Too soon to scrap the Census - Public Administration Committee Contents


2  Options for the future of the census

6. Every country in the world wants to know the size and characteristics of its population, and the United Nations and other bodies require the data to be collected.[4] But the detail of what is collected and how it is collected varies considerably between countries. Some have a one-off census at a given moment in time, every five or ten years, or sporadically. Others conduct a census in a different part of the countryevery year, covering the whole area in rotation. Some predominantly use administrative records including population registers rather than collecting data on forms. Relatively infrequent censuses are supplemented in many countries by sample surveys or other information to derive more frequent, often annual, statistics. The questions asked, methods of data collection and publication details have evolved in the UK but the fundamental core of the country's population estimates-the ten-yearly census-has been unchanged since its inception in1801.

7. In 2008, the Treasury select committee, which had responsibility for scrutinising statistical matters before this responsibility was transferred to PASC, looked into the adequacy of population statistics in England and Wales.[5] In their Report, "Counting the population", they expressed concerns that there were "substantial problems in generating accurate population estimates in some areas during the 2001 census", and that the 2011 census should be "the last census in the UK where the population is counted through the collection of census forms."[6]

8. The Science and Technology select committee also looked at the census in 2012. Their inquiry considered "the use of data from the census by the Government and whether there were elements of the census that would be irreplaceable by other means if the business of Government would be seriously impacted if census data was lost or changed."[7] The Committee concluded:

    We have concerns that social science could suffer if the census was to be discontinued without serious consideration as to how this data would be replaced. We have been told that surveys and administrative data can be used effectively but we have concerns that this would not easily scale to a nationwide coverage. There would need to be a serious consideration of how administrative data could be brought to a national standard to allow it to be more easily used as a replacement for census data.[8]

However, they envisaged some benefits if the census were to be discontinued:

    However, we anticipate that the absence of a census would also potentially stimulate a considerable amount of innovation in social science and examination of how to produce social data of an equivalent standard, but to much quicker timescales, than the current census data.[9]

9. The Minister for the Cabinet Office, the Rt HonFrancis Maude MP, has said "the census was an expensive and inaccurate way of measuring the number of people in Britain", telling The Telegraph newspaper that the Government was looking for a "fundamentally" better way of doing the census.[10]John Pullinger, the President of the Royal Statistical Society, told us

    [The Minister for the Cabinet Office] starts from the position that the plethora of open data and administrative data ought to make it possible to doa census in a much cheaper and quicker way. We all agree with that. The test is one of practicality now.[11]

10. As it is currently conducted, the census does not in fact count every member of the population. In 2011, the census response rate in England and Wales was estimated by the ONS to be 94%, which implies several million people did not complete the form.[12] It is likely that many of those who did not complete the form were harder-to-reach people-the poor, elderly, non-English speaking or less well-educated in deprived areas, damaging the accuracy of the data in the very places of arguably most interest to policy makers. Policy makers cannot be certain of the accuracy of the data as not all forms are fully completed and even when responses are given, they might be questionable; for example, in 2011, 176,632 respondents to the census described their religion as "Jedi knight".[13]

Office for National Statistics' work on the future of the census

11. The ONS has always had an eye on how the census could evolve or be enhanced. More recently it has been carrying out a programme to look at the options for the future of the census since April 2011: the Beyond 2011 programme. A three-month consultation, launched in September 2013, proposed two options for census taking in future:

·  Once a decade, like that conducted in 2011, but primarily online; or

·  Using existing Government data ["administrative data"] and compulsory annual surveys.[14]

The ONS defines administrative data as that "data already held by Government [which] would be used by ONS to produce an annual estimate of the population in local areas. Sources currently being researched include those held by the Department of Health, the Department for Work and Pensions, HM Revenue and Customs, the Department for Education, the Higher Education Statistics Agency, NHS Wales and the Welsh Government."[15]

12. The ONS set out their analysis of the "strengths, weaknesses, opportunities and risks" of the two options in their consultation document:[16]
Online census once a decade Census using administrative data and surveys
Strengths
Delivers a rich set of statistics for a range of geographic areas.

Proven ability to delivery detailed statistics for small geographic areas.

A familiar and tested approach

High degree of continuity.

Single, high-quality "snapshot" of the nation.

Annual; changes and trends could be identified much more quickly.

Would cost less than an online census.

Would reduce the burden on households.

Weaknesses
Only every 10 years.

Online census costs more than using administrative data and surveys.

Scale and cyclical nature of the census makes the build-up and run-down of the operation challenging.

Imposes a burden on all households.

Would never produce the most detailed statistics available from an online census once a decade.

The date to which the statistics refer would be less clear.

Would not result in the detailed historical record of people and households used by family historians and other historical researchers.

New legislation would be required.

Opportunities
Faster and more efficient data processing if online. It should be possible to extend the use of administrative data to cover topics not covered by the survey.

Potential to be more flexible than an online census in the questions that are asked.

New opportunities for historical research in 100 years' time.

Risks
Increasingly difficult to achieve high levels of response to a census.

Online approach risks excluding some people and households.

Some consider census to be an invasion of privacy.

Would depend on a number of new and partially untested methods.

Would lead to some discontinuities from statistics produced previously.

Would rely on access to the required administrative data.

Would be difficult to achieve the required response rate to the new surveys.

Would require public acceptance of the use of administrative data for statistical purposes-privacy concerns.

The role of Parliament

13. Parliament has an explicit role to play in determining the future of the census. This reflects both the requirements for legislation and the cost of the exercise. The 2011 census cost about £500m for England and Wales, more than double the annual ONS budget. The Telegraph has recently suggested that the 2021 census could cost "more than £800 million in today's money, or £1billion in practice".[17]The ONS explains some of the legislation required: "Existing primary legislation allows a census to take place, but does not require a census.As with the 2011 Census, an online census in 2021 would require Parliament to agree specific secondary legislation, setting out the census date and the questions to be asked for example.A census method based on administrative data and surveys would require Parliament to agree new primary legislation, to enable easier data access for ONS and to make it a legal requirement for households to respond to any new surveys."[18]

The ONS's recommendation on the future of the census

14. The ONS consultation received over 700 responses and a report on the consultation was published in March 2014, together with the consultation responses.[19] Around four in ten the responses were from individuals concerned about the loss of information for family research purposes.[20] Such significant public interest in and support for the census should not be lightly dismissed. The ONS analysed the main messages as follows:

·  there was continuing widespread demand for detailed information about small areas and small populations offered by the decennial census;

·  there was a strong concern that a much smaller annual survey, supporting the use of administrative data, would not meet these needs;

·  more frequent statistics that could be provided between censuses by the use of administrative data and annual surveys would be welcomed, but not at the expense of the detailed statistics;

·  the methods for using administrative data and surveys were not yet mature enough to replace the census; and

·  there should be a hybrid approach, making the best of both approaches.[21]

15. Also in March 2014, on the same day the ONS consultation report was published, the National Statistician set out her recommendation to the Board of the UK Statistics Authority (UKSA) "for the future provision of population statistics and the next census".[22] She recommended that there should be an online census of all households and communal establishments in 2021, and increased use of administrative data surveys "in order to enhance the statistics from the 2021 census and improve annual statistics between censuses".[23]

Increased use of administrative data

16. Many of our witnesses agreed with the analysis in the ONS consultation document that the current system, of a census every 10 years, had many disadvantages. John Pullinger, President of the Royal Statistical Society, expressed the concerns of many around timeliness and cost of the current census. He told us

    It is becoming harder to do a census purely in a traditional way. Sometimes it is going to be necessary to think laterally about whether there are cheaper, certainly, but also more creative ways of counting the population, particularly in complex areas such as Westminster, in a timely way. If it is possible to harness administrative data that is available in real time then so much the better.[24]

17. Juliet Whitworth, from the Local Government Association, said that the administrative data option "seems a really sensible way to go. Local Government would really like more frequent data, if that was possible".[25] However, witnesses also described to us some of the possible disadvantages of the administrative data option too: that it might be difficult to get the kind of "small, granular" estimates at a very local level which the traditional census gives, that it might be hard to compare differences over time, that there is a lack of clarity about what legislation would be needed to enable greater data sharing within Government.[26] Juliet Whitworth told us that "we should not rush into it before we know it is good enough."[27]

Scope of the administrative option

18. Whilst some witnesses were generally positive about the way in which the ONS had carried out the Beyond 2011 programme and the recent consultation, many of them thought that the administrative future option set out in the consultation was "unambitious".[28] Keith Dugmore, founder and Director of the Demographics User Group, which represents the needs of commercial users of information, described it as "a very cautious option".[29] Professor Chris Skinner, from the London School of Economics, who carried out a an independent review of the Beyond 2011 programme methodology, told us

    In some ways, it is a slightly unambitious option, in that it is only looking to the administrative data for population counts-the number of people within age group, sex group, and areas. At this stage, it is not attempting to make user of other information, such as income from tax records, or the administrative data that some users think would make this option much more attractive. It is basically sticking with the administrative data for population counts, and then having this large, rolling, annual survey for all the other information.[30]

19. Professor Mayhew, from City University, said "there is absolutely no doubt that central Government data systems could provide a good population count and even a household count if you did it properly". He cautioned however that "whether you would get all the attributes you needed that you get in the census, I doubt. However, you would also get other attributes that you do not currently get".[31] He outlined his own research work, which he argued demonstrates the potential of administrative data.[32] Keith Dugmore told us "one thing about administrative data [...] is that it can offer us new topics that we do not get at the moment. It is not just a matter of replacing questions in the census".[33]

How advanced are the systems for exploiting administrative data for population statistics?

20. Witnesses thought that much more work needed to be done in order to exploit fully the administrative data already held by Government, and cautioned against ending the ten-yearly census just yet. John Pullinger told us

    We ought to do much better [...] we have got data coming out of our ears in all sorts of places, and we ought to be able to bring them together to create the kind of data that we need from the census. That is tantalisingly close, but it is not there yet and that is the risk.[34]

Professor Skinner agreed, telling us that that

    the research that has been undertake so far has not yet demonstrated that [the administrative option] would definitely be fully reliable in terms of the estimates that would be produced [...] to ditch the census entirely, and to proceed with this administrative option when it has not yet been fully demonstrated, would be reckless.[35]

Sarah Henry, Head of Research and Intelligence at Manchester City Council, said "we are nowhere near replacing the census with administrative data yet".[36] Although Professor Skinner thought that the ONS "have been doing a lot of innovative research on possibilities for administrative data" it was "still at a relatively early stage".[37] John Pullinger agreed, telling us that "the ONS needs to make much more rapid progress with looking at the potential for the administrative data option".[38] Professor Falkingham, from the University of Southampton, was more optimistic; she said she was "fairly confident" that the ONS was tackling the issue with sufficient urgency. She said they "know what the problems are and they are working hard to find the solutions".[39]

21. The Office for National Statistics has not provided detailed information about what data, other than a head count, could be harvested from the various administrative sources. We recommend that the Office for National Statistics lists all the public and quasi-public sources that could be tapped for data, the data that could be forthcoming from them and the administrative, technical and legal barriers to the use of, and ultimately linking of, that data.

A hybrid approach?

22. Several of our witnesses felt that, rather than choosing between either a traditional ten yearly census, conducted primarily online, or using administrative data plus smaller scale surveys, a hybrid approach would be more sensible. This is indeed what the National Statistician, endorsed by UKSA, now proposes.[40] Professor Falkingham told us that "we can draw strength from some of the administrative data to improve the census estimates. Equally, we can draw strength from the census to improve some of our administrative data."[41]

23. The hybrid approach, but one more ambitious than that proposed by the National Statistician would involve much more extensive use of administrative data, to give more frequent and richer information about the population. This could be complemented by smaller scale, more local surveys, to verify the accuracy of the centrally produced data and to add information in key areas that would not be accessible from administration records.

24. The ONS, in consultation with the Market Research Society and others, estimated that the economic benefit to the private sector of continuing with a traditional 10-yearly census, largely online, as £3.67bn over the 10 years, compared to £2.1bn over 10 years for the alternative of moving to administrative data.[42]

25. In one sense the 2011 census was already a hybrid. Responding to the criticism that the census in 2001 omitted some entire blocks or housing developments, the ONS built an address register for use in 2011. Sarah Henry explained that in 2011, the ONS worked with local authorities to help improve the censusaccuracy, using administrative data in addition to the address registers.[43]

26. The use of an address register in 2011 was a very good example of using administrative records to enhance the accuracy of population statistics. Other administrative data was also apparently used. We recommend the Office for National Statistics sets out what data it used in 2011,the impact it had on the resulting estimates, the lessons learnt from this experience and how such additional sources can be used more widely and effectively.

27. Population estimates are of fundamental importance to the statistical system, policy makers and society more widely, but the days of the traditional, ten-yearly, paper-based census are numbered. The Government has a wealth of detailed administrative data which is currently unexploited and which could provide a rich seam of information to improve the nation's knowledge of its population and boost the quality of public services. Data from administrative sources can be richer, broader, cheaper and timelier than the equivalent from a traditional census; it can be made available far more frequently than every ten years. The National Statistician has recently recommended that there should be a census in 2021, albeit conducted where possible online, and that there should be greater use of administrative data and surveys. It is too soon to decide whether to scrap the census. We believe that it is right to have a census in 2021; as insufficient effort has been made in recent years, the alternative options for the collection of population statistics have not been adequately tested and plans are not sufficiently advanced to provide a proper replacement, given the importance of the resulting data.

28. However, in order to get the most use out of the information already held by the Government, for the purposes of high quality and granular population statistics, and before we can be sure that there can be, eventually, a full and proper replacement for the traditional census, much more work must be done. We are concerned that the work on the future of the census has been done in isolation.

29. We recommend that the Office for National Statistics, under strong leadership from the board of the UK Statistics Authority, now scope and set out a more ambitious vision for the creative and full use of administrative data to provide rich and valuable population statistics.The Office for National Statistics should explain how the outputs will be different if administrative data were to be used in place of much of the census, explaining clearly the advantages and disadvantages.


4   See 2010 World Population and Census Programme, website of the United Nations Statistical Division Back

5   Treasury Committee, Eleventh Report of Session 2007-08, Counting the population, HC183-I Back

6   As above, p4 Back

7   Science and Technology Committee, Third Report of Session 2012-13, The Census and social science, HC322, incorporating HC1666-i to -iii Back

8   As above, Summary, p3 Back

9   As above, Summary, p4 Back

10   National census to be axed after 200 years, The Telegraph, 9 July 2010 Back

11   Q71 Back

12   Office for National Statistics, Response rates in the 2011 census, December 2012, accessible at Response and imputation rates Back

13   Office for National Statistics, 2011 census, table QS2010EW "2011 Census: Religion (Detailed), local authorities in England and Wales" accessible at Key Statistics for Local Authorities in England and Wales Back

14   The census and future provision of population statistics in England and Wales, ONS consultation document, September 2013 Back

15   As above, p7 Back

16   As above, pp8-11 Back

17   Telegraph online, Officials ask for access to personal data to improve national census, 28 March 2014 Back

18   Beyond 2011 programme website, Office for National Statistics Back

19   Office for National Statistics, The Census and Future Provision of Population Statistics in England and Wales: Report on the Public Consultation, March 2014 and Office for National Statistics website, Responses to the consultation Back

20   As above. 306 individuals out of 715 total responses (or 43%) stated that their main use of population statistics was for family history research. Back

21   Office for National Statistics, The Census and Future Provision of Population Statistics in England and Wales: Report on the Public Consultation, March 2014 Back

22   Office for National Statistics,The Census and Future Provision of Population Statistics in England and Wales: Recommendation from the National Statistician and Chief Executive of the UK Statistics Authority, 27 March 2014 Back

23   As above Back

24   Q44 Back

25   Q111 Back

26   Q48 [Professor Skinner] , Q53 [Professor Skinner], Q62 [John Pullinger], Q74 [Keith Dugmore], Q93 [Les Mayhew] Back

27   Q111 Back

28   Q48 [Chris Skinner], Q49 [John Pullinger], Q122 [Professor Mayhew] Back

29   Q122 Back

30   Beyond 2011: Independent Review of Methodology, by Chris Skinner, John Hollis and Mike Murphy, accessible atReports and Publications page of Beyond 2011 section of ONS website. See also Q48. Back

31   Q93 Back

32   FOC0002 [Professor Mayhew] Back

33   Q111 Back

34   Q46 Back

35   Q53 Back

36   Q77 Back

37   Q67 Back

38   Q57 Back

39   Q68 Back

40   Office for National Statistics,The Census and Future Provision of Population Statistics in England and Wales: Recommendation from the National Statistician and Chief Executive of the UK Statistics Authority, 27 March 2014 Back

41   Q46 Back

42   Market Research Society response to ONS consultation on the future of the census, December 2013 Back

43   Q96 Back


 
previous page contents next page


© Parliamentary copyright 2014
Prepared 17 April 2014