Planned intervention: On Wednesday April 3rd 05:30 UTC Zenodo will be unavailable for up to 2-10 minutes to perform a storage cluster upgrade.
Published May 1, 2019 | Version v1
Report Open

On the challenges of using administrative data from social care: experience using a new, linked whole-population dataset

  • 1. UCL Legal Epidemiology Group
  • 2. UCL Department of Statistics

Description

General background: the need to make better use of data within family justice

This short briefing note provides an overview to the full report of Children in Family Justice Data Share (CFJDS) pilot.

Recent children’s social care and family justice initiatives have attempted to exploit routinely collected data for service planning and delivery and informing or driving decision-making at the individual case management level. For example, the new Nuffield Family Justice Observatory (FJO) aims to encourage the use of administrative data to better understand the family justice system and to ‘develop, update and test guidance and other tools based on that knowledge’ (Nuffield Foundation, 2018). The use of administrative data has the potential to be transformative of children’s social care, especially given that this is an area where population-level research evidence is lacking, partly due to the fact that the children and families affected are ‘hard to reach’ with traditional research methods. There is also interest in the development of risk prediction or decision tools, often done using artificial intelligence or machine learning techniques. The better use of administrative data and research evidence within the family justice system has repeatedly been called for, for example by the 2011 Family Justice Review (Ministry of Justice [MoJ], 2011) and the FJO data scoping report (Jay et al, 2017).

In this context, the MoJ, as part of a process that began in 2011, recently piloted a new dataset—the CFJDS—along with its access arrangements. The MoJ invited a selected group of academics to submit brief proposals for projects to this end. The CFJDS contains linked data from the MoJ’s family courts administrative database named FamilyMan, Children and Family Court Advisory and Support Service (Cafcass) data, Children Looked After (CLA) data and National Pupil Database (NPD) data—these latter two being held by the Department for Education (DfE).

 

The background to our project

The UCL Legal Epidemiology group submitted a proposal examining the overlap between public and private children law. The courts deal with a range of disputes about the upbringing of children. These could be between private parties, such as parents, and concern matters such as with whom a child will live following relationship breakdown. This class of disputes is often called private children, or private family, law. Most private disputes that reach court are initiated under s 8, Children Act 1989, and hence are frequently referred to as s 8 applications/cases. The other class—public children law—refers to situations where the state, in its manifestation as local authority, intervenes to make decisions about children’s upbringing. This often (though not always) occurs in the context of child protection and has as its most dramatic manifestations the granting of care orders (by which the local authority may remove a child from her parents) and adoption. These can occur without parental consent where the court agrees that relevant threshold criteria are met and that the decision treats the child’s welfare as the paramount consideration. Care orders are granted under s 31 Children Act 1989 and adoption occurs under the Adoption and Children Act 2002. Children can also become accommodated by local authorities under s 20 of the 1989 Act in situations where there either are no parents or where the parents do not object; no recourse to court is required.

It has been suggested (Bainham, 2013), however, that this neat distinction often breaks down in practice in that many cases are in fact hybrids. Indeed, this can be seen to a limited extent from the routinely published figures in that many ostensibly public cases end with private children law orders; and it is suggested, though it is unknown the extent to which it happens, that many apparently private cases are driven by local authority impetus or support.

If there is overlap between private and public cases, then it is likely that those who appear in private cases are more likely to appear in public. This is the central question that we aimed to investigate using the CFJDS. Our study had three objectives:

1)         Identify yearly cohorts of children subject to section 8 applications and describe these children and the cases.

2)        For children subject to private law proceedings in 2011, quantify the proportion who return to court (public or private) or become looked after through extra-judicial legal routes within five years.

3)        Following completion of the pilot, to provide feedback to the MoJ based on our experience of the CFJDS and its associated secure Micro Data Lab. This feedback (which is provided in the present report) will be considered as part of an internal MoJ review of the pilot.

 

Main findings and discussion

Our main finding in response to the substantive legal research question was that of the 60,008 children involved in s 8 applications in 2011, 18% had further private law proceedings and 3.4% were involved in a public family law case by March 2015 (i.e. within 4.25 years). This is higher than expected, given that 3.3% of all children enter care at least once throughout the whole of childhood (Mc Grath-Lone et al, 2016).

Further detail is available in the attached report, which is also available on our website (https://www.ucl.ac.uk/child-health/research/population-policy-and-practice/child-health-informatics-group/legal-epidemiology-group) and which will feed into an internal MoJ review. The report makes a number of recommendations concerning the dataset and access procedures. Those which concern the dataset can e summarised as follows:

  • Parents: data on parents, who are central to children’s lives and who therefore are key components of any court action, are absent and could be included from the Cafcass database.
  • Data linkage transparency: the MoJ (and other data linkers) could provide sufficient information to enable researchers reporting results to comply with the GUILD guidance (Gilbert et al, 2018).
  • External area-level data linkage: mechanisms could be put in place to allow for external, publically available and anonymous data (such as deprivation indices) to be imported into the data analysis environment; this was not possible during the pilot and therefore certain analyses we wished to carry out were precluded.
  • Continued dialogue between data providers & researchers: continued dialogue between researchers and each of the data providers involved in the CFJDS could be supported and encouraged for the mutual benefit of the data provides and researchers. Examples of such interchange could include the Children’s Social Care Data User Group, seminars, webinars and user support groups. This could contribute to fostering a virtuous circle whereby data quality are improved, making better research possible which simultaneously informs service provision and data providers.
  • Data-specific recommendations: we also made a range of data-specific recommendations such as the manner in which data tables are made available, and specific variables (such as age, which is currently provided in age bands rather than year integers) to enhance analysis capability.

 

This work demonstrates the importance of data linkage to enable researchers to view the whole picture of family justice, for children involved in private family law proceedings. More generally, data linkage—both across life courses and between different data sources—can provide us with a fuller picture of an individual’s experiences. Data linkage, if low quality (i.e. high record linkage error), can introduce selection bias to the cohort. Therefore, it is important that the quality of data linkages are evaluated so that the results can be considered by all using the linked resource and also presented alongside any subsequent analyses.

 

Disclaimer

This work is entirely that of the authors. It does not necessarily represent the views of the Ministry of Justice or any other government or public body.

 

Author contributions

The authors designed the study and analysis plan and applied to the Ministry of Justice for access to the data. MAJ performed all statistical analyses. All authors contributed to the preparation of this report. Neither the Ministry of Justice nor the funder played any role in study design, analysis, interpretation or the preparation of this report. The views expressed are those of the authors.

 

Acknowledgements

We would like to thank the Ministry of Justice for providing access to the Children in Family Justice Data Share. In particular, Alexy Buck, Amy Summerfield, Sarah French and Lisa Robinson all helped to facilitate access. We also thank Charlie Smuliman, Nick Newcomen and James Wickenden for escorting MAJ while at the Ministry. Finally, we thank Ross Black for providing assistance with understanding the data and linkage process.

The Medical Research Council via a UCL-Birkbeck Doctoral Training Partnership Flexible Training Supplement funded MAJ for the duration of this project.

Files

Jay et al - CFJDS pilot report.pdf

Files (4.7 MB)

Name Size Download all
md5:0c86b5011f608bc48adde42581003155
957.2 kB Preview Download
md5:b371abdb2473779476582435631599e6
3.7 MB Download