Apologies were received from Stephen Aldrige and Marcus Bell. The Chair welcomed Siobhan Dickens who attended for Marcus Bell.
Members approved the minutes of the November meeting.
2. Population Statistics System and Census SRO update
The Executive Director for Population, Census and Social Statistics gave an update on changes to the organisational structure to include the Director for Social, Local and Census Statistics, whose role would focus on statistical aspects of the Census. Members discussed the composition of teams and staff within each directorate and asked to see an overview of the resourcing and headcount plans.
Members heard that ONS would not make a wholesale move to admin-based population estimates (ABPEs) as originally planned, but were committed to continuous iterative improvement that would improve the data. The ONS affirmed that while the set criteria had not yet been met, this was still a positive decision in the context of holding a census in 2031 and aligned with the Devereux Review which advocated continuous improvement. ONS would bring a substantial item on ABPEs to the March Committee meeting.
The Committee heard that the dialogue with HM Treasury and Cabinet Office on the strategic outline case had been positive and constructive. Discussions were ongoing and the ONS agreed to share a summary of the business case with the Committee.
3. Census timeline and constraints
ONS introduced a paper to facilitate a discussion around the decision points within the timeline of the Programme. ONS wanted to ensure the Census was relevant to the context of 2031 and that operational elements were properly tested for such a large-scale event.
The Committee heard of the major constraints within the programme timeline and potential challenges, including elections in the devolved nations, a general election due by 2029 and the timing of necessary legislation to authorise the running of the Census.
The Committee considered the update. The following points were raised in discussion:
the White Paper and legislation straddled the expected general election period which may cause challenges, but also provided opportunity to garner cross-party support prior to the election;
there needed to be careful consideration over the communications approach and how to ensure public support to make the census a success;
the legislation window was challenging and it would be difficult to bring this forward as it could not be submitted during the general election period and it required the questionnaire to be included in full;
the risk of judicial review was acknowledged;
there was discussion of the risk of the social and political context changing significantly between 2027 and 2031 and how to adapt. The 2027 Test and other trials would test the design operations and statistical elements, but there was flexibility built in to be able to adapt to the context. ONS were able to do this in 2020 and 2021 in response to the Covid-19 pandemic for the previous Census round, and were learning from international statistical institutes;
the scenario planning should be relevant and actionable to ensure it was most effective. The ONS could learn from other areas of governments that use scenario planning to stress test policy;
members discussed how to plan most effectively to maximise digital responses and reduce the reliance and financial burden on the field force;
ongoing work on digital ID could have a similar timeline to the census, and members discussed the impact of digital IDs on public attitudes and census response;
the outcomes of the Topic Consultation and 2027 Test could indicate where admin data could provide the most value alongside or compared to a Census question; and
there was consideration of what the population statistics system would look like beyond Census 2031.
The Committee emphasised the importance of ensuring the public and politicians were supportive through external communications and work with political parties. A substantive item on communications and public engagement would be added to the forward agenda. Members discussed areas of government that may provide benefit to the census teams, regarding digital IDs, artificial intelligence (AI) and scenario planning.
4. 2027 test
ONS presented information on the 2027 Test, a large-scale statistical and operational test critical to the design validation phase. The Test would provide an opportunity to implement operational processes, test methods to increase responses and improve data quality, and provide evidence-based improvements to the overall design. An equivalent test was held in 2017. National Records Scotland (NRS) and Northern Ireland Statistics and Research Agency (NISRA) would also run their own tests.
The Committee heard that the Test would test ways to improve field force effectiveness and retention, test new ways of using admin data to target resources, improve linkage and reduce respondent burden, and, improve response for priority communal establishments. Specific local authorities had been targeted to take part in the test had been chosen based on criteria including Welsh speaking, urban and rural areas, low digital propensity, and university locations. In total, 220,000 households would be sampled.
Members discussed the update. The following points were raised in discussion:
members wanted further insight on the respondent journey from end-to-end;
members discussed whether the extent of experimentation was ambitious enough, and if more could be tested given we are targeting such a large sample. ONS noted that there was a risk that including too many experimental treatments could dilute the conclusive results. The 2027 Test aimed to validate statistical and operational assumptions, and other experimental aspects would be tested at iterative points throughout the Programme;
the Test had to manage cost and capability demands without retracting from other ONS priorities;
a digital first approach was discussed, including the SMS messaging which would be tested, as well as the potential to investigate the use of OneGov apps and AI to support respondents;
once available, the harmonised standards would support the Programme on questions on ethnicity, sex and gender identity (depending on the outcome of the Topic Consultation). The agreed questions would be subject to specific tests;
there was a risk of respondents using AI to complete the Census, members discussed the likelihood of this and how it would be possible to test for this and how to mitigate associated risks, as well as considering whether it had any benefits;
ONS was liaising with the relevant local authorities who could support the Test;
there was opportunity for cross government collaboration with the Department for Work and Pensions to support the recruitment and retention of field workers; and
different types of training for field workers were being tested to maximise effectiveness and value for money, including the possibility to complete the survey at the doorstep and how to target hard to reach groups.
The Committee emphasised the potential and value of the 2027 Test, and appreciated that ONS were implementing lessons learned from the 2017 Test. Members recognised the importance of testing innovation in encouraging response in a number of ways. The Chair asked that the test strategy returned to the Committee to ensure members were updated on the progress of the Test planning.
