27th July 2020


Chair: Sir Bernard Silverman

13:00 – 13:15
(15 minutes)
Introductions, actions reviewPam Everett
13:15 – 14:15
(60 minutes)
Target for variability in response
Kamila Martyna
14:15 – 14:30
(15 minutes)
14:30 – 15:30
(60 minutes)
RAG Status and RCA forecasting methodology
(EAP139 & EAP140)
Alexandra Christenson & Brendan Davis
15:30 – 15:45
(15 minutes)
15:45 – 16:15
(30 minutes)
Discussion of the role of MARP in live census processingJon Wroth-Smith
16:15 – 16:30
Any other BusinessPam Everett

Attendee List

External Panel Members
Sir Bernard Silverman (Chair)
Prof Natalie Shlomo (External Panel Member)
Prof David Martin (External Panel Member)
Dr Nik Lomax (External Panel Member)
Dr Oliver Duke-Williams (External Panel Member)

Office for National Statistics
Pam Everett (Vice-Chair)
Cal Ghee (ONS Panel Member)
Owen Abbott (ONS Panel Member)
Jon Wroth-Smith (ONS Panel Member)
Orlaith Fraser (Presenter)
Kamila Martyna (Presenter)
Alexandra Christenson (Presenter)
Brenden Davis (Presenter)
Christopher Lydiat (Secretariat)
Harry Wrightson (Secretariat)

Sir Ian Diamond (ONS Panel Member, National Statistician)
Sarah Henry (ONS Panel Member)
Gareth Powell (Secretariat)


Agenda ItemActionOwner
[8,1]A65 – Establish whether a method to assess the combined variability of census elements is requiredOwen Abbott
[8,1]A66 – Cost the acquisition of data from other government departmentsSarah Henry
[8,5]A67 – Produce paper on field address checkJon Wroth-Smith


1.1 – The panel agreed to close actions A43 and A58.

1.2 – After requesting to close A40, it was agreed another action to investigate uncertainty should be raised, but A40 can be closed.

1.3 – The panel suggested a standing item on COVID, with a verbal update on its implications for methodology.

1.4 – It was agreed that the request by the panel to cost the acquisition and linkage of hashed and un-hashed HMRC/DWP data be formally added as a panel action.


A65 – Establish whether a method to assess the combined variability census elements is required.

A66 – Cost the acquisition and linkage of hashed and non-hashed data from other government departments.


The estimation strategy for 2021 Census is based on the 2011 Census, with more emphasis on characteristics than an area based estimation strategy. In order to meet census quality targets, the variability of response is important. The panel were taken through the details of this and proposed targets and how operations in the field will be by adapted by data to minimize variability in local areas.

Discussion & Suggestions

2.1 – The panel stated the need for careful explanation of the percentages used in communications.

2.2 – The grouping of all hard to count groups as a single set was questioned by the panel, as there could be certain hard to count groups that are meeting targets while others are missing. ONS responded that during the census rehearsal, there was significant variability within the hard to count groups. It was also stated that there would be limited operational impact, are there would always be prioritisation of hard to count areas. Additionally, the group represented 2% of all groups. The panel was content with this explanation.

No actions given.


The Red, Amber, Green (RAG) Status is an indicator of how each local authority is performing in census collection, with outputs informing prioritisation of workloads. The current indicators are a revision of 2011 with optimisations informed by data from the 2019 census rehearsal.

The Response Chasing Algorithm (RCA) forecasts responses to census operations and was tested in the 2019 census rehearsal. The panel were taken though the results of this, with results for operations largely verifying assumptions of RCA and operational values.

Discussion & Suggestions

3.1 – The panel suggested the use of hard to count data in the indicator. The small numbers of these individuals at local levels led to viewing this at a national level. ONS felt this could be too simplistic and specialists in hard to count groups would be better able to measure hard to count response levels.

3.2 – The panel questioned the weighting of various impacts. ONS clarified that the weights were standardised, which the panel were happy with.

3.3 – The panel asked about the assumptions of demographic differences in responses. ONS explained there were differences in male and female, which vary across age, and in hard to count groups.

No actions given.


Should contingencies be required, alterations to the statistical methodology may become necessary. Experiences other national statistics institutes and COVID led to more focus on this topic, and this discussion starts to establish the way ONS should work with MARP members to ensure these potential alterations are methodologically robust.

Discussion & Suggestions

4.1 – The panel were content to contribute to contingency procedures, and glad this was being considered ahead of issues.

4.2 – The panel recommended thinking about documentation of meetings where alterations were discussed.

4.3 – ONS stated the speed of decision making may limit the number of people on the panel available to respond. The panel accepted that this was a possibility but did not feel this would be an issue.

4.4 – The need to separate advice from assurance was emphasised by the panel.

4.5 – The panel raised that the terms of reference would need to be reviewed to ensure the panel would be able to commit time to this.

No actions given.

5.1 – An action was raised for a paper on field address check


A67 – Produce paper on field address check.