Thursday, 11 August 2011

The dishonest Indices of Deprivation map doing the rounds

There is a map doing the rounds which is said to depict level of deprivation and areas where riots took place. This map has seemingly impressed many people and has been tweeted and re-tweeted all over Twitter. It can be seen here.

As someone who has spent a number of years working in statistics and research at the Office for National Statistics, I'm disappointed in this map. I'm disappointed because it represents either a deliberate attempt to mislead, or an ignorance of how to fairly and properly display data. The reason that this map is misleading is because it depicts deprivation for London only. People looking at this could be forgiven for looking upon the map as incontrovertible evidence that deprivation leads to the kind of behaviour that occurred. However, if the data had been displayed for the rest of England, it would be quite apparent that the association between deprivation and rioting is actually quite weak. Failing to map the data in its entirety removes the need for an explanation of why a few deprived areas got involved in the riots, but most deprived areas did not. 

I can only speculate on why someone would go to the effort to produce their own map when a perfectly useful one is made available by the Department for Communities and Local Government, who produce the Indices of Deprivation. This map can be seen below and can be found, along with methodological information in this document.

This map quite clearly depicts high levels of deprivation in many regions of the country. Most of these were not affected by rioting. For example, there was no rioting in the North East. I would recommend looking at the actual data, as it carries a much greater level of detail than a map. For example, from the map above it isn't apparent that the most deprived Lower Layer Super Output Area (LSOA) in England is in Tendring, Essex, or that no LSOA in London is within the most deprived 400 in the country. I talk about this data more in a previous blog entry.

Ultimately, excluding items from analysis because they don't show what you want to see is not just bad practice, but it is dishonest. Statistics such as these are created at considerable expense with the purpose of informing policy, not for people to bastardise them in order to support their pre-existing opinions and beliefs. At the very best, the map in question indicates that there were riots in some deprived areas in London. That's it - that's all it does. It doesn't prove that deprivation was an important causal factor. The data as a whole indicates that it isn't actually an important factor at all. This is consistent with what we know about people who aren't financially wealthy - most of them are decent, moral people who wouldn't commit such criminal and immoral acts.

No comments:

Post a Comment