Members

• Professor David Hand (Chair)
• Sue Bateman (Central Digital and Data Office)
• Sam Cannicott (Centre for Data Ethics and Innovation)
• Colin Godbold (Independent member)
• Professor Donald Simeon (University of the West Indies, St. Augustine, Trinidad & Tobago and Royal Statistical Society)
• Tom Smith (Data Science Campus, Office for National Statistics)
• Jennet Woolford (deputising for Emma Rourke, Office for National Statistics)

UK Statistics Authority

• Alice Toms
• Emma Walker
• Simon Whitworth
• Lily O’Flynn (for item 6)

Apologies

• Emma Gordon (UKRI ESRC)
• Roger Halliday (Scottish Government)
• Reema Patel (Ada Lovelace Institute)
• Emma Rourke (Office for National Statistics)

1. Minutes and Matters Arising

1. The Chair welcomed the members to the second meeting of the Centre for Applied Data Ethics Advisory Committee, and members were thanked for attending.
2. Members were introduced to Tom Smith, Managing Director of the Data Science Campus, who is a new member of the committee, and Jennet Woolford who was deputising for Emma Rourke with the permission of the Chair.
3. The committee were made aware that the minutes for the previous meeting, for which there were no further comments received prior to the deadline, have now been published on the Centre webpages. This is in line with the CADEAC Terms of Reference.
4. Emma Walker updated the committee on the progress of actions from the previous meetings. All actions are complete or otherwise in progress.
5. Members of the committee were thanked for providing feedback on potential ethics topics after the previous meeting.

2. Brief Update on Centre Activities

1. Simon Whitworth presented a summary overview of the progress of the Centre, with specific reference to activities to date, and plans for the next 3 months.
2. The Centre launched in February and has been well received. The establishment of the Centre was a listed aim of the UK Statistics Authority’s Five Year Strategy, published in July 2020, to enable the UK Statistics Authority to become a recognised leader in the applied data ethics space.
3. Centre activities to date include:
   • A landscape review was published in March, which outlines what the Centre will offer in the context of the wider data ethics landscape, which was well-received.
   • The Centre’s first piece of expert ethics guidance was published in collaboration with ONS Geography as an open early draft in May. This focuses on the ethical challenges of geospatial data in the research and statistical context.
   • A roundtable event was hosted by the Centre in June, which discussed how ethical issues in research and statistics can best be addressed. Discussion from this event has contributed to the Centre’s future work plan and engagement strategy.
   • The team set up, and is leading, an ethics workstream as part of the UNECE Machine Learning 2021 group and is working on a piece of machine learning guidance as part of this work.
   • The self-assessment tool is showing increasing use, with over 100 project applications having been received by the team so far this year.
   • The Centre continues to provide ethics training across government.
4. The Centre has ambitious plans for the next three months and these include:
   •  A guidance piece on articulating public good will be published as an early open draft shortly. A need for this emerged from the support provided by the ethics services team to researchers completing the self-assessment tool.
   • Guidance on the ethical considerations of machine learning in the production of official statistics is currently being drafted and will be published as an early open draft shortly.
   • Online ethics training module is currently being produced by the team which outlines the UK Statistics Authority’s ethics services. This will be made available openly online for both internal and external users. The current timeframe for this is for the module to be published within the next 6-10 weeks.
   • Further guidance on the ethical use of data from third parties, linked data and genomic data is also planned.
5. The committee were impressed by the outputs and activities of the Centre to date and were interested to learn how activities would be prioritised going forward. The committee were advised that the Centre attempts to make activities collaborative, and that in many instances the Centre takes on a coordination and collaborative role, working with various stakeholders to produce guidance which is useful to the research community. The Centre is user led in many ways and remains open in terms of the future work plan as to what ethics topics guidance it will cover.
6. The Centre is keen for its work to focus on the applied ethics space, and to offer support for researchers and statisticians. The Centre understands the importance of having direct impact on research projects in order to empower researchers and enable ethical research. The self-assessment ethics tool has been very successful in achieving this, and the Centre has identified a demand to further improve this tool, as it has been useful in applying ethical principles to practice. This is a major way in which the Centre can demonstrate impact.

3. Centre Impact: What does good look like and how can we measure it?

1. Emma Walker presented this item, which was developed as an action following the previous CADEAC meeting and outlines how the Centre intends to measure its impact.
2. Members of the committee were invited to provide feedback and approve the proposed Centre impact indicators. Specifically, the Centre asked the committee to identify which metrics they thought would be most beneficial for the committee to view in dashboard format on a regular basis.
3. Indicators have been chosen to encompass a range of target audiences and beneficiaries of the Centre, grouped around the four main types of Centre activity which are: the provision of guidance; leading international engagement, collaboration and engagement with UK stakeholders and; the provision of user support.
4. The Centre suggested to the committee that a process of regular review of these indicators would be beneficial. It was proposed that initial informal review would occur every 6 months, and more formal review on an annual basis as part of the Centre’s report to the National Statistician’s Executive Group.
5. The Centre will collect data related to the indicators, and so feasibility for the existing data ethics team and their resources needs to be considered. The committee raised concerns around the temptation to be overly ambitious in the number of indicators measured and suggested that caution should be exercised in this respect and the various indicators prioritised. However, it was acknowledged that the Centre is well positioned to measure impact, due to its collaborative nature.
6. Members of the committee identified that metrics related to user support- given the focus of the Centre on the applied data ethics space and the provision of practical support for researchers- would be particularly beneficial for the Centre. There was a suggestion that case studies of positive impact for user groups would also be beneficial in understanding impact.
7. The committee highlighted a need for evidence of the impact that the Centre has in relation to influencing user behaviour with regards to the ethical use of data within projects, and that narrative evidence from researchers illustrating that they like what the Centre is doing would not be sufficient to demonstrate this.
8. It was suggested by attendees that whilst the impact measures outlined focus on the positive impacts of the Centre, it could also be useful to measure/highlight any criticisms from users relating to the Centre. The committee felt that it may be helpful to consider how projects which have been negatively received, either by NSDEC, or more widely, can be used to develop the Centre’s understanding of its potential impact. For example, could difficulties with projects have been mitigated if researchers had used the Centre’s guidance?
9. The committee emphasised the need to distinguish between short term and long-term impact, with the concern that impact could be slow to visibly materialise. As a result, it will be important to consider how often these metrics are assessed and the different types of impact achieved.

4. Stakeholder Engagement Plan

1. This item was presented by Emma Walker and was developed as an action following the last meeting.
2. The plan highlighted 7 main stakeholder groups which the Centre most commonly engages with, and three broad types of engagement – inform, consult and collaborate- which each utilise a variety of different mechanisms to engage with the needs and interests of each stakeholder group.
3. Members of the committee were invited to provide feedback and approve the stakeholder engagement plan.
4. The committee felt that the plan covered most main stakeholder groups, but that other groups should also be considered, even though their relationship with the Centre may be more indirect. Specifically, the committee highlighted the following stakeholders:
   • The public. There was discussion as to how the public would make use of the service, and whether they would have direct engagement with the team. The committee acknowledged that the public are not direct consumers of the services offered by the Centre, but that this group does have an interest in the way that government uses public data. The Centre’s website, (and the online training module currently in draft) is available to the public to view, so there is a minimum requirement to at least be transparent in communicating the aims and activities of the Centre.
   • The media.  The committee felt that the media are an important actor in ensuring public trust in the use of data (particularly by the government), which is one of the intended outcomes of the Centre. Whilst the media may not be in direct contact with the Centre, it is still important to consider their potential impact.
   • Industry and commercial sectors. The committee felt that it would be beneficial for the Centre to consider what they can both learn from, and contribute to, the commercial sector. It is important that the Centre does not work in a silo, as much can be learnt from other sectors when considering the ethics of data use. The Centre has been in discussion with several commercial organisations, and collaboration with these companies is ongoing. This highlights the need for this group to be considered within the engagement plan, though this might mean using different principles and frameworks of engagement than those set out for other stakeholders.
5. It was acknowledged that the data science community is a wide and diverse one, and despite the team’s concerns that some groups within this community may not feel that the work of the Centre is relevant to them, the committee felt that the engagement plan did a good job of including these groups in their strategy.
6. The committee were interested in knowing the extent to which the Centre hopes to engage with each stakeholder group and fulfil the various aspects set out in the engagement plan. It was highlighted that the Centre will predominantly use the plan as a guideline for consideration when undertaking Centre activities rather than engaging specifically with each group listed.
7. The committee highlighted that there would need to be prioritisation in regard to engagement with different groups, due to finite resources. It was suggested that as the Centre is not the only actor within the data ethics space, it will be beneficial to collaborate with others when planning engagement activities.
8. The committee considered how the Centre could effectively measure the success of engagement activities, and it was highlighted that some of the metrics previously outlined will allow the Centre to measure impact on/with different user communities. It will be necessary for the Centre to iteratively reflect upon this over time.

ACTION: The Centre team will reflect on the comments made regarding additional stakeholders to consider and amend the engagement plan to reflect this.

5. Ethics Guidance Pieces: Ethical considerations in the use of geospatial data for research and statistics and Ethics guidance for Public Good

1. This item was presented by Emma Walker. Emma provided the committee with an overview of the first two pieces of guidance produced by the Centre – geospatial ethics guidance and public good guidance.
2. Committee members were reminded that the geospatial guidance is currently published in draft form for wider comment from the user community, and the public good guidance is in development.
3. The committee were reminded that guidance has been produced in response to feedback and discussion with the user community. The geospatial guidance was identified as important as a result of internal user need, and an acknowledgement that there was a gap in this area.
4. Committee members were thanked for comments that they have already provided for the geospatial guidance. Suggestions, comments, and feedback are currently being collected and collated for both pieces of guidance, and members of CADEAC were invited to provide further feedback and comments.
5. Members congratulated the team on the production of both pieces of guidance and were particularly impressed by the public good document.
6. The committee were considerate of the legal challenges surrounding the definition of public benefit and were reassured that the Centre had sought legal review of the public good guidance. It was emphasised to the committee that in considering these challenges, the Centre made the deliberate decision to frame the guidance in the context of the ONS/UKSA strategy for providing data for the public good.
7. Members of the committee were aware that the definition of public good and the articulation of this concept has been a long-standing challenge for researchers, and were pleased that this guidance addresses these issues, whilst presenting a practical approach for researchers to help navigate through the articulation of public good, and guided by the principles set out in legislation.
8. The committee were positive in their feedback relating to the structure and format of the guidance and were keen to share both pieces of guidance with colleagues. There were some concerns that the checklist questions in both pieces of guidance, but specifically for the geospatial guidance, could be difficult to answer due to the nature of the issues addressed, and that the provision of more substantive answers or examples of how these questions could be answered might be helpful, although it was recognised that this could be challenging to provide.
9. Members were reminded that the public good guidance will be published as an open early draft for comment on the Centre website, and that this will be accessible to all.
10. The committee felt that it could be useful to get an international perspective on the guidance pieces, and how they translate to other contexts.
11. Discussion surrounding how the Centre outputs relate to the service side of the ethics team emphasised the importance of making it clear that the guidance pieces are relevant to the services offered by the team. More detail regarding how the guidance and the service relate could be beneficial.
12. The committee were reminded that draft guidance produced by the Centre is sent out to specific expert groups and interested parties for feedback, prior to publication. The Centre has made a concerted effort to search out targeted user feedback to ensure that outputs are helpful, but this of course takes time and feedback is still being collated for the geospatial guidance.
13. There was a consideration that whilst the public good guidance provides examples of how principles should be applied, the guidance does not identify negative behaviours, or examples of what would not be an appropriate consideration of public good. It could be beneficial for researchers to know what they should not say/do, as much as it is beneficial to know what they should. This is a technique used at an operational level with researchers via NSDEC, so could be very beneficial if presented appropriately.
14. The committee identified ethical challenges associated with genomic data use, the inclusivity of data sources and admin data as being topics for consideration for future guidance.
15. The committee felt that using stress-testing techniques could be useful in helping to identify current gaps in ethics guidance topics and inform Centre priorities. This could help the Centre to identify guidance topics in preparation for future needs, rather than waiting for these needs to arise.

ACTION: Members of the Centre will update the relevant guidance pieces where possible to reflect feedback from the committee.

6. Centre Activities: Ethics Services update

1. Lily O’Flynn presented this item. The presentation provided members of the committee with an update on the data ethics services. The service side of the team supports researchers in understanding and applying the UK Statistics Authority’s ethical principles.
2. NSDEC met last week, and the meeting considered projects form ONS, across government and academic researchers. The committee considers specific project proposals, wider ONS processes, and provides comment on the guidance produced by the Centre.  The committee were shown a breakdown of which types of users apply the self-assessment tool to their projects. The tool has sustained engagement from a wide number of organisations and sectors, and this is expected to increase, particularly as a result of the ONS integrated data plan.
3. Lily outlined the key metrics related to the self-assessment tool. These tell us that guidance from CADE is in demand from users of the tool. The ethics service team can track the types of data and methods that researchers using the self-assessment tool are using, which provides the Centre with the opportunity to identify areas where additional support and guidance would be beneficial.
4. The ethics self-assessment tool has been used more than 110 times this year so far. 25% of researchers using the tool have been from the Office for National Statistics, whilst others using the tool have come from a variety of areas including academia, other government departments, and commercial organisations.
5. The self-assessment user review showed that 3 in 4 researchers surveyed understood how the UK Statistics Authority’s principles relate to the tool and agreed that the tool made it easy to apply principles to their work. 4 in 5 found the range of user support helpful for completing the tool, and 2 in 3 thought that the guidance provided by the Centre was easy to find. Data sharing, understanding public good, and algorithms were areas researchers commonly identified as needing further guidance.
6. Going forward, the Centre hopes to improve the tool to make it more user-friendly. The Centre aims to provide more guidance for researchers as to what information is needed from them to get the most out of the tool and the ethics process. It is hoped that this will assist the team in providing more timely, focused, advice on projects. There is a need to provide more information on how each principle is weighted, and how scoring items differently will change the overall ethical ranking of a project, so that researchers better understand this process. An effort will also be made to make the tool and associated guidance more accessible, and easier to find, increasing visibility throughout the user community.
7. The Committee were invited to ask any questions or provide any comments on the existing data ethics services.
8. Members of the committee who have a role within NSDEC were invited to share their thoughts. The service was discussed very positively, and it was recognised that very rarely does NSDEC get a project presented to them which is thought to be ill-founded for ethical reasons. This was thought to be a result of the work of the ethics service team and the self-assessment tool. Projects that are ethically concerning appear to be identified very quickly, and ethical issues are being considered early in the research process, rather than later down the line when they could be problematic.
9. It was highlighted that projects presented to NSDEC are often done so at one of three different stages – concept stage, as a feasibility study, or once the research design has been finalised, and so advice is tailored based upon stage.
10. Members of the committee felt that it would be beneficial to provide more ongoing ethics support. This was a big consideration when the tool was created, but because the researchers themselves own their version of the tool, the researchers are actively encouraged to revisit their application, and are always welcome to contact the team at a later date, though this puts the responsibility on the researcher.
11. The committee felt that it may be beneficial when collecting metrics to break down use cases and users into typologies, persona’s or sub-groups to better identify gaps in knowledge or support and to better understand differing user needs and engagement.
12. The committee requested further information as to how approved research is monitored, and how the services ensure that researchers do not change their research without revisiting the tool. The team tries to manage this by creating hard checkpoints, where researchers accessing data are asked to provide evidence of their ethics self-assessment though this only happens within the ONS or accredited infrastructures. A selection of projects are also audited by colleagues from the data protection team on an ongoing basis, with the results of this presented to NSDEC. However, there remains a responsibility on the Centre to engage with researchers to remind them that they should be revisiting their ethics assessments regularly.

7. Any Other Business

1. The Chair thanked the committee for their involvement, and their contribution to the meeting. Members were invited to raise any other business.
2. The next meeting of the Centre for Applied Data Ethics Advisory Committee will be held on 6th October 2021.

ACTION: The committee were invited to email any further contributions to the discussion should they wish to.