DAViT

The Domestic Abuse Visualisation Tool (DAViT)

Background

Domestic Abuse (DA) is a public health concern and safety issue that can have severe consequences on individuals, as well as wide societal costs spanning across multiple sectors (e.g. Police, health, and other services, such as Social Care and others). A coordinated response across statutory agencies is vital to ensure DA survivors’ needs are fully met. This multi-agency coordination can be impeded by the inaccessibility of quality DA data.

In Surrey, DA data is produced and held independently by multiple organisations with limited sharing between them due to local culture and / or issues surrounding information governance and policy, considering the sensitivity of DA data. However, when DA data is viewed in fragments and in isolation, it may provide insufficient context to enhance service provision locally. For example, police departments countrywide use descriptive data analytics to ‘hotspot’ crime localities and mobilise resources to those areas. However, it is difficult to track DA trends with police data alone due to under-reporting. In this case, looking at police data in tandem with DA data collected by other agencies, like social services or health, might help to provide a more complete picture of the situation.

For more information about Domestic Abuse in Surrey, including help and support to those experiencing it, please visit the Surrey Against Domestic Abuse homepage.

The Domestic Abuse Visualisation Tool (DAViT) published here has been created by SODA partners to create a shared understanding of the intelligence that different agencies hold on Domestic Abuse instances in Surrey.

What can this tool be used for?

  • The intelligent commissioning of services for DA survivors and perpetrators.
  • The provision of relevant information to support planning and decision making, including bidding for additional resources.
  • The improvement of a multi-agency response, by making sure agencies complement what they individually know with intelligence from other sources.

The interactive Tableau DAViT tool is presented below. Please click on any one of the page buttons (“MSOA Maps”, “LAs over time”, or “MSOAs over time”) to begin using, or scroll down for more information relating to the tool. You may need to adjust the Zoom settings on your browser to get the best experience.

Description of the tool

DAViT is a Tableau-based visualisation suite that enables different teams to contribute aggregated DA data from their own records. These are used to produce maps and other data visualisations aimed at improving the multi-agency understanding of DA in Surrey.

Data mostly relate to survivors / victims (except for police data, where victim and perpetrator data cannot be separated), and can be filtered by:

  • Financial year – the UK government financial year starts on April 1st and ends on March 31st of the following year.
  • Sex – this is a binary indicator (male / female) due to the lack of more specific gender data from most datasets contributing to the tool.
  • Age group – Discrete banding categories from 0 to 90+, 15 groupings in total, including the category ‘unknown age’.
  • Geography – Middle Layer Super Output Areas (MSOAs) are units of geography, developed and released by UK Statistics, that show geographical areas with a consistently sized population. They have a minimum size of 5,000 residents and 2,000 households, with an average population size of 7,800. There are 151 Middle Super Output Areas (MSOA) in Surrey, that fit exactly within local authority boundaries.

Datasets included

Data sources included in this version of the DAViT tool derive from:

  • Surrey County Council Children’s Social Care (Contacts to Children’s Single Point of Access where DA is identified as a factor; and, Child and Family Assessments where DA is identified as a factor).
  • Surrey County Council Adults’ Social Care (Adult safeguarding enquiries with DA identified as a factor; and, Adult safeguarding enquiries with DA identified as a factor).
  • Surrey Police (DA recorded crimes and non-crimes).
  • Multi-Agency Risk Assessment Conference (MARAC) referrals
  • NHS Surrey Heartlands (Monthly Hospital Secondary Uses Service (SUS) dataset)

In this section, a definition of DA (which may vary for each organisation) for each dataset included in the tool is provided, including a description of how DA cases are recorded by each organisation and service.

Surrey County Council Adults’ Social Care

Adults’ Social Care is responsible for receiving and triaging safeguarding concerns raised about adults (these can come from private individuals, or professionals).
After the initial triage, the level of risk is determined, and a decision is made by the caseworker about whether the safeguarding concern should progress to a Section 42 safeguarding enquiry.
DA is one of the factors (risk or allegation) that are considered in the decision and is therefore a flag that is raised within the safeguarding concern data.

Surrey Safeguarding Children’s Partnership

The Children’s Single Point of Access (C-SPA) team is responsible for triaging all incoming concerns and referral requests for CFLL (Children, Family and Lifelong Learning). It is down to the professional judgement of the worker completing the form to identify whether DA is a factor based on the information they are given in the contact or during assessment.

In the DAViT tool we have included two indicators:

  • A contact is the initial request for social care involvement. The figures for all contacts include all C-SPA contact forms in the period, whereas those with DA as a factor count only those where one or more of the 3 DA related factors has been ticked (1. the child has been a direct or indirect victim of DA; 2. the parent or carer is a victim of DA; and, 3. another person in the household is a victim of DA).
  • A C&F assessment is carried out to establish whether a child has a high level of need. (Children with lower levels of need will usually be referred to Early Help or universal services, which don’t require a C&F assessment). We have included the number of all C&F assessment forms in the period, and specifically those where one or more of the 3 DA-related factors has been ticked.

Both C-SPA and social care assessment datasets include children who are impacted either as direct victims or due to their parent/carer or other member of their household being a victim of DA.

Multi-Agency Risk Assessment Conference (MARAC) referrals

The DAViT tool includes data on referrals to the MARACs (Multi-Agency Risk Assessment Conference), which are local meetings where professionals from several agencies (including police, probation, health, child protection, housing practitioners, and other specialists from statutory agencies and voluntary sector) discuss cases of high-risk domestic abuse survivors. After sharing all relevant information about the survivor, representatives discuss options for improving their safety and turn these options into a coordinated action plan.

NHS Surrey Heartlands

The dataset shows the number of hospital admissions where a patient has disclosed being subject to domestic violence. It includes data from hospitals across England where the patient has declared a Surrey address. Data is from the NHS England Digital Monthly Hospital Secondary Uses Service (SUS) dataset. This is a nationally mandated dataset that all hospitals submit to NHS England (data from community providers is not included at this point).

Surrey Police

The DAViT tool includes two indicators from Surrey Police having a domestic abuse marker against them: DA-crimes and DA non-crimes.
An initial investigation will establish whether a criminal offence has taken place. Where there is no criminal offence, but the person reporting perceives that the incident was motivated wholly or partially by hostility, the incident will be recorded as a ‘non-crime’ incident. Police officers may identify a non-crime incident, even where no victim or witness has done so.
A ‘crime’ event is where a criminal offence has been committed.

The Surrey police data relates to incident location of DA, regardless of victim or offender, so it will include data on perpetrators as well as victims.

Notes on Scores and Indexing Score

We have rescaled the number of records or incidents for each indicator to protect the privacy of individuals who, however unlikely, may be identified from the presentation of small case numbers. This approach transforms the actual values for each unique combination of period, age group, and gender for every indicator to a 0-100 scale, where 0 is the lowest number of cases observed and 100 is the highest for that combination.

This makes the scores easy to interpret and comparable across indicators.

The first step in doing this is to standardise the indicator values so that the scores are all measured on the same scale. Because we start with indicators that are measured in different units, they are standardised so that they fall in comparable ranges. We then identify the minimum and maximum values and use these to rescale the range of standardised values from 0 to 100.

We have also created an index combining all the indicators. To do this, we have taken the standardised and rescaled values for the indicators above and applied an equal weight to them to create an overall score and rank.

An alternative method of weighting was considered (inverse covariance-weighting), which generates an index by balancing the different effects present in a group of indicators, but this was rejected because the data were not invertible and required regularisation, which happens if the numerical columns are linearly dependent or there are too few data points relative to the number of features.

The combined index gives a score of 0 to 100 for each area, where 0 is the lowest possible score and 100 is the highest. This combined index permits the visualisation of hotspots with greatest prevalence of recorded domestic abuse.

How to navigate the tool

Three views are available on the DAViT tool.

Each view shows the distribution of indicator scores across 4 quartiles.

Quartiles are used to categorise values into four equal parts, each representing 25% of the selected dataset. In our visualisations, areas are classified into quartiles based on their index score values. The 1st quartile includes the bottom 25% of values, the 2nd quartile covers the next 25%, the 3rd quartile the following 25%, and the 4th quartile the highest 25%.

A colour is applied to the scores to aid visualisation, from clear purple for a lower score to darker purple for a higher score. White areas are areas where there is no data recorded for that geography (i.e. a recorded count of no DA incidents). This provides a clear visual representation on areas where reporting is not available.

1. Double Map View

Two identical maps are presented side-by-side, allowing users to compare results across different indicators using filters for both geographical and/or personal characteristics. The MSOAS have been classified into quartiles represented by a gradient of colour. There is also a dynamic scorecard below each map showing the same quartile classification for each MSOA for every indicator, based on the filters selected.

The user can select filters of interest from the drop-down menu in the navigation pane and results will be presented in a tool tip, where applicable. The tool tip highlights the selected criterion, as well as an allocated DA score.
The score shows the magnitude of how much or little information agencies hold on DA and we can compare across MSOAs in Surrey (please see below to understand how the DA score is calculated).

2. Local Authority trend bars

This section includes two Local Authority bar charts side-by-side displaying time series by year and quarter, using the same quartile colour classification, to illustrate changes over an extended period. At the top of the page, a set of filters allows you to slice and dice the data to observe trends over time and make area comparisons based on characteristics. This section presents actual values rather than scores.

3. MSOA Historic trend lines

It is also possible to view historical trends using values (where available) for the same set of overarching characteristics but presented as a line chart at MSOA level. Similar to the previous section, this one includes two identical charts over a period of time which can be modified using a set of filters.

Current limitations of the tool

  • It must be stressed that not every incident of domestic abuse will be known about by any service area. For that reason, areas where we have recorded no known DA incidents do not necessarily mean that no domestic abuse has happened there – only that it has not come to our attention. Areas with consistent lack of records most likely indicate we need to improve our detection methods for those communities.
  • The tool is helpful but we would like it to improve through incorporating a wider range of datasets from other agencies, such as probation data, or from outreach services. Adding further health-related datasets such as those recorded by mental health services, acute services, or general practice, for example, might provide a more complete picture. Considering that a great proportion of DA incidents are not reported to the Police, the integration of this data would provide a more comprehensive overview of the situation in Surrey.
  • Data labelling is not standardised across agencies, and there are gaps in information recording. For example, police data may refer to victim or perpetrator, and gender data is not available (i.e. only the binary male / female sex option is available, not reflecting the experience of LGBTQ+ survivors).
  • The ability to share data on DA is limited by information governance requirements that are not the same across multiple agencies. For this reason, different approaches have been developed to align with such requirements (e.g. concealing actual values through indexing). Where not possible, data is not shown.
  • The data shown is collected from contributing organisations, that aggregate it at source. This means data is not linked, and therefore what is shown are episodes, rather than individual people affected by DA. In children’s services data in particular, multiple records may represent one single incidence of domestic abuse – with a separate record for every child within that household.
  • Some sections of the dashboard use scores while others present actual values. Where values have been used, health-related data is concealed to permit the protection of individuals. This has been done since the underlying values for this dataset are small.
  • Where scores are selected, no scores will be displayed unless a single option has been selected for each of the available filters (quarter, sex, age group).
  • Where indicators lack variation and include a large number of Zero values, partial quartile classification can be observed (not using the full range of colours).