by Ros Edwards, Sarah Gorin and Val Gillies
The National Data Strategy encourages the UKās central and local government to team up with the private sector to digitally share and join up records to inform and improve services. One example of this is the area of troublesome families, where itās thought that the use of merged records and algorithms can help spot or pre-empt issues by intervening early. But there are questions over this approach and this is something our project has been looking into. In our first published journal article, we have been examining the rationales presented by the parties behind data analytics used in this context to see if they really do present solutions. Ā
The application of algorithmic tools is a form of technological solution; based on indicators in the routinely collected data, in an effort to draw out profiles, patterns and predictions that enable services to target and fix troublesome families. But local authorities often need to turn to commercial data analytic companies to build the required digital systems and algorithms.
In our paper we analysed national and local government reports and statements, and the websites of data analytic companies, addressing data linkage and analytics in the family intervention field. We looked in particular at rationales for and against data integration and analytics. We use a āproblem-solvingā analytic approach, which focuses on how issues are produced as particular sorts of problems that demand certain sorts of solutions to fix them. This helps us to identify a double-faceted chain of problems and solutions.
Seeking and targeting families
Families in need of intervention and costing public money are identified as a social problem and local authorities given the responsibility of fixing that problem. Local authorities need to seek out and target these families for intervention. And it is experts in data analytics that, in turn, will solve that identification problem for them. In turn companies are reliant on citizens being turned into data (datafied) by local authorities and other public services.
We identified three main sorts of rationales in the data analytic companies promotion of their products that will solve local authoritiesā problems: the power of superior knowledge, harnessing time, and economic efficiency.
Companies promote their automated data analytics products as powerful and transformational. They hand control of superior, objective and accurate, knowledge to local authorities so that they can use profiling criteria to identify families where there are hidden risks, for intervention. And their systems help local authority services such as social care and education collaborate with other services like health and the police, through data sharing and integration.
Data analytics is presented as harnessing time in the service of local authorities as an early warning system that enables them quickly to identify families as problems arise. It is the provision of an holistic view based on existing past records that local authorities hold about families, and the inputting of āreal timeā present administrative data on families as it comes in. In turn, this provides foresight, helping local authorities into the future ā predicting which families are likely to become risks in advance and acting to pre-empt this, planning ahead using accurate information.
Another key selling point for data analytics companies is that their products allow economic efficiency. Local authorities will know how much families cost them, and can make assured decisions about where to put or withdraw resources of finances and staffing. Data analytic products produce data trails that cater for local authorities to prepare Government returns and respond to future central Government payment-by-results initiatives, maximising the income that can be secured for their constrained budgets.
Questions to be asked
But there are questions to be asked about whether or not data linkage and analytics does provide powerful and efficient solutions, which we consider in our article. Concerns have been raised about the errors and bias in administrative records, resulting in unfair targeting of certain families.
Particular groups of parents and families are disproportionately represented in social security, social care and criminal justice systems, leading to existing social divisions of class, race and gender built into the data sets. For example, there is evidence that racial and gender profiling discriminations are built into the data, such as the inclusion of young Black men who have never been in trouble in the Metropolitan Police Gangs Matrix. And automated modelling equates socio-economic disadvantage with risk of child maltreatment, meaning that families are more likely to be identified for early intervention just because they are poor. On top of that, studies drawing on longitudinal data are showing that the success rates of predictive systems are worryingly low.
All of which raise a more fundamental question of whether or not algorithms should be built and implemented for services that intervene in familiesā lives. In the next stage of our research, we will be asking parents about their views on this and on the way that information about families is collected and used by policy-makers and service providers.
- Problem-solving for problem-solving: data analytics to identify families for service intervention by Ros Edwards, Val Gillies and Sarah Gorin has been published in Critical Social Policy, 2021, 1-20 .