Methodological Assurance Review Panel minutes: 17 May 2022

Agenda

13:00 – 13:05 – Introduction – Sir Bernard Silverman/Oliver Duke-Williams
13:05 – 13:20 – Census Update – Jon Wroth-Smith/Owen Abbott
13:20 -13:50 – Bayesian Methods for Demographic Estimation (Research Update) -Louisa Blackwell/Duncan Elliot/John Bryant
13:50 – 14:05 – Social Statistics Transformation Forward Look – Becky Tinsley
Break opportunity
14:10 – 15:05 – EAP178 – 2021 Census Cohort Study – Nicky Rogers/Louisa Blackwell
15:05 – 15:10 – Actions Update – Dominik Thompson
15:10 – 15:15 – AOB – Sir Bernard Silverman

The chair welcomed attendees to the session.
Prof Ana Basiri and Dr Nik Lomax both were absent for this meeting and sent their apologies.

ONS provided an update on Census operations.
The panel discussed the methodological adaptions that have been made to the coverage estimation processes during operations. They also discussed why some population subgroups can be missed by the census and the possible mode effect. ACTION: ONS to evaluate the use of digital prompts in the online data collection for Census 2021 to increase coverage.

Louisa Blackwell and Duncan Elliott presented an update on the latest research and future plans for the Demographic Accounts using Bayesian Methods.
The panel highlighted the need for a reasonable narrative if there are notable differences between the published annual Local Authority research outputs that do and do not use Census 2021 data as inputs.
ONS agreed to make the technical paper for the design and implementation of the Demographic Accounts available to the panel.

Becky Tinsley presented an update on the latest plans to utilise MARP in assuring the evidence for the 2023 Recommendation. The panel were asked to highlight any gaps in the proposed MARP forward agenda.
The panel recommended planning an evaluation of how effectively MARP performs its assurance role in this period.
The panel highlighted the need to consider how UK-wide data is assembled in the future.

Nicky Rogers and Louisa Blackwell presented on the 2021 Census Cohort Study.
The panel ensured all comments left in advance had been appropriately responded to.
The panel questioned whether the ONS needed a 2021 Census Cohort within the current longitudinal landscape, especially if there may still be a traditional Census in 2031. ONS responded that experience of the Health Data Asset demonstrated a need for further linkage with or without a traditional Census.
The panel questioned whether ONS had considered pursuing only a sample cohort, rather than 100% of the 2021 Census cohort as this may make the design easier to maintain and quality control. ONS clarified that the cohort will not be truly 100% due to factors such as Census undercount and weightings to maintain representativeness. ONS also explained the need for a larger cohort to do more granular analysis than is currently possible through the 1% Longitudinal Study, where sub-national groups quickly reach single figures. The panel questioned whether ONS could find a sufficiently large sample, which would allow for these analyses.
The panel highlighted the ethical concerns of a linked sample of the population and sought more clarity on the advice given by the National Statistician’s Data Ethics Advisory Committee.
The panel sought further information around the Health Data Asset. The ONS provided a publication for information.