Censuses have come a long way since the vagaries of Roman population enumeration led to the Christian Nativity story ending in a stable rather than a local maternity unit. In modern times however, they pose a different set of challenges.
In 2010, Cabinet Office Minister Francis Maude decried the UK Census as "an expensive and inaccurate way of measuring the number of people in Britain". The coalition government have since adopted the policy that the 2011 Census, from which new data is being released all the time, would be the last of its kind. But how likely is this to come to pass?
This summer, the Social Research Association hosted "The Census: Now and in the Future", an event aimed at exploring what might happen next. Ian Cope from the Office of National Statistics provided the context by presenting the progress of the ONS response to governmental policy: Beyond 2011, a scoping project to find the UK's Next Top Population Model.
The current unpopularity of the Census is not necessarily a new phenomenon. As event chair BBC News' Mark Easton pointed out, there has been a long held suspicion in Britain of such centralised data collection. But current governmental scepticism seems borne out of political expediency, the desire to seek cost savings at every level of public administration. The major arguments made against the Census in its current form have broadly clustered around two main issues: high, and constantly rising costs; and the infrequency of data collection in an environment of accelerated demographic change.
While some have criticised the Census for failing to move with the times, Mark Easton also identified that it is this exact continuity which is the strength of the Census. While it seems that some form of change is inevitable, the challenge now is to protect quality and comparability with previous data.
A variety of options being explored by Beyond 2011 were presented by ONS, including various permutations of rolling data collection and short-form censuses in combination with more substantial surveys with large samples. However the most likely alternative to the current Census is the linkage of administrative data, coupled with a large rolling survey. This option would involve combining NHS, Department of Work and Pensions, Electoral Roll, Education and Birth & Death data to generate population estimates. While this would produce data at much more regular intervals than currently, this approach would only provide very limited population-wide information, most likely just age and gender.
While ONS have declared themselves pleasantly surprised by the degree of accuracy of the estimates generated while testing this approach, there remain considerable margins of error. Substantial numbers of people are not covered by admin databases, especially those on the fringes of society. The words of a local authority representative were also quoted at the event, that “accurate but infrequent data is more valuable than more frequent but inaccurate data”.
The current Census provides an invaluable benchmark against which almost all other survey data is calibrated. At NatCen we use these methods extensively in both the sample design and weighting of our surveys. Without this benchmark, systematic bias can creep in, undermining survey estimates, regardless of sample sizes.
Danny Dorling, speaking at the Radical Statistics Conference in February, has also highlighted the perils of relying on modelling and estimation. The 2011 Census figures for England and Wales demonstrated there to be almost half a million more people than predicted by the population estimates. Ludi Simpson of the British Society for Population Studies has similarly emphasised the importance of “being able to be surprised”.
Simpson identified another major advantage of the current census, exhibiting the appropriateness of the Local Government Association as the venue for the SRA event, and that is the Census's unique ability to generate accurate and detailed small area statistics. Extremely precise population statistics for small areas are crucial for local government planning and businesses alike, and this flexibility of the data would be unparalleled by alternative approaches. The Census is about much more than just a headcount. This was also identified as potentially undermining the current move towards localism led by Minister Eric Pickles that is at the heart of the Coalition’s strategy. Furthermore, there are significant political and legislative challenges to any type of grand data linkage, and no guarantee that the general public would be any more in favour of such a move than of the current Census.
Beyond 2011 will continue to run until September 2014, when final recommendations will be made to the government. In the meantime ONS are inviting submissions on the future of the UK Census from all interested parties. Despite appearances, the Census's fate seems far from sealed, with many strong arguments in favour of an evolved, rather than abolished, census. Amid the options on the table is a mixed mode, web-first full census, a choice which delivers substantial cost savings while retaining everything most valuable in the current Census.
Censuses throughout history have provided invaluable and unique data sources for countries and civilisations. There is enormous value to the UK Census, for government, local administration, business, social researchers, and ordinary citizens, a value that is often hard to enumerate. While change is inevitable, to ensure that this value is preserved, the rationale should be progress and improvement, not merely cost-cutting. After running for more than 200 years interrupted only by World War, the UK Census deserves better.