In this blog, Tej Nathwani introduces a new area-based measure of deprivation created by the Higher Education Statistics Agency (HESA). He sets out how and why this was formed, alongside the possible uses of the variable to those who utilise the UK Data Service.
Introduction
One of the criticisms often made of area-based measures of deprivation is that they may not reliably inform us of the circumstances faced by individuals/households within a vicinity. For example, it is possible that a household living in a deprived neighbourhood could have high levels of income. Consequently, using area-based measures for policy-making may lead to the implementation of proposals that are ineffectively targeted, with support provided to those who may not require it, while those who could have benefitted miss out. However, as we have attempted to demonstrate through our latest research, it appears place may matter, with the locality one lives in potentially inhibiting individuals from fulfilling their abilities.
Yet, with existing area-based indicators having their drawbacks, we decided to see if we could generate a measure that may help address some of these limitations. Before we go into the details of our work, we begin with a brief overview of how we define deprivation and why it remains a pertinent topic.
What do we mean by deprivation and why does it matter?
[The concept of deprivation] is used extensively not only in the analysis of social conditions but also, in an applied form, as an instrument of policy in allocating resources to particular regions, areas and services.
From Deprivation by Peter Townsend in 1987
Our understanding of deprivation today is predominantly thanks to the work of a sociologist called Peter Townsend. Back in the late 1980s, he published a paper in which he outlined that deprivation was relative and that an individual/household was likely to be experiencing this if they were unable to fully participate in what had become seen as customary activities within society.
At present, the need to tackle long-standing economic and social inequalities between/within regions of the UK has led to this issue being at the forefront of objectives set by governments across all four nations.
Of course, devising effective interventions requires data that can enable researchers and policymakers to accurately identify those areas experiencing higher levels of deprivation. As Townsend himself acknowledged, this was not a straight forward task.
How do we capture deprivation through data?
Currently, the primary data source used to examine this (Households Below Average Income) is based on the Family Resources Survey in which respondents will be asked questions relating to (material) deprivation, such as whether they can afford to go on holiday once a year or repair broken goods within their home. The main problem with this is that a survey will not offer sufficient coverage of all parts of the UK.
That leaves researchers with two options to proxy for deprivation – use a survey that must be completed by virtually the entire UK population (i.e. the Census) and/or rely upon administrative data sources.
Townsend’s approach around forty years ago was to build a measure using the Census, given this contained a number of variables that were likely to be correlated with deprivation, such as housing tenure and job market outcomes. He chose to use four fields. These specifically related to unemployment, overcrowding, household tenure and car ownership. The Carstairs index surfaced at a similar time and drew upon comparable fields from the Census, with the use of social class instead of home ownership being the main distinction.
In recent decades, we have seen administrative datasets on matters such as health, income and employment gradually emerge. As they are also expected to be associated with deprivation, they are increasingly used in this space. Indeed, it is partly such sources of information on which the Indices of Deprivation have been created – the most commonly used area-based measure in all four nations.
Given we have numerous area-based measures already, why do we need another one?
Over time, the inclusion of car ownership in both the Townsend and Carstairs index has been questioned. The reason behind this is that it may well be that those from rural communities may have no choice but to purchase a car if they live in a location with limited public transport. Hence, despite these measures having been created on a UK-wide basis, they may lack the ability to capture deprivation effectively in rural spots.
The Indices of Deprivation also have their own limitations. Firstly, they are nation-specific and therefore do not enable researchers to compare statistics across countries or provide a UK-wide picture. Indeed, the lack of a suitable UK-wide deprivation metric has been raised as a concern by the Office for Statistics Regulation and the Social Mobility Commission, given that it inhibits the possibility to answer questions on social mobility. Furthermore, the size of the areas used to form the Indices can make it difficult to identify pockets of deprivation in otherwise less deprived neighbourhoods. The final index (typically used for research and policy-making) is also known to be less effective in picking up deprivation in rural areas.
Consequently, there remains a need for relevant data in this field.
How did we form and validate our measure?
Like Townsend, we relied on the Census, given the harmonisation (as far as possible) of questions across the UK opens the opportunity to create a UK-wide measure. As this work began a few years ago, it has initially been based on the 2011 Census, though we will be updating this with 2021 data once this is available for all four nations. The other major advantage of using the Census is that data is released at a geographic domain, which tends to contain less than 500 individuals. These are much smaller than the size of the areas used to form the Indices of Deprivation and thus can help to mitigate the risk of not being able to identify more deprived neighbourhoods in otherwise less deprived areas.
Our literature review highlighted that education and occupation were key determinants of whether households experienced persistently low income. Though poverty and deprivation are slightly different concepts, it is recognised that those with little financial resource have a higher likelihood of experiencing deprivation. Consequently, our measure is derived based on these two Census variables.
We then validated our measure by assessing its correlation with household income estimates (which take into account housing costs and family size), as well as its association with housing tenure, family composition and self-reported health data from the Census. The rationale behind evaluating its link to these three variables is that previous research has noted that those living in social housing, single-parent families and/or in poor health are more likely to experience material deprivation.
Our results were as expected. That is, those areas we identified as being most deprived had lower levels of income and higher proportions of households living in social housing or reporting that they were single-parent families. Residents in these areas also had a higher probability of stating that they were in bad/very bad health.
Additionally, given the issues noted above with existing area-based indicators in terms of picking up deprivation in rural areas, we also examined whether our measure captured a greater proportion of such localities in the bottom quintile. We found this to be the case.
How might this measure benefit UK data service users?
If you are a researcher interested in utilising one of the UK-wide studies available from the UK Data Service (such as the Millennium Cohort Study or Understanding Society), there is currently no appropriate UK-wide area-based measure of deprivation available to you to the best of our knowledge. Instead, you may have to rely solely on an individual-measure such as parental occupation or household earnings for a UK-wide analysis, despite the possibility that areas have a role in whether or not individuals reach their potential. Additionally, for those carrying out country-specific research, you may be looking for additional/alternative variables to the Indices of Deprivation as part of your study.
We therefore welcome enquiries from data depositors on how to include our measure in their datasets, as well as from researchers who may be interested in using this variable in their analysis (e.g. by linking it to an existing data source available through the UK Data Service). Please send your comments and queries to pressoffice@hesa.ac.uk.
About the author
Tej is a researcher at the Higher Education Statistics Agency (now part of Jisc). Since joining in 2016, he has led the establishment of a research function within the organisation. The purpose of this activity is to generate outputs that support key stakeholders (e.g. students, employers and policymakers) in their decision making, as well as to ensure official statistics on higher education remain relevant and valuable to data users.