Using Census data to derive a new area-based measure of deprivation - Section 7: Discussion

Updated 19 March 2024

Section 7: Discussion and concluding remarks

Our aim in this paper was to firstly outline the creation of a new UK-wide area-based measure of deprivation using Census 2011. Having done so, we then assessed its correlation with factors associated with deprivation and the extent to which it differs to the Indices of Deprivation. Overall, the findings provide evidence in favour of our measure being related to deprivation as expected and also offering added value to users of deprivation statistics through overcoming some of the known limitations of the Indices of Deprivation. Furthermore, given it is UK-wide, our measure can also enable comparable statistics to be developed across nations – a mark of quality as highlighted in the UK Code of Practice for Statistics.

In England, Wales and Scotland, our measure captures a greater proportion of rural areas in the bottom quintile when compared with a multiple deprivation index. Within Scotland, all council areas emerge in the bottom quintile of SEISA, which is not the case for SIMD, where some known areas of deprivation are not found in the bottom quintile. For England, the lowest quintile of the variable we have created also has the advantage (over the same quintile of IMD) of picking up medium/large towns in northern and central England. As in Scotland, the bottom quintile of our measure covers all local authorities/LGDs in Wales and Northern Ireland, respectively. In Wales, one of the key differences between our measure and the bottom quintile of WIMD is the greater ability of our variable to catch deprivation across two local authorities in the south. Meanwhile, for Northern Ireland, the benefit of our measure relative to NIMDM (based on our assessment of the lowest quintile) appears to lie in picking up parts of the north and/or east of the country.

Although taking part in the Census is obligatory, it should be acknowledged that the data is still self-reported and could therefore be subject to measurement error, as individuals may be unwilling or unable to give an accurate response to questions on their qualifications and occupation. Indeed, the ONS (2014) completed a Census Quality Survey between May and August 2011, in which volunteers were requested to participate in a face-to-face interview. Those who agreed were asked the same questions that they responded to approximately two to five months earlier, with the key difference being the mode of survey completion. Agreement rates were generally found to be lower (around two-thirds) for the questions on occupation and education, which tended to be for reasons such as respondents giving different job titles or struggling to recall the qualifications they had attained. Additionally, in this paper, our evaluation of the degree to which our measure is correlated with income relies upon estimates at a relatively aggregated level of geography. We recognise that within such areas, there is likely to be variation in the incomes of households. In the case of Scotland and Northern Ireland, we had no relevant data that we could draw upon.

The 2021 Census has now been administered across all nations of the UK. Once this data is released into the public domain, we will seek to produce an updated version of the measure we have presented here.

Next: References