2015年经济学人 反对人口普查 十年一度制度被保留（在线收听）

Official statistics

Con census

Britain's decennial population count has been saved. Now make it work better

WHEN the first results of the 2011 census were published, almost two years ago, the most striking discovery was straightforward.

Britain's population, it turned out, was around 500,000 bigger than statisticians had thought.

For all that Britons are relentlessly surveyed, only a full count could reveal the size of the country,

yet alone that of individual cities, towns and neighbourhoods.

Given that finding, the proposal made in 2010 by an austerity-minded coalition government to scrap Britain's decennial census seemed odd.

It now looks as though the census will survive after all.

On March 27th the Office for National Statistics (ONS) concluded that a full population count is still necessary,

though many people will in future be asked to fill in the census online.

In all likelihood, the government will follow its recommendations.

Academics, market researchers and social historians breathed a sigh of relief.

Scrapping the census in favour of rolling surveys of a portion of the population was always a silly idea.

A full count can provide information at the level of a single street, which even very large surveys could never do.

The census also supplies detailed data about poorly understood groups of people, such as some ethnic minorities.

Each year billions of pounds of government spending are allocated according to estimates derived from the census.

Its data determine where new schools and hospitals are built, where planning permission for housing is granted and where money is spent on transport.

The census is hardly perfect. It is expensive—the 2011 edition cost around 480m (800m).

Its findings go out of date quickly. And it is increasingly difficult to conduct,

as a rising proportion of people do not fill in their forms.

In many European countries, including Germany and the Netherlands,

the authorities use administrative data collected by public bodies in place of a traditional census.

In Britain that would be tricky. The nation lacks a central population register or an identity-card system that would allow administrative data to be linked up.

By northern European standards, Britain also has lots of irregular migrants who have little interaction with the state.

That makes a full count unavoidable.

Yet the authorities should still be investing in working out how to use administrative data better,

argues Chris Skinner, a statistician at the London School of Economics.

That would help to provide a check on the accuracy of census and survey data,

as well as providing timelier and more precise estimates in between census years.

The ONS agrees—and it is researching the possibility of using government data better.

The trouble is that doing so will mean spending money,

which is precisely what ministers wanted to stop doing when they pushed the ONS into thinking about cancelling the census.

Data users will be hoping that they have a change of heart. If not, Britain might be stuck with inadequate numbers for decades to come.