Who's counting?

When your 32-page census form drops onto your doormat this March, be thankful that it’s only your type of central heating, number of bedrooms and overnight visitors it wants to know about. It could have been much worse. Among the topics considered but not included this year were fertility, internet access, pet ownership, email addresses, property value and income.

The census is a vast bureaucratic exercise that is becoming increasingly hard to justify. If all the government wants to do is count the population and provide “the main family, social, economic and housing characteristics of persons” required every ten years under EU law, there are much cheaper ways of doing it than recruiting over 35,000 people to collect forms and help people answer questions in more than 50 different languages including Punjabi, Tigrinya and Twi-Fante (no, me neither).

There are also more accurate ways: the methodological flaws in the census are manifold. Young men, non-English speakers and the poor have a tendency to disappear at census time, and those of us who do fill in the form aren’t that careful. Last time, more than one in four of us skipped parts, while 6 per cent of the population was thought to have been missed altogether. Their details were estimated and added later.

Even the follow-up survey in 2001, to test how complete the just-finished census had been, came back incomplete: not surprisingly, it missed the same people the initial census had missed. So another 230,000 people were added; then, later, another 278,000. After all that care, the data then has to be muddled up so that researchers are unable to identify individuals from it.

The cost of this year’s census—£482m—has doubled since the last ten years ago. It’s still cheap compared to the US census, which asks far fewer questions yet costs £8.7bn. Tellingly, the director of the US Census Bureau himself has described it as “close to the edge in terms of usefulness.”

Many experts expect this census to be the last one carried out in this way in Britain too. The Conservative cabinet office minister Francis Maude has said there are better and cheaper ways of doing it in future. Yet all of them quickly run into problems. The Office for National Statistics (ONS) could follow the lead of Scandinavia, and link all of our health, education, tax and benefits records into a super-database at a fraction of the cost. But imagine the outcry from privacy campaigners.

Yet that is exactly what the ONS is considering doing—but quietly. A process called “Beyond 2011” is examining the potential for linking administrative with survey data to produce a national population statistics database. Initially, the plan was to combine census, survey and administrative data (name, address, income, exam results, haemorrhoid operation), and make it available to researchers, government and the private sector. Councils and market researchers, unsurprisingly, loved the idea. But the brakes were applied when concerns were rightly raised that if people suspected the information they provided in the census could be used for multiple purposes, it might put them off participating.

In the meantime, an academic project is collecting information of far more use to many researchers than the census. “Understanding Society,” run by the Institute for Social and Economic Research, is the largest long-term survey in the world, visiting 40,000 households year after year. With the participants’ permission, it already links their answers to their official health and education records. In future, it will link to HMRC data as well, and may even include biomedical information. It will be a huge source of detailed information about our changing society, including much of the information the census wants about step-families and changing living arrangements.

What it won’t include, though, is much about the wealthy, as this group tends not to have the time or inclination to participate in surveys. Yet these are the people who market researchers really want to know about. So are we spending half a billion pounds in order to tell businesses where to site the new Boots or Waitrose? It’s hard to avoid that conclusion. For local authorities, the increasingly detailed statistics provided by the ONS are arguably of greater use than the census because they are continually updated—local council tax bands, life expectancy, crime levels, hospital admissions and benefits claimants. Anyone not visible in this official data is unlikely to fill in the census, either.

The only information that the government gets from the census which it couldn’t already isn’t any of its business: the layout of your house, the details of your commute, overnight visitors on 27th March 2011—and your religion. None of this is asked in the US and, given that 390,000 Britons replied “Jedi” to the religion question last time, this may be a very good thing.