Executive summary

The National Statistician’s Inclusive Data Advisory Committee (NSIDAC) welcomes the opportunity to comment on the ONS proposals for the Future of Population and Migration Statistics in England and Wales (FPMS). In summary, following the UK Government’s ambition, as set out in 2014, that “censuses after 2021 will be conducted using other sources of data and providing more timely statistical information”, ONS has proposed and is consulting on a system primarily using administrative data such as tax, benefit and border data, complemented by survey data and a wider range of data sources. It is suggested that this could produce more timely data both at a national and local authority level and would replace the current reliance on and need for a census every ten years.

Our aim is to review the FPMS proposals in light of their implications for inclusivity of the data. A fundamental principle of inclusive data as set out in the Inclusive Data Taskforce recommendations report: Leaving no one behind is that ‘everyone in society counts and is counted and no one is left behind.’ More specifically, following the criteria established by the Inclusive Data Task Force, we evaluate the proposals in the light of whether they address and remedy (or the reverse) critical data gaps the Taskforce identified, barriers to data collection and challenges to conducting analysis relevant to inclusion and equality.

We recognise the excellent work that the ONS is doing to improve the quality and timeliness of population estimates by drawing on administrative data and support the ambition for more timely data, providing it is accurate with respect to measurement of at-risk groups. We also recognise that many elements of the proposals are unproblematic and we strongly support ONS’s programme of work to improve the quality and coverage of administrative data, to create linked datasets such as the Statistical Population Dataset (SPD) and the Admin-Based Ethnicity Statistics (ABES); and to improve inclusiveness in areas which are not currently counted or fully considered, such as looked-after children and veterans. Indeed, we are in favour of changes which advance inclusive data generally. But we are concerned that the census still has a role to play in ensuring the inclusivity of official statistics and that there are risks for inclusivity if a census is not conducted in 2031, in particular that there might be deterioration, rather than improvement, with respect to critical data gaps, barriers to participation, and challenges to analysis.

Specifically, we are concerned about four main issues, as set out in summary below.

1. Critical data gaps – Under-counting of at-risk populations

Currently available administrative data appear to leave more people behind than the census, and to undercount some at-risk populations. We acknowledge that ONS has already made substantial progress in improving the coverage of ethnicity in administrative data, and expect that more progress can be made going forward. But since ONS analysis is only at an exploratory stage, it is not yet clear to us how large the current undercounts in administrative data are for other protected characteristics, and how much progress can reasonably be anticipated by the time the 2031 census would be due. We welcome the ONS proposals to develop appropriate programmes of work to reduce the undercounts in the administrative datasets but the very early stage of planning these programmes means that there is no sure method or timeline for avoiding critical data gaps, especially for the characteristics that are not covered in administrative data.

2. Critical data gaps – Lack of validation sources

A further concern about discontinuing the census in 2031 will be the difficulty of validating the accuracy and coverage of the linked administrative data constructed by that date. We welcome ONS proposals for a programme of work to find alternative ways to validate the accuracy and coverage of the linked administrative data. However, this programme is also at an early stage of planning, and a proper method of evaluation in the absence of a census has not yet been established. We are concerned that the ONS might be suffering from what the National Audit Office (NAO) terms ‘optimism bias’, and that policy-makers and data users might be ‘flying blind’ after 2031 with respect to the numbers and geographical distribution of at-risk populations across the UK.

3. Challenges to equality and inclusion analysis – Lack of harmonised standards

A further concern is that administrative definitions will likely evolve over time, reflecting policy changes, rather than being linked to the concepts enshrined in the 2010 Equality Act. There could also be a possible divergence between the data available across the four nations of the UK. Some key outcome measures for assessing the extent of disadvantage of at risk populations may also be unavailable in administrative data sources. These changes and divergences could have important implications for understanding of the situation faced by at-risk populations across the UK, as well as limiting intersectional analysis. We do not consider that these issues are adequately addressed in the FPMS proposals.

4. Barriers to participation – Public trust

The Inclusive Data Taskforce report emphasised the importance of building public trust in statistics and in the bodies responsible for collecting them. Over time the ONS has built an enviable reputation for trust and this is reflected in the very high levels of cooperation that has been achieved with the census. In contrast some at-risk groups have lower trust in some administrative bodies. We believe ONS must consider the wider implications for its work of a move towards a potentially lower-trust statistical framework, including possible spillover effects for ONS of reliance on frameworks that are less transparent and more vulnerable to challenge and to changing political preferences than the census.

Conclusion

We trust that our comments will be of assistance in consideration of the proposals generally, including information on how Canada is addressing issues of inclusivity of data in a range of ways. We consider any future arrangements should seek to enhance inclusion and equality in the UK’s framework of official statistics and are not assured that the proposals achieve this. In summary, the Advisory Committee is concerned that:

• in the proposed admin-based statistical framework there are likely to be critical data gaps, particularly around some of the most vulnerable and disadvantaged communities in the UK;
• for some under-represented communities, barriers to participation in data collection, such as distrust in government, could well be greater in the proposed new framework than they are in the case of the regular census;
• there will be additional challenges to conducting analysis relevant to inclusion and equality because of a range of technical issues such as differing data standards and a range of practical issues such as sample sizes, where sample surveys have to be used to fill gaps in the admin-based data.

We recognise that ONS has proposed various programmes of work to mitigate some of these issues. However, experience suggests that there is a serious risk of reputational damage to ONS if the proposals are implemented before the mitigations have been fully developed and shown to be effective.