Enhancing insights and coherence through collaboration

This is a case study for Principle V3: Clarity and Insight

While the Department for Work and Pensions (DWP) has responsibility for administering most benefits available in Great Britain, some benefits such as Tax Credits and Child Benefit are administered by HM Revenue and Customs (HMRC).

In March 2020, DWP and HMRC produced a new joint-release for the first time on children in low-income families (CILIF) at a local area level. This release has replaced DWP’s Children in out-of-work benefit households and HMRC’s Personal tax credits: Children in low-income families local measure releases. The new statistics provide a more coherent picture of children in low-income families by drawing together administrative data from both producers to provide insights on benefits, tax credit and employment incomes within families from which local area estimates of children in low-income families are published. From 2021, this release is now solely produced by DWP.

DWP also produces Benefit Combination statistics, which offer a picture of the number of individuals claiming at least one benefit as well as the number of claimants for each combination of benefits. Currently users need to look at statistics produced by both the DWP and HMRC to gain a complete picture of how many families and households are claiming benefits.

DWP and HMRC have been working together to develop a joint publication which would bring together the Tax Credits and Child Benefit statistics with those produced by DWP. There have been challenges with delivering this work due to the difference in timing of the data feeds that DWP and HMRC work with. However, they are working towards producing annual statistics on the numbers of individuals claiming common combinations of benefits which DWP and HMRC are responsible for, on a common snapshot date.

The Heads of Profession for Statistics in DWP and HMRC have discussed this joint publication and are supportive of the work to develop the experimental Benefit Combinations statistics. In the interim, DWP has provided details in the DWP benefit statistics background note of benefits administered by HMRC and signposted to relevant HMRC statistics. While work is still ongoing to develop the joint publication, it is great to see the joint focus on aiding public understanding of benefit provision by developing these statistics.

This example shows how producers can work together across departmental boundaries to make greater use of existing data sources through linking, and the benefits this brings in terms improved insights and coherence across a statistical topic.

Improving quality assurance and its communication to aid user interpretation

This is a case study for Principle Q3: Assured quality.

The Department for Work and Pensions (DWP) publishes statistics on new National Insurance number (NINo) registrations to adult overseas nationals, on a quarterly basis. In 2017, following an assessment by OSR, the statistics had its National Statistics designation suspended. Since then, DWP have implemented a range of improvements leading to re-designation of the statistics in November 2020. One of the areas of greatest improvement is their quality assurance processes and how they communicate these procedures with users.

The team have worked hard to review their quality assurance processes and apply guidance from OSR’s Quality Assurance of Administrative Data (QAAD) toolkit, as discussed in their Quality report. Applying guidance from QAAD supports good practice in monitoring the quality of data over time and in identifying data quality issues, whilst the Quality report itself demonstrates transparency about the quality assurance approach taken.

More recently, the production team have focused on harnessing aspects of Reproducible Analytical Pipelines (RAP). DWP have a separate dedicated team focused on introducing RAP across the department. Working closely with them, the production team – recognizing the benefits associated with quality assurance, streamlining and automating the production of the publication – have started to adopt many of the RAP principles. This has also involved team members upskilling and developing their programming skills. These are DWP’s first statistics to be produced using RAP and the team continues to develop and further integrate RAP principals into its production processes.

To further explain the quality assurance processes, the administrative data and the strengths and limitations of the statistics, an informative background and methodology document is available for users. This includes illustrative figures of the data journey – showing how different data sources fit together – and flow diagrams documenting the steps in the quality assurance process.  The inclusion of these helpful diagrams, alongside descriptions and further detail within the text, offers users a clear explanation of the strengths and limitations of the data, and enables an understanding the underlying data and processes.

This case study shows how DWP has made improvements to how it assures the quality of its NINO statistics and how it communicates this assurance transparently to aid users in their appropriate interpretation. The enhanced quality assurance processes and their clear communication through new documentation, reflect a positive and engaged approach to enhancing the quality of the statistics and improvements to their overall public value.






Making government data available for reuse

This is a case study for Principle V5: Efficiency and proportionality.

The Department for Work and Pensions (DWP) introduced Stat-Xplore, an open data tool which provides a guided way to explore DWP benefit statistics and helps to ensure that DWP statistical data is used, re-used and understood by a wide range of users.

Stat-Xplore currently holds data relating to 16 different DWP benefit programmes with plans to extend this to others in the future. The data on Stat-Xplore is the same data used in DWP statistical publications, so it does not impose any additional burden or require any additional supply of information from DWP sources, making efficient use of the statistical data DWP already holds.

DWP data on Stat-Xplore is available made in a number of ways to suit many user types, (including the Enquiring Citizen, Information Forager and Expert Analyst) with a helpful user guide, an interactive tour and extensive metadata provided to support users get the most value out of the data available.

The data contained in Stat-Xplore draws on recognised standards, classifications, definitions and methods. For example, many of the data sets draw on standard geographical units, and data perturbation/statistical disclosure control is applied to avoid the release of confidential information.

In conjunction with statistical publication consultations, user feedback and queries relating to Stat-Xplore datasets help DWP statisticians to determine whether new or extended data requirements are needed, thereby helping to inform decisions concerning the possible impacts of changing data requirements against their potential value.

Stat-Xplore has been upgraded to better meet user needs since its inception. User feedback has informed the further development of:

  • Ready-made tables available for most sets of benefit data, for ease of use
  • User defined tables which can be saved in a variety of formats (Excel, CSV, XML, ZIP), with additional large table functionality available to registered users
  • Derivations – allowing users to create new calculated items within a table such as adding together values in other columns or using mathematical and statistical functions
  • Linked data visualisations for 5 key benefits, hosted directly onto Stat-Xplore
  • Mapping – a relatively new feature which allows users to view results from geographical fields inside an interactive map
  • Open Data API – Registered users also have access to all datasets via the Open Data API. Users can build their own third-party interactive reports and applications that retrieve data directly from Stat-Xplore and automatically update every time DWP releases new data. The API also caches new requests to make the future retrieval of the same data much quicker

DWP has itself been making use of the Stat-Xplore Open Data API and JavaScript to build interactive visual dashboards (this tool is no longer available but you can access the fraud and error in the benefit system hub here)  that give a live summary view of the data in Stat-Xplore. DWP Data Scientists are also making use of the Stat-Xplore Open Data API in their products and have begun to promote this work more widely.

Stat-Xplore makes and is a valuable resource which makes efficient use of DWP statistical data and promotes its re-use both within DWP and more widely. Users access is well supported through extensive guidance, recognised standards, classifications and metadata. The analysis of users’ queries also supports DWP statisticians in determining whether value can be added to DWP statistics by adding new or extending existing data requirements.

Voluntary Application: Department for Work and Pensions

Why does the Department for Work and Pensions (DWP) voluntarily apply the Code?

DWP doesn’t just publish official statistics. We also produce a wide range of “other” data and analysis that are important in designing and evaluating policy. Core activities in DWP involve using data to: make decisions, track policies, forecast spending, and much more.

Our data are collected, reused, and analysed by many other groups and organisations outside government – for example, our interactive tool Stat-Xplore enables users to delve into the data and produce bespoke analysis.

The mix of data we use, produce and publish provides a rich range of data that contribute to the evidence base available to individuals and decision makers – information that can ultimately affect each person in Great Britain. So, we want to ensure that any use of statistics in our communications, from a tweet to writing a press notice or creating a presentation, should align with the Code, to ensure messages are clear, measured and appropriate for the story.

What does voluntary application of the Code mean?

It means that we have publicly committed to demonstrating trustworthiness, reassuring users about the quality of our information, and ensuring that the information serves the intended purposes.

Confidence and trust in our data and statistics are fundamental in how the DWP Strategy is delivered:

  • From understanding and managing the performance of the business and helping inform the direction of economic and commercial activities;
  • Allowing the formulation of better public policy and the effective measurement of those policies, including anticipation of risks and pressures on the system;
  • Enabling the public to hold to account all organisations that spend public money, and informing democratic debate and public understanding of our society.

Ensuring that our data and statistics reflect user needs, support better decisions and gain public trust is a shared responsibility for everyone working in DWP. It is essential that those of us who are actively engaged in this area understand our roles and responsibilities.

We have an aspiration to further embed voluntary application of the Code, with the aim of having a shared aspiration across all analytical areas in the Department around when and how we demonstrate our commitment to trustworthiness, quality and public value.

How do you do this – and is it easy?

We apply the four voluntary application stages that aim to help produce analytical outputs that are high quality, useful for supporting decisions, and well respected.  The stages are Understand, Review, Consider and Publish. The commit stage (Publish) is publishing the statement about why users can be reassured that the product achieves Trustworthiness, Quality and Value (TQV).

In leading up to that decision, it is helpful to understand the three pillars (TQV), review the approach to producing and publishing statistics/analysis in relation to the pillars, and consider if there are ways of improving practice. By publishing this statement, we are being transparent.

Our key is to think through each of the pillars:

  • How are we trustworthy? Should people trust what we provide?
  • What is the quality of the data? Is what we are doing the best estimate? Is it misleading? Could people make the wrong decision?
  • How do we provide value? Is it helpful? Can users make decisions using the information?

We have had conversations across all analytical areas to raise awareness. We shared guidance and given advice to analysts who have voluntarily adopted TQV. We also obtained senior backing for applying TQV in the department – our chief analyst said:

“I am really keen for us to be as ambitious as possible in applying the ‘Voluntary Compliance …’ approach to as many outputs as appropriate from the Analytical Community.”

Here are some examples of where we have applied the Code beyond official statistics

Benefit Expenditure tables: We have added a statement of voluntary compliance with the Code to our release of the expenditure tables showing historical and forecast benefit spend. Ministers see these tables well in advance of publication as they’re part of Budget and Spending Review negotiations – so we can’t achieve all the requirements for official statistics. Instead, our statement sets out why we think this is high quality analysis, why it is trustworthy and how it adds value to the debate.

Social Research Reports: Voluntary application has started to be used in DWP research reports, as agreed with the Head of Professional for Social Research, for example, the GSR Automatic Enrolment Evaluation Report (see page 116).  The report brings together the latest evidence and new analysis to show what has happened to workplace pension membership and contributions since automatic enrolment began.

Management Information (MI) is used for operational delivery and the running of the business. We have now begun to apply the Code pillars in our preparation and release of the MI, for example ESA Underpayments and Support for Mortgage Interest MI.

What do you think are the benefits of applying TQV?

It gives you an opportunity to highlight your commitment to demonstrating trustworthiness, reassuring users about the quality of your information, and ensuring that the information serves the intended purposes.

It provides a quick win to improve users’ understanding of the data presented. It also demonstrates a common-sense approach to sharing numerical information and prompts you to review your own processes.

“I found the voluntary Code helpful. The headings (TQV) made sense and it was a useful structure, in fact it reassured us as the producers as much as anything. We also enlisted an analyst from another team to quality assure the peer review of our analysis, which again proved valuable”.

[DWP statistician]

Ultimately, it builds confidence and trustworthiness.