What Deprivation Ranks Can’t Tell Us
With the recent release on the 2014 Welsh Index of Multiple Deprivation, Wavehill’s Steven Donbavand gives us his thoughts on the pitfalls and perils of using deprivation ranks.
Deprivation ranks are used across the public sector as a quick and intuitive performance measure – a sort of socioeconomic health check – for every neighbourhood in Wales. The ranks give each ‘lower super output area’, to use the precise jargon of the Office of National Statistics, a simple number whereby ‘1’ indicates the most deprived area (‘St. James 3’ in Caerphilly) and ‘1909’ the least (‘St. Kingsmark 1’ in Monmouthshire).
But although deprivation ranks seem straightforward, they can easily be misinterpreted. For example, it is actually possible for deprivation in an area to steadily improve over the years but for the deprivation rank to fall. Equally it is possible for an area to move to a higher rank even if the situation on the ground is deteriorating! This is because ranks are relative in nature – it is not only the performance of your area which determines its deprivation rank, but also how all the other areas in Wales are doing.
Another difficulty comes with interpreting the magnitude of change. We might imagine that a neighbourhood that has improved its rank by 50 places has seen a bigger improvement than one which has only made a 10 place increase, but this may not be the case. The change in the deprivation score between ranks 1 and 2 is not necessarily the same as the change between ranks 1908 and 1909. The interpretation of moving through the ranks can change depending on which part of the deprivation continuum we’re referring to.
Not only are ranks more difficult to interpret than might be thought, but they are also tricky to work with when undertaking analysis. For example, if a Local Authority wants to know its overall level of deprivation, the obvious approach would be to calculate the average rank across all of its neighbourhoods (i.e. the arithmetic mean of the deprivation ranks). But this approach is actually incorrect from a statistical perspective. A good way to think about this is to see ranks as a type of simplified summary measure which cannot then be further summarised into an average. Instead, Statistics in Wales recommend that Local Authorities look only at ‘quantiles’, such as how many of its neighbourhoods fall in the bottom 10% nationally.
The difficulty in correctly interpreting the Index of Multiple Deprivation actually goes much further than the use of ranks. The index is composed of eight deprivation ‘domains’:
Access to Services
Each one of these domains is given a weight which determines its importance in producing the overall deprivation score. It’s an interesting thought experiment to consider how you would personally rate each one of these domains, and then compare it to the weights actually used. For example, the Index considers Access to Services to be twice as important as Housing, and the combined importance of Health and the Physical Environment are judged to be less important than Employment. We’re all likely to have our own opinion on these trade-offs, but it becomes clear that these are value-judgements. If we were to reconstruct the Index on the basis of alternative value judgements the deprivation ranks would change.
Ultimately, it’s important to remember that whilst deprivation ranks are a quick way to get an impression of how an area is doing, the extent to which they can offer deep insight is limited. To really know what progress an area is making it is necessary to go back to the simpler indicators which underlie these composite measures. This allows us to speak in absoluterather than relative terms and to build a nuanced picture of what is happening in a particular area; understanding how different issues interact with each other and accepting that some indicators may be more relevant in some areas than in others.
The 2014 Welsh Index of Multiple Deprivation can be accessed here:
If you would like to discuss any of the issues raised in the article, please contact Steven at email@example.com