click on image to view larger):

EnglandRegionsCities

The big green line is the ‘North-South’ divide. Exact placement and angle is open to interpretation. Those in the North usually think it is higher, and those in the South usually think it is lower. Those of us in the middle think the band should be wider given the North think we are southerners and the South think we are northerners.  The red spec on the big green line is where your author currently resides 🙂  The SoA stands for Stratford on Avon, not Sons of Anarchy… tho’ a modern-day Shakespeare might like the alternative. The region boundaries are visible on the visualisations below.

Changes in rank over time

Let’s take a look at the impact of the recession. The image below shows the change in rank by local authority from 2004 to 2007, 2007 to 2010 and 2010 to 2015: (Click on the image to view larger)

Comparing changes in rank for England local authorities

Comparing changes in rank between local authorities in England

It’s important to remember that all 326 local authorities (LAs) are ranked from 1 to 326, with 1 being the most deprived and 326 being the least deprived. For some to go up, others have to go down. The ranks say nothing about the distance in deprivation from 1 to 326 or the movement up or down at each level. Being the most deprived in 2004 may have been a bigger or smaller issue than being the most deprived in 2015.

The LAs in cream had a less than 1% change from one index to the next, i.e. minimal movement. The pale red and pale blue represents an increase or decrease of 1 to 10%. The dark red and dark blue represents an increase or decrease of more than 10%.  The recession clearly had an impact from 2007 to 2010 but there is no clear winner or loser, with LAs in the north and south being affected. The most recent changes are perhaps more interesting. From 2010 to 2015, the majority of gains have been in London and the South East when viewing the average score. The only two LAs to have a more than 10% increase outside of those areas are Cambridge and Manchester. The worst drops in rank occurred in the East of England.

Biggest fallers 2007 to 2015 Biggest risers 2007 to 2015
Region Local Authority Movement Region Local Authority Movement
East of England Forest Heath -62 South West Isles of Scilly +103
East of England Breckland -45 London Greenwich +50
East of England Chelmsford -39 South East Eastbourne +45
East of England Tendring -36 South East Oxford +44
West Midlands Wychavon -34 South East Wycombe +39

Geographical distribution of ranks

The second perspective is the spread of ranks across the country. Is there a pattern to the distribution of high and low ranks? If so, has the pattern changed as a result of the recession? The image below shows the distribution of ranks in 2000 when the first IMD was published, in 2007 at the beginning of the recession, and in 2015 when the debate is whether or not the recession is over.

Distribution of ranks

For this visualisation, cyan blue represents the bottom 10%, fuchsia pink represents the top 10%, and each shift in colour represents a 10% shift from blue to pink.

Well… truth be told, the visual came out a lot less interesting than I had hoped. Focusing just on the overall IMD rank, it does look like the spread of high ranks is concentrating more in the south east but this is one of those instances when the plain numbers provide a stronger picture than a more detailed analysis. The table below shows the top and bottom 5 ranked local authorities pre- and post-recession:

Bottom 5 in 2007 Bottom 5 in 2015 Top 5 in 2007 Top 5 in 2015
Region Local Authority Region Local Authority Region Local Authority Region Local Authority
North West Liverpool North West Blackpool South East Hart South East Hart
Greater London Hackney North West Knowsley South East Wokingham South East Wokingham
Greater London Tower Hamlets Yorkshire & Humber Hull South East Surrey Heath South East Chiltern
North West Manchester North West Liverpool East Midlands South Northamptonshire South East Waverley
North West Knowsley North West Manchester East of England South Cambridgeshire South East Elmbridge

And the following chart shows the make up of the top and bottom 10% by region for each year the IMD has been published:

Chart showing region groups

The northern regions are shades of blue, east and midlands are shades of green, south and London are shades of orange. The top charts show the number of local authorities grouped by region in the top and bottom 10% of ranks for each year. The bottom charts show the proportion of local authorities per region. For example, the South East has 67 regions whilst the North East has only 12. In 2015, both the North East and the Yorkshire & the Humber had  3 LAs in the bottom 10% but that represents 25% of the North East versus 14.3% of Yorkshire & the Humber.

So back to the original question. Has the past five years seen investment focused on improving London and the South East at the expense of the North? Well the initial signs are not great. Not strong either, but not great.

In reality, there is insufficient open data to work with. To really get a feel for changes in deprivation for each factor, the data needs to be normalised by population for each local authority. And that highlights a challenge with many data sets. Whilst it is fantastic that governments are committing to make much more data openly available for analysis, a lot of the data sets are incomplete. To take this study further requires sifting through other data sets to try and find some useful population figures by local authority. They must have been captured for the IMD study in order to be able to do the weighting. But the underlying data and calculations have not been shared. Without them, useful analysis is limited.

For those interested in the tools used. The data was prepped in Excel and mapped in ESRI ArcGIS. Took a break from the R vs Python debate 🙂 (tho’ R would have won for the mapping…)

References

Category:
Blog, Data Science, Visual Analytics
Tags:
,

Join the conversation! 3 Comments

  1. Although I just wrote about the problem(s) with metrics, I have been fascinated by your recent exploration/work and what I think is a very thoughtful of going about it. Not meaning it as a criticism, but just so you are aware, relatively mild color blindness makes it nearly impossible for me to perceive the gradations of blue to pink. I’m not sure what you can do with that issue in such a tight graphic, but I thought you might like to know.

  2. Thoughtful “way” of going about it 🙂

  3. Hi Dan

    Thanks about the colour tip – very useful to know as its something I try to avoid. For this one, I actually just picked one of the default sets out of ArcGIS and vainly thought it looked nice :-/ I normally use the colorblind safe sets from http://www.colorbrewer2.org

    And as always, thanks for the feedback – much appreciated!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: