• 10:00 – Arrival
  • 10:15 – Welcome and introductions – Sir Bernard Silverman
  • 10:20 – Updates on progress towards delivering the evidence in June and the consultation/recommendation – Becky Tinsley
  • 10:35 – EAP194 – Transition to a new methodology for the production of national mortality projections – Pamela Cobb and Camille Harrison
  • 11:30 – EAP195 – ABME Uncertainty Estimation Options – Mark Edward
  • 12:30 – Lunch
  • 13:00 – EAP196 – Calculating the excess mortality baseline – Sarah Caul
  • 13:45 – EAP197 – Census outputs and statistical disclosure control – Iain Dove
  • 14:30 – Update on Gender Identity – Jennet Woolford and Pete Large
  • 14:45 – Any other business
  • 15:00 – End of meeting

Panel Members

(In person)

  • Sir Bernard Silverman (Chair)
  • Dr Oliver Duke-Williams
  • Carl Emmerson

(Via teleconference)

  • Prof David Martin
  • Prof Natalie Shlomo

Office for National Statistics

(In person)

  • Owen Abbott (ONS Lead)
  • Magdolna Lorinc (ONS Secretariat)
  • Sarah Caul (presenter)
  • Lucia Barbone
  • Julian Buxton
  • Charlie Wroth-Smith

(Via teleconference)

  • Gareth Powell (ONS Secretariat)
  • Amy Curtis (ONS Secretariat)
  • Becky Tinsley (presenter)
  • Pamela Cobb (presenter)
  • Mark Edward (presenter)
  • Iain Dove (presenter)
  • Jennet Woolford (presenter)
  • Myer Glickman
  • Camille Harrison
  • Sarah Henry
  • Elizabeth McLaren
  • Manuela Naprta
  • Katie O’Farrell
  • Kirsten Piller
  • Keith Spicer
  • Nick Taylor

1. Welcome and introductions

  1. The chair of the panel, Sir Bernard Silverman welcomed all attendees.

2. Updates on progress towards delivering the evidence in June & the consultation/recommendation

  1. Becky Tinsley, the Deputy Director for Social Statistics Transformation, Analysis and Research, provided an update on the 2023 recommendation, communication and engagement plans. She mentioned that in addition to the list of publications in February, further evidence will be published in June. As part of the engagement plans, stakeholder roundtables are conducted prior to the consultation and until October. The consultation will run during the summer.
  2. MARP asked what the ONS expect from their review of the draft Consultation Document in May. ONS explained the document will provide an early sight of the direction of travel, so MARP can assess if this is reasonable and consistent with previous discussions.
  3. ONS described the aims of the consultation in more detail. ONS will present their vision for the future of population statistics and the evidence produced so far. They are seeking feedback on whether these meet user needs, and to set priorities for the next phase of work.

3. Transition to a new methodology for the production of national mortality projections EAP194

  1. ONS briefly presented the paper, next steps in the project, their questions to the panel and key points of the comments from the panel.
  2. MARP asked about the differences in population projections across different years between the current and new method. ONS explained that the difference partly reflects how the Covid pandemic impact is treated in projections. There was a discussion on how to include the impact of Covid and other shocks in mortality projections.
  3. MARP noted that uncertainty around the estimates is increasing over the years which needs to be reflected in future confidence intervals and communicated appropriately.
  4. The panel raised a question about the 1.2% annual mortality improvement rate quoted in the paper. ONS clarified that it is based on observed rates over the years and reflected on the likelihood of this rate to continue in the future.
  5. MARP highlighted the confusion about terminology such as life expectancy and the need for clear definitions provided to the public. ONS brought attention to exiting resources on the ONS website.
  6. The panel agreed that ONS should work towards making code available through GitHub at an appropriate time, followed by a discussion on the practicalities of this, including the necessary documentation and maintenance costs.
  7. MARP sought clarification on how the new mortality projections integrated with other statistics and the proposed future population and migration statistics. ONS explained they are exploring how to move towards a more integrated approach.
  8. This was followed by a discussion on the viability of running both old and new methods for a period and, as an alternative, applying the two mortality projection methods to previous estimates to assess the new method. It was generally agreed that providing uncertainty measures around projections from the new model would be more useful to users than dual-running previous estimates. The ONS would work on developing these.

4. ABME Uncertainty Estimation Options EAP195

  1. ONS presented the key points from the paper, highlighting the main sources of uncertainly in international migration estimates and recommendations on future work, followed by initial responses to the comments received.
  2. The panel noted that illegal migrants might have been living and working in the UK for a long time and asked if they are captured in admin data. ONS replied that if they are interacting with administrative systems then they are counted. They also stressed that the ONS is looking at data sources to quantify population and migration size, not to identify individuals. There was a discussion on the limitations of admin data to capture stocks and flows adequately.
  3. MARP asked for more evidence on calculating uncertainty.
  4. There were discussions on the limitations of admin data to measure international migration and on using survey data to adjust for coverage problems.

5. Calculating the excess mortality baseline EAP196

  1. ONS outlined the main points of the paper, then presented a summary of the comments and questions received from the panel, with responses.
  2. Responding to the comments received, ONS clarified they are investigating different time periods to use when calculating the baseline (the number of ‘expected’ deaths).
  3. The panel highlighted the importance of including ethnicity and other characteristics such as age, sex and levels of deprivation when providing measures of excess deaths. ONS explained that ethnicity is not collected as part of the death registration process, thus the analysis produced on ethnicity is based on linked data which impacts on the timeliness of results.
  4. ONS and the panel discussed the methodological implications of including/ excluding higher mortality periods in calculating the baseline, focusing on the Covid pandemic. ONS presented different options, highlighting the pros and cons for each.
  5. The panel advised the ONS to consult with the National Statistician and clarify what the exact definition of excess mortality baseline should be: (1) the average number of deaths for a certain period or (2) the most likely estimate that includes a certain possibility of adverse events, such as a high mortality flu season or another pandemic. The panel acknowledged that both solutions are satisfying if we are transparent about the limitations of the methods and urged ONS to test all options.

6. Census outputs and statistical disclosure control EAP197

  1. ONS presented a summary of the paper and their response to the questions and comments received from the panel.
  2. The panel acknowledged the usefulness of the custom dataset builder tool and asked for more details and further assurance that disclosure control risk have been managed appropriately.
  3. ONS agreed to submit a revised paper to the panel for further review.

7. Update on Gender Identity 

  1. ONS explained to the Panel that they were intending to carry out some research into the results of the gender identity question on the Census, and were seeking support from MARP to review and assure this research. The panel decided to provide this support by setting up a subgroup of the panel which could work promptly with ONS and would report back periodically to the main panel.

Action:

ONS to submit a revised paper on Census outputs and statistical disclosure control.