Possible new approaches to the collection and use of homelessness data in Wales

On 1 September CaCHE published a feasibility study into a new case level homelessness data system for Wales. In this blog, Dr Ian Thomas provides an overview of the four potential designs for a new homelessness data system that emerged from the study.

In their strategic response to homelessness released in 2019, the Welsh Government set out their vision for a Wales where homelessness is rare, brief if it occurs, and non-recurrent. Data play an important part in achieving this vision for Wales by enabling policymakers and practitioners to measure and monitor the scale of homelessness and to test the effectiveness of their efforts to reduce it.

At present, the main sources of information on homelessness used by the Welsh Government and the homeless sector come from mainly aggregate data. Sources include counts of households assisted by local authority housing teams, and local authority level counts of rough sleepers bedded down on a single night. However, these data sources lack the detail and information needed to assess whether homelessness in Wales is rare, brief, and non-recurrent. To measure ‘success’ against these strategic aims, the sector would require data on:

  • the number of unique individuals experiencing homelessness—not simply the number of households which can include any number of people, some of whom may go on to become homeless themselves
  • some indication of the length of time homeless—whether that be the time between assistance start and ‘successful’ housing outcome, or the ability to link between housing team and rough sleeper data to identify when someone returns to the streets after being housed
  • the ability to identify or follow people through different systems—such that their re-use of the same service, or appearance within a different part of the homeless service sector can be assessed

To (re)design aggregate collections to achieve these data needs would be impractical to say the least, with an individual level data collection being the most effective and flexible approach. Individual-level data collection has the added benefit of increasing the sector’s capability to conduct research into homelessness and evaluations of their response to it. Research and evaluation of what works have the potential to be a game-changer in ending homelessness.

Welsh Government asked UK Collaborative Centre for Housing Evidence (CaCHE) and researchers based at Cardiff University/Administrative Data Research Wales, to explore different options for a new individual-level data collection on homelessness in Wales. Four options emerged from their feasibility study:

Transformed reporting

In its simplest form, homelessness data collection in Wales could be transformed by requiring local authority housing teams to provide individual-level data to the Welsh Government, rather than aggregated data as they currently do. As with other individual-level data collections, such as Looked After Children data, the Welsh Government would be responsible for managing a centralised data set, including tasks such as data quality monitoring.

Data warehousing

Rather than individual-level data being provided directly to the Welsh Government and stored within their local systems, a ‘data warehouse’ could be created external to the Welsh Government, whilst still being under their governance structures. The benefit of an external data warehouse is that software that interface with the warehouse can be designed by the homelessness sector, as and when they are needed, without the restrictions of having to operate with(in) a Government IT infrastructure.

Federated data

Both the transformed and data warehousing options require data to leave organisations before being combined. As an alternative, ‘federated’ data models enable organisations to retain data within their local systems. A ‘data broker’ acts as go-between for anyone who has a request for information—such as researchers or policymakers—and the data owner. If a data owner agrees to a request for information, the broker is authorised to access and prepare extracts of data and can combine data from different organisations. Federated systems are heavily adopted in Statewide Longitudinal Data Systems in the United States, to enable access to linked education data within particularly restrictive data sharing environments.

Pan-Wales information platform

The options presented so far facilitate uses of data with a time-lag between the creation of the data and any impact on people experiencing homelessness. For the collection of data to have more immediate impacts on people’s experience of homelessness, a final option is proposed whereby a single information-sharing platform is used by local authority housing teams and the homeless sector. Individuals who are homeless would have a single entry on the platform, from which all organisations involved in resolving that person’s homelessness would work from. Communities in the United States adopt this type of shared sectoral information management system, which functions as both a way of helping people in the moment by enabling them to access the most appropriate services for their needs, as well as providing data for national reports to the government on the scale of homelessness.

The full details of these options are presented in the CaCHE report which was published on 1 September 2020.

Dr Ian Thomas is a Research Associate based at the Administrative Data Research Centre Wales/Wales Institute of Social & Economic Research, Data & Methods, based at Cardiff University.


Date: September 15, 2020 11:26 am


Categorised in: