16 May 2008
...does anybody know who's there? The public sector relies on accurate population data for its budgeting and decision-making. But the problems that have plagued past census surveys mean that 2011 could be the last of its kind. Michael Cross reports
March 27, 2011 will observe a tradition going back more than 200 years. It is the next UK Census Day, when, in theory, every household in the realm declares how many people are under its roof, along with data on their origins, health, beliefs and (possibly) income.
Almost certainly, it will be the last such count-up. Among demographic experts a consensus is growing that population statistics based on a ten-yearly household survey no longer reflect the realities of a mobile society especially if the figures emerge a year or two after the count.
Most parts of the public sector use the census figures from the Office for National Statistics. However, the NHS and local government are particularly affected by any shortcomings in the data as their funding is allocated largely by population. Mike Heiser, senior policy consultant at the Local Government Association, says that the census is probably the single most important source of data for local government funding.
The formula grant, which represents half of all non-earmarked funding available to councils, is calculated by the Department for Communities and Local Government largely according to ONS population statistics. Last year, the LGA called on the government to consider new ways of calculating population figures, including taking an interim census every five years.
ONS census data also underpins the largest component of the NHS's operating budget, Department of Health allocations to primary care trusts on the basis of the relative needs of their populations. The weighted capitation formula in use is based on the age distribution of the target population. Health demographics expert Professor Alison Macfarlane of City University, London, says that although the link with funding is not as direct as with local government, a failed census would be 'a huge, huge loss to the NHS'.
Meanwhile, central government's plan for 'transformed' public services, based around citizens rather than institutions, will create more demand for accurate and up-to-date demographic data. Sir David Varney, the prime minister's adviser on transformation, last week contrasted the public sector's one-size-fits-all approach with the ability of companies such as Tesco to use information on segments of customers. 'The public sector is bereft of information,' he said.
In the long term, at least, alternatives are available. One of the last acts of the old Statistics Commission, abolished last month, was to call for the census ritual of visiting every household in the country literally, Domesday book methodology to be replaced by more modern techniques, involving collecting data from operational databases.
However, any such decision would be controversial in the current climate of opinion about the government's competence in securely handling personal data. Neither would it solve the immediate problem of the public services' need for more up-to-date figures on groups undercounted by the ten-yearly census, especially migrants.
The fact that the 2011 census is likely to be the last of its kind places the ONS (and its counterparts in Scotland and Northern Ireland, which carry out separate though simultaneous censuses) under special pressures to get it right. It 'will be the standard against which new ways of estimating the population in the future will be judged', the commission said. In other words, if the last conventional census does not produce credible data, any new methodology will lack credibility from the start.
It is a tall order. The 2001 census, in the words of an official evaluation, was plagued by 'budgetary pressures, public protests, problems with enumeration in hard-to-count areas, poor quality address lists and other technical difficulties'. The ONS's census director, Glen Watson, has warned that the 2011 exercise will need to tackle these difficulties in the face of profound social changes, including a more mobile society and a greater tendency to mistrust officialdom.
Although the big day is still three years away, major decisions about the content and administration of the £500m exercise need to be taken over the next few months. A white paper setting most of the procedures in administrative stone will be published this autumn.
Complaints about the inadequacy of census data have been mounting since 2001. Westminster council estimates that official statistics miss at least 24,000 people in the borough at any one time. One particular category left out is short-term migrants, people in the country for less than 12 months. Registrations under the government's Workers Registration scheme and applications for national insurance numbers suggest that the population is far higher than that shown by figures used to calculate the council's underlying formula grant, equating to a reduction of £12m each year.
If this were not bad enough, demographers say the 2001 census fell well short of ideal in counting the permanent population. Notoriously, it undercounted significant sections of the population in inner-city areas with high proportions of multi-occupied homes and those with high student populations, in particular Manchester.
Three years on, the ONS conceded that enumerators had missed about 15,000 homes, and adjusted population estimates upward by 104,000 people. In 2002, the Commons Treasury subcommittee called for a radical overhaul. Public impressions of incompetence were increased by the resignation of the then census director after a National Audit Office report found evidence of 'serious misconduct' in the placing of a consultancy contract, and the botched 2003 launch of a website holding data from the 1901 census.
In 2001, most of the 'disappeared' labelled the 'missing million' by the press were single people living in hard-to-enumerate accommodation, such as multi-occupied homes. Many were missed because of problems with recruiting the 70,000 census field staff (and paying them on time). In some pockets, poor control of field operations and follow-up meant that coverage fell below 70% of households. Meanwhile, local delays in handling forms posted back by householders meant that unnecessary follow-up visits were made.
Glen Watson says that lessons have been learned. 'Key stakeholders must be engaged earlier and final processes and systems fully tested,' he told the Market Research Society's Future of Geodemographics conference in March.
The 2011 census will also feature several innovations designed to improve the quality of information collected and the use of resources but these might also increase risks.
The hottest current debate is over the content of questions. Census household forms have grown steadily longer, and more controversial, since they were introduced in 1841. The 2001 form contained 40 questions in England and 41 in Wales (with one on the Welsh language). The content of the 2011 form has not yet been decided, but is likely to be substantially longer. In March, Treasury minister Angela Eagle told MPs that the ONS would receive funding to add a fourth page to the 2011 form.
The main focus of the new questions will be to get a grip on migrant numbers. The LGA describes the census's migration statistics as 'not fit for purpose'. Today, the main source of migration demographics is the International Passenger Survey, a voluntary face-to-face sample survey of passengers arriving at or departing from UK ports.
The survey, originally set up to capture statistics on tourism, is widely seen as flawed. Although it interviews more than a quarter of a million people a year, this is only 0.2% of all international movements and an even smaller proportion of migrants.
The political sensitivity of the issue is encouraging ministers to angle the 2011 census more towards migration data. Questions are likely to ask how long people have been resident in the UK and which main language they use (including British Sign Language).
But such efforts might be in vain. Keith Dugmore, of the Demographics User Group, which represents users of data, says that migration questions would be of limited value because of reluctance to respond. He would prefer more use to be made of administrative databases, such as applications for national insurance numbers.
Commercial users of census data are lobbying for a question on income; the General Register Office for Scotland plans to include one in 2011 after a successful test in 2006. However, a similar question is unlikely to be included in England and Wales, because of ONS fears that it might reduce response rates.
Operationally, the 2011 census will be handled very differently to its predecessor, Watson says. However several of the innovations are controversial. These include outsourcing the census operation. A procurement exercise has reached a short list of Lockheed Martin and T-Systems, part of the Deutsche Telecom group. The winner will be announced in June. The outsourcing has already attracted criticism because of the possibility that the US authorities might be able to demand access to data.
Another controversial proposal is to deliver forms through the post, rather than by enumerators on foot. The Statistics Commission warned last November that this presented a two-fold risk. First, reducing contact between enumerators and households tends to reduce response rates. Secondly, everything depends on the authorities having an accurate and up-to-date list of addresses. At present, this does not exist.
'Without a good quality national address register, all the census risks are amplified,' the commission warned.
To tackle these challenges, the 2011 census will try to be smarter than its predecessors. The ONS plans to rank areas into five different 'hard to count' categories. These will be based on factors such as the type of housing stock and data from the Department for Work and Pensions on numbers of residents claiming benefits. The next census is also likely to use more information technology, to manage the field force and offer an option of completing forms online.
However, new technology can create its own headaches. Earlier this year, the US Census Bureau gave up plans to issue field staff with handheld terminals for its next census when the cost of each unit rose above $8,500.
The hardest category of risk to handle might be political. Censuses everywhere attract the attention of political and lobbying campaigns. In the UK, these have included campaigns on national identity, council tax (the 1991 count coincided with the poll tax protest), disability and religion. The fear is that the 2011 count will be held against a background of the introduction of identity cards and the roll-out of electronic medical records.
'Privacy campaigners may latch on to the census simply because it asks a lot of questions of individuals,' the Statistics Commission warned. It recommended early engagement with campaigning groups, noting that government statisticians, on their own, are not the best people to communicate the census's benefits.
The commission also echoed the Treasury subcommittee in calling for the 2011 census to be 'the last census of the current kind'. Beyond 2011, there are several options. One possibility is a 'short-form' census, backed up by comprehensive surveys of a sample of households. Another is a rolling census of comprehensive surveys.
The most dramatic change would be to abandon form-filling entirely, in favour of a regular snapshot of databases such as the DWP's customer information system and the newly computerised central registers of births and deaths. In 1990, Finland became the first country to carry out a census in this way. Statistics Finland now compiles information from some 30 databases every five years.
Some continental countries, such as Germany, the Netherlands, Sweden and Norway, maintain rolling censuses simply by downloading data from national population registers.
A national data-match would cost a fraction of the cost of a conventional household survey but would certainly attract controversy as the next step towards a 'big brother' state. Eagle told the Treasury subcommittee in March that she was 'agnostic' about whether the 2011 survey would be the final UK census. While the identity card scheme 'would provide us with a national register in due course' it was 'far too early' to say what part it would play in future censuses.
Alison Macfarlane says there are pros and cons of both a traditional and a database census. While a census based on administrative data might catch more of the population, the quality of data would suffer. 'You wouldn't be directly asking people questions, and that would be a loss.'
In practice, the most likely choice for 2021 is a combination of administrative data with household surveys. But work on whatever system is chosen will have to proceed in parallel with work on 2011, placing yet more strain on resources. In the meantime, the LGA says that improving migration statistics should progress 'with greater urgency', with better use made of administrative databases.
Heiser says that whatever decision is taken about the future of the census, the sources of ONS data must be as fair and as transparent as possible.
Meanwhile, there will be a practice run for 2011, with census rehearsals taking place in 2009 in Lancaster, the London Borough of Newham and the Isle of Anglesey. In the rehearsal, 110,000 households will receive questionnaires by post or hand delivery, with temporarily recruited enumerators following up households that do not respond. Although the experience will allow the ONS to fine-tune operational procedures, it will by then be too late to change any fundamental details.
Officially, the verdict on whether 2011 is a success will come in 2012, when the first outputs become available. The full results are due for publication the following year. But if the last count-up turns out to be a disaster, the ONS and its political masters will know about it unofficially well before then. Censuses might grab the headlines only once a decade, but those headlines are big ones.
Countdown to 2011
2004 ONS proposed design for Census 2011
2005 Harmonisation across UK agreed
2006 Census test in Scotland. Test questionnaire published for England and Wales
2007 Census test England and Wales
2008 Test evaluation published. 2011 questionnaire finalised. Ministers to approve white paper
2009 Census rehearsals in England and Wales and Scotland
2010 Census regulations come into force. Field staff recruitment begins. Address check
2011 March 27, Census Day
2012 First results reported (autumn)
PFmay2008
