OPINION2 October 2013

Making sense of the census proposals

Opinion

The Office for National Statistics (ONS) has recently launched a consultation on approaches for providing census information and population estimates. Barry Leventhal weighs the pros and cons of the two options on the table.

Res_4010559_census

The Office for National Statistics (ONS) has recently launched a consultation on approaches for providing census information and population estimates in the future. Two methods are being put forward – an online census and a combined method using administrative data supplemented by sample surveys. Users are now being asked to give their views on these very different approaches. This article aims to give you a “heads up” on the two routes, together with a SWOT analysis for each, the likely “best bet” at this stage and why you should bother to respond.

ONS started to explore and research alternative methods immediately after completing the fieldwork phase of the 2011 Census. The ‘Beyond 2011’ project was initiated in order to examine different and cheaper ways to count the population more regularly in the future.

“ONS has estimated the cost of each approach: £625m per decade for the online census solution vs. £460m for the admin data route. The cost saving from dropping the online census is therefore fairly small – £165m over a decade, or just 30p per person per year”

The Beyond 2011 project and its current consultation are focused on the provision of census information for England and Wales, for which ONS has responsibility. Equivalent projects are underway in Scotland and Northern Ireland, led by their respective census agencies, with collaboration taking place between agencies in order to manage the consistency of statistics across the UK.

The UK is not alone in reviewing its approach to census-taking – over the last decade other countries have been changing the ways that they conduct their censuses. For example, amongst EU nations, a traditional full population survey was carried out by 18 out of 27 countries in the 2000 census round. Of those 18 countries, seven switched to a different method for the 2010 round.

The ONS Beyond 2011 project has investigated a wide range of approaches, including various census-based methods, options based on existing administrative sources and an address register option. The shortlist has steadily been reduced, until now we’re down to just two alternative solutions.

The online census solution

Under the online census solution, we would continue to have a census once a decade, however completion would be primarily online. In the 2011 Census, households were sent paper forms for self-completion and return, with online entry available as an alternative – the outcome was that 16% of all completed questionnaires came in online. For the future, online completion would become the priority mode, with other modes available for the unconnected population.

Other countries have been moving to online census operations – for example, the 2011 Census in Canada was completed primarily online, backed up by paper questionnaires where necessary, and achieved an internet response rate of 54.4%.

The main strength of the online census solution is that it would continue to deliver a rich set of outputs – a detailed picture of the population – for a wide range of topics and geographies down to very small areas and population groups. However, its main weakness would be that, as in the past, the census and its outputs would continue to be on a ten year cycle – which some might consider to be too infrequent, given the current rate of change in other data sources. The principal opportunity of this solution would be to achieve cost savings and efficiencies through online completion; however the primary threat would be the increasing difficulty of achieving a high response rate.

The administrative data solution

The administrative data solution would make use of existing sources, such as the NHS Patient Register and the DWP/HMRC Customer Information System, to produce annual population estimates. As part of the process, ONS would carry out an annual coverage survey, on 1% of the population, in order to adjust the estimates for biases present in administrative data. Much of the research, that’s been carried out by the Beyond 2011 team, has been to investigate the accuracy that may be achieved in producing population estimates via this route – the indications look good.

This estimation process would be supplemented by an annual survey of 4% of the population, in order to capture information on the characteristics of people and households. There are no specific details about this survey or the topics it would cover, however it could be similar to a census survey and would again be conducted primarily online. The results could be used to produce statistics and profiles on areas, mainly at local authority level, possibly combining several years of survey data together.

A number of other countries conduct their censuses using a combination of administrative registers and sample surveys. For example, in the EU, 11 countries used this approach in the 2010 round. The Netherlands has been most advanced in adopting this technique; its methods have evolved over a number of decades – the last traditional Dutch census was in 1971.

The main strength of the administrative data solution lies in its ability to generate top-line population statistics on an annual basis. Therefore changes and trends might be identified more quickly. However, the weakness is that the solution will never produce the level of detail provided by a census, and will never provide small area statistics for census Output Areas or other small geographies. The main opportunity is that the use of admin data could be extended over time, and that the annual characteristics surveys could potentially be more flexible in the questions they ask. The key threat is that this approach is new and untested in the UK – other countries have taken decades to introduce this solution.

Summing up

ONS has estimated the cost of each approach: £625m per decade for the online census solution vs. £460m for the admin data route. The cost saving from dropping the online census is therefore fairly small – £165m over a decade, or just 30p per person per year.

Therefore, on both richness of output and cost of provision, the online census looks like the better option.

However, ONS has demonstrated that administrative data could help to improve their annual population estimates – this work ought to continue, in my opinion, and ultimately be implemented alongside an online census.

ONS needs as wide a response as possible, in order to understand the balance of user preferences – the choice is essentially between detailed data every ten years and top-line data every year.

Therefore all users of census data should carefully consider the options, which are set out clearly in the consultation document, and respond with their views. Responders should tell ONS about the uses and benefits obtained from census data, and the impact that the options would have for them.

Failure to respond could have serious consequences.

The Beyond 2011 Consultation may be found online here.

Barry Leventhal is chair of the MRS Census & Geodemographics Group

3 Comments

11 years ago

People with limited vision should never speak about what will happen in hundred years

Like Report

11 years ago

This is a good summary of the options. The cheaper annual survey option whilst having some adavantages in terms of being more uptodate, will not provide small area data down at output area level even if yearly results are combined and that is a big disadvantage. How will the planners in education, health, housing, transport etc. cope without reliable small area data? It seems to me this is a big price to pay for such a small saving.

Like Report

11 years ago

The tivial cost saving of an electronic census is far outweighed by the historic value of originl documentation in "own handwriting" of the population. Future genealogists would be deprived of these sources and the state lose huge potential income.

Like Report