The second option being proposed by ONS is to scrap the census and replace it by a very small 4% annual survey with a reliance on cobbling together various administrative records in order to mimic what a census might have produced. Unfortunately, and confusingly, this second option is being presented as an alternative administrative-records-based census. It would be far better if the word ‘census’ was not corrupted in this way. A census is an attempt at a complete enumeration of the population of an area, all at the same time. Plus the collection of descriptive information about the population in each locality, their households, families, homes and lifestyles; a collection that has been planned and designed, not a by-product of other administrative records.
The reasons for maintaining a census are so numerous that, when you look further into the choices, it becomes clear that scrapping the census entirely is not an option. This is why the small survey of one in twenty five of us, combined with admin records, is being presented as the alternative. But it is not a safe alternative. Both the 1991 and 2001 census revealed that our admin records were including a million people who were not actually here anymore. The 2011 census found half a million extra people that were not here according to the official estimates. The census corrects and finds faults in admin records. Admin records are not a safe replacement for it.
As more and more professional and academic bodies begin to consider the choices being presented by the ONS, more and more people are recognising that the two choices that are being offered need not be the only option. Running the census as it has been run before, even if a little modernised, may not be enough on its own. People are becoming harder to count as many more in the population become increasingly mobile. Far more people are migrating in search of work or are moving between families. Many children of parents who have split up are included equally in both families’ household arrangements so they might easily be double-counted as they are put on two separate census forms.
Many young people from Europe are living in the UK and often also have a home elsewhere on the continent. They can easily be double counted, or it is not realised that they are only here for part of the year. We need to start combining census data with administrative records more cleverly if we are to better enumerate a more complex society and a more mobile population. There are also growing calls for more timely data so that we are not, for instance, planning to build too few hospitals because we have not noticed an influence of our elderly returning from retirement homes abroad. I would suggest that the ONS considers presenting to government the need to hold both a proper modern census in 2021 and also plan to begin an annual survey too. That survey could begin in 2020 so we could see just how much more the full census tells us and also how the countries of the UK are changing in the year up to census day.
Not holding a decent census will cost far more money, dashed dreams, lost school places, overcrowded homes, poor health, poor care and in the end will cost lives. Increasingly people responding to the ONS consolation are indicating that the preferred option is not one or the other, but for both options. What many of those consulted want is a continuation of the census which has run every decade since 1801, other than in 1941. Plus an annual sample survey to update estimates, along with the increased utilisation of administrative records to increase both the accuracy of census based estimates and to widen the scope of what the census enumerates. Combining census and admin data would allow us to avoid asking questions in the census we do not need to ask and would help us ask those questions which admin data has no good answer for.
There is, for instance, no need to ask an income question on the census. Data from the Inland Revenue and Department for Work and Pensions could be used to impute it. However, there might be a need in the future (in 2021) to ask if people have made any personal provision for a pension as the administrative records on that issue are such an awful mess. All we know are total numbers which show that ‘The collapse in pension provision is almost entirely in the private sector. In 1967, more than eight million workers in the private sector had company pensions, but this has fallen to fewer than three million today.'
If admin records from the DVLA were combined with census data there would be no need to ask people how many cars and vans they had access to. We could instead ask how much space they had to park cars in their property, in a garage or on hard standing. That would help us work out just how many people are trying to park vehicles overnight on the public street and how much space there is which could be used for housing, if fewer people had to rely on being able to park a car, or two or three at their home. We are running out of space for housing in many cities. Arguably we have too many cars.
Censuses have always been used to address the key problems of the day. The 1911 census asked women how many of their babies had died. Infant mortality at the time took up to 1 in 10 of the richest of peoples’ new-born infants. The 1971 census asked for details about housing quality, hot running water and toilets. Slums were still being cleared then and the authorities needed to know what was left that was still in poor condition. The 2011 census asked, for the first time ever, how many bedrooms were in every home. It revealed that there were more bedrooms in central London than people. It showed us that we have enough bedrooms to go round, but that we are increasingly sharing out space badly. In short the 2011 census reveals that the rich have been taking too much of our most limited resource: space1.
- 1. Results forthcoming in 'All That is Solid' to be published by Penguin in February 2014, relying very much on recent work by Professor Becky Tunstall of the University of York and her analysis of 1911-2011 census data.
The views expressed in the Opinion section of StatsLife are solely those of the original authors and other contributors. These views and opinions do not necessarily represent those of The Royal Statistical Society.