The Senselessness of Census Data

In March next year, everyone in Britain will be ploughing through a 32-page census form answering questions about who they are, where they live and what they do. If this £500m venture was of value, companies would clamour to buy the data. Instead, the 10-yearly market research exercise tells neither government nor commerce what it needs to know.

The 56 questions ask about marital status, place of birth, race and religion. They ask about nationality and education; if you live in a caravan or a bedsit; if you work, what as, what hours and whether you supervise anyone. They want the employers' name, address and type of business and how you travel there.

Yet the only question of a mildly financial nature is whether you have a mortgage. We will be specifically told not to give job grade or pay band. There are no questions about consumer loans, investments or if you paying into a pension scheme or receiving income from one. There is not even a query about whether you are self-sufficient or on social benefits.

Government statisticians can spend the next decade devising matrices to show if people who think of themselves as English live in England, if divorced people inhabit detached homes and what central-heating Christians have. But they will never be able to say if the poor lack pensions, whether the self-employed have savings or if the Tube travellers have rent arrears.

And never mind the meaninglessly subjective questions such as: "How good is your health -- very good, good, fair, bad, very bad?" Or the logically silly "Not very well" tick-box to reply to "How well do you speak English?"

It is impossible to join a bingo club or apply for a store card without answering more relevant questions than that.

That's what the Americans think too. The US used to produce a long and complex census form but on 1 April this year there will be just 10 questions, including name, sex, phone number and age. (And as it also asks date of birth, the age question is superfluous.) Instead the US government will do what any company would do -- put detailed questions to sample groups.

Any commercial researcher knows long forms deter people from responding -- despite the state's threat of punishment. About 3 million people may have escaped capture in Britain's 2001 census, making detailed analysis dangerous. Even as a simple headcount of the UK population the exercise is a poor proxy - illegal immigrants are unlikely to fill in forms -- and it will take 18 months to deliver even that headline figure.

Government departments already hold much of the information the census seeks -- not just births and deaths but peoples' pay, benefit payments, dividend income and house sizes. Much could be obtained from banks, business or trade associations without compromising individual privacy. And much is simply not worth knowing -- which is why commerce does not rely on census data.