Census figures can no longer be relied onBy Simon Briscoe, Statistics EditorPublished: July 8 2004 18:46 | Last Updated: July 8 2004 18:46

The government statisticians' latest set of changes to the population estimates derived from the 2001 census, announced on Thursday, came as no surprise to researchers using the data. As the figures have been subjected to more analysis since their initial release in September 2002, it has become clear that all is not well.

Jayne Mills of Bristol City Council said at a Local Authorities Research and Intelligence Association conference, that Bristol's population fall contradicted their understanding of the city's dynamics with knock-on effects for service planning and performance indicators. Recorded crime, for example, had risen by nearly 20 per cent due only to the population change. She blamed bad enumeration and an over-reliance on imputation to plug the gaps in enumeration. Bristol had its population revised up by 5,800 on Thursday due, according to the ONS, "enumeration issues" in three local authority wards.
Lindsay Brook, a social researcher who was a census enumerator in London's borough of Hackney, said it was "an absolute nonsense for inner city census results to be taken seriously". He said enumerators did a "difficult job in terrible conditions" with little support from their managers. Language problems - many people in the area do not speak English - and the mobility of the population - few people felt like filling in forms - meant that the census in his area was a "shambles".

A BBC radio programme, "Where have all the men gone?", exposed the casual attitude of many people, especially the young and non-British, to filling in census forms. The failure was highlighted in a study of the unemployed in Hull, where three inner city wards had nearly 400 more unemployed benefit claimants than jobless working age men recorded in the census. The number of claimants would not normally be lower as not all unemployed people can claim benefit.

A researcher in Hull Council said: "These are deprived inner city areas with cheap rented accommodation traditionally attractive to unemployed young males who may be the least likely to have returned a form." The conflicting conclusions from the data means it is "impossible" to establish firm foundations for policies to increase employment rates in the most deprived areas. Hull's population was raised by 6,600, (2.6 per cent) on Thursday, in recognition of the errors in adjusting for undercount.

Paul Bivand, head of research at the centre for Economic and Social Inclusion, thinks many of the missing men - up to one-third of a million - could be partners of lone parents. He says the benefit system is unfriendly to families depending on a male breadwinner with unstable employment due to significant "waiting days" and administration delays for benefits.

For families who need income and rent payment, a claim for income support on grounds of lone parenthood provides a much more stable income flow with earnings by the male partner being an addition whenever they occur. "Men in this position will be used to 'disappearing' whenever anything official arrives, or anyone official such as an enumerator."

John Hollis, census expert at the Greater London Authority, said that the enumeration in some areas of London "was very poorly run - it just didn't work". He has, however, been impressed by the efforts of the ONS in the reconciliation exercise that led to Thursday's revisions and is hopeful that the next population count scheduled for 2007 will be better.

Most of the census data have been released but some have not. The micro data of anonymised records which are heavily used by researchers have still not been published, despite being expected a year ago.

http://news.ft.com/