We have applied the analytical methodology developed in our 2010 report 3D Poverty to data from the Understanding Society longitudinal survey, in order to segment the low-income population into the dominant groupings or ‘types’ of poverty. This was achieved in four main phases.
1. Selection of the poverty indicators
In the first stage of the project, we held a series of stakeholder workshops, where we sought the views of poverty experts and those working ‘at the coalface’ of tackling poverty and deprivation.
We asked what they considered to be the key indicators that should be used to measure poverty – looking broadly at risk factors, symptoms and essential features that underpin definitions of poverty. In addition, we revisited the results of our polling in 3D Poverty, and looked at existing academic models. The list of indicators that we eventually selected married these three perspectives.
2. Analysis of the Understanding Society dataset
We wanted to understand how this list of indicators overlapped and interacted at the level of individual households.
To do this, we used Understanding Society, a comprehensive panel survey of 40,000 British households, which replaces the British Household Panel Survey. Using a latent class analysis, we examined where different sets of indicators frequently occur together, or ‘cluster’ into distinct groups. These formed our poverty ‘types’ – each type is characterised by a combination of the 20 different indicators.
3. Follow-up interviews
Once we had identified the poverty types in the data, the next step was to verify this, and explore some of the dynamics within each type by asking people about their real-life experiences of poverty.
NatCen carried out a series of thirty 2-hour long interviews with families experiencing each of our 5 child poverty types, and the findings have been incorporated into our report and our recommendations.
4. Testing the local response to poverty types
In order to test how our poverty types could be used on the ground to tackle poverty and disadvantage, we worked with three local authorities. We began by selecting two of our national types which seemed to fit best with the local situation.
To do this, we gathered, with the help of the local authorities, all of the data available relating to the 20 indicators in our poverty model, and identified the indicators which seemed particularly problematic for the area (e.g. much higher than average sickness benefit claims; much higher than average overcrowding; etc). We then compared these flagged indicators with our poverty groups and isolated where there was the largest overlap. For example:
Local authority A has:
- Small households with single parents
- Problems of social isolation
- High rates of mental health problems
- Mixed skill rates and mixed labour market
It is likely, then, that Local authority A will have managing mothers in their local population.
Of course, the limitations of this approach means we can only state that Local Authority A is likely to have managing mothers, and we do not know if they predominantly have this type or if they have a type which is wholly different to all of the national average types we have identified.
However, we then tested this process by asking the local authorities and a range of local agencies and service providers whether the groups we thought would be common locally (based on local data) ‘sounded like’ the types of families they encountered and helped on a day to day basis.
When then refined further these local types with these stakeholders’ insights before using them to develop local toolkits – suggesting ways in which such groups might be identified, targeted and helped with existing local resources and joint working, based on the model outlined above.
Demos has been researching poverty measurement since 2009, when we embarked on research for our 3D Poverty report, published in 2010.
This first wave of research set the groundwork for a broader approach to poverty measurement and presented the methodology for an annual, multi-dimensional analysis of poverty and social exclusion. Our research included polling and scoping workshops with members of the public to better understand public attitudes to poverty measurement, as well as a detailed review of existing measures.
3D Poverty had several key findings. Though various academic models for measuring multi-dimensional poverty exist, they have various drawbacks – firstly, they are targeted at an academic audience, and thus are not designed to be used. They also fail to communicate a meaningful vision of what poverty actually looks like.
Secondly, despite covering a range of indicators, the existing models do not allow an examination of the overlap and incidence of different dimensions of poverty. Where this has been attempted (e.g. by the Social Exclusion Taskforce at the Cabinet Office), it has not been possible to drill down to the particular sets of characteristics of individuals and households, and so has limited use for developing multi-dimensional poverty responses.
Our survey of public views on poverty and how it is measured found that there is a lack of public understanding of the scale of income poverty, and of how it impacts on the daily lives of those experiencing it. What did emerge was that poverty could never be fully described or explained by referring to income alone.
In polling commissioned by Demos for the report, we asked which indicators were most important for measuring poverty. Income was the top response (69 per cent of people polled chose this option). However, in a separate question, more people disagreed (48 per cent) than agreed (30 per cent) that it is adequate to measure poverty solely by assessing household income alone.
In scoping workshops carried out by Demos with members of the public, distribution of income through the benefits system was also the source of the most resentment and negativity towards people in poverty. Focusing on income alone can thus be a barrier to building greater public support for tackling poverty.
These are the flaws that we have sought to address in this second wave of the research, culminating in Poverty in Perspective.
Scroll over the icons below for a definition of the twenty indicators we used to identify the types of poverty.
About the indicators
All the poverty groups have a household income below 70% of median income. Within this, they have been subdivided into five income bands.
Lacking material goods
This captures the extent to which families lack typical consumer durables such as a TV or washing machine.
This is the proportion of working age household members who are not in employment (either full or part-time).
Behind on bills
This captures whether households have been behind on one or more bill payments in the last 12 months.
This reflects whether households contain too few rooms for the number of occupants.
Lone adult households
Single person households vs. households with multiple adults. (The presence of more than one adult in a household with children does not necessarily mean that this is a two-parent household.)
We have chosen to code lone adult households as negative and presence of multiple adults in a household as positive, thinking of multiple adults as an asset (as it increases the potential for household income). However, lone adult or lone parent status is not necessarily a disadvantage.
This indicates whether households report low levels of trust in strangers (used as a proxy for community participation).
Disinterest in politics
Extent to which people describe themselves as interested in politics.
Carer for a child
Whether anyone in the household is caring for a child with a health condition or disability.
This distinguishes between those who rent and those who own their homes. We have chosen to code renting as negative and home ownership as positive, thinking of ownership as an asset. However, ownership is also associated with additional costs, e.g. mortgage payments and maintenance, which may contribute to poverty.
Limited car access
This is based on the ratio of cars to adults in a household. We have chosen to code lack of access to a car as negative and car ownership as positive, thinking of ownership as an asset. However, car ownership is associated with additional costs that may contribute to poverty.
This reflects the educational attainment of the head of the household, from having no qualifications to having a degree.
Physical ill health
This is a self-reported measure of physical health for all adults in the household.
Mental ill health
This is a self-reported measure of mental health for all adults in the household.
Lack of neighbourhood support
This indicates whether respondents report feeling poorly supported by their neighbours and neighbourhood.
Lack of family support
This indicates whether respondents report not living near to, or having only infrequent contact with, family.
This shows the extent to which respondents feel they are struggling financially.
This captures whether people are struggling to afford to heat their home.
This illustrates how deprived the household’s neighbourhood is on a national ranking.
This shows the extent to which households cannot afford common services and leisure experiences such as replacing worn out household items, or meals out.
Scroll over the graphic to find out about the different indicators