The information in this section was first released in September 2002, and amended in December 2002 and March 2003. It is retained here to illustrate developments at particular stages of the Census programme, and may not represent a final position.

Information on the One Number Census issued to accompany Census results
First Census results were published on Monday 30 September 2002. Details of the One Number Census methodology were also provided at the time. However, there was a need for further and more detailed information on aspects of the One Number Census, particularly at the local authority level. The package provided information on:

• the Census Coverage Survey - sample sizes, response rates and dual system estimation components for each local authority
• quality assurance at the national level, including a paper on the dependency adjustment
• quality assurance at the local authority level, including quality assurance charts for each local authority
• the administrative data used in the quality assurance process (these are the comparator data used to calculate the diagnostic ranges found in the local authority charts)
• maps showing the location of postcodes sampled by the Census Coverage Survey

There was no risk of disclosure of personal information through release of these maps, as no information from the CCS was published that could identify personal characteristics.

The One Number Census
A new and more robust method, known as the One Number Census (ONC), was used in the 2001 Census to give a complete estimate of the population in addition to improvements in the conduct of the Census itself, with new initiatives such as community liaison and the post back of forms. The ONC comprised four key elements

1. A large Census Coverage Survey (CCS), conducted independently of the Census and undertaken because response to the Census was likely to be lower, than in the 1991 Census. It covered around 320,000 households.
2. A comparison and combination of the data collected by the Census and CCS, to estimate and adjust all results for Census non-response, resulting in Census data and estimates covering the whole population.
3. A national quality assurance process, resulting in an additional adjustment for dependency, to account for the impact of statistical dependence between the Census and CCS.
4. A comprehensive quality assurance of the Census estimates for every local authority

The Census Coverage Survey
The Census Coverage Survey (CCS) was designed to enable census population counts to be adjusted for underenumeration at the national, local ,and small area level. It consisted of an independent and intensive face to face survey of some 16,000 sample postcodes, containing 320,000 households drawn from all local authorities in England and Wales. An evaluation of the CCS can be found HERE.

Further information is available on:

• CCS sample sizes by local authority.
  
• CCS response rates and dual system estimation components by local authority. (The household response rates achieved by the CCS were updated on 27/10/04. This was to reflect a processing error identified in the previously published response rates for the CCS.)
• areas covered by the CCS and hard to count information.

Adjusting for under-enumeration
The ONC methodology for adjusting the Census database can be found HERE.

Quality assurance - national
Dependency
As part of the One Number Census process, an adjustment was made to account for statistical dependency between the Census and Census Coverage Survey. An overview of this adjustment can be found in this paper (in pdf format). Further technical detail can be found in this paper (in pdf format), currently in final draft form.

Quality assurance themes
During the One Number Census process, it emerged that outcomes for some subgroups of the population were consistently differently from expectations. These were described in commentaries, along with any actions taken to try and resolve the differences and any conclusions drawn:

These commentaries have also been included with the quality assurance charts for those local authorities to which they apply. They were written generically, so give a good guide to the issues faced in these areas during quality assurance, but do not necessarily reflect the actual extent of the issues or action taken for any particular Local Authority.

Quality assurance - local authority
The ONC quality assurance process compared the Census estimates to a range of data from administrative sources.

The illustrative QA pack demonstrates the materials available to the quality assurance panel, whilst the specific charts provided below for each local authority are those that were used during the quality assurance part of the One Number Census process.

Each chart compares the original 2001 Census count with the One Number Census estimates by age-sex group for a local authority district. They also show the upper and lower boundaries of the corresponding 95 per cent confidence interval for the estimate, as well as the upper and lower boundaries of the diagnostic range that the estimate was compared against.

The diagnostic ranges were calculated using the 2000 mid-year population estimates, patient records, pension data, child benefit data, school census information, and approximate 2001 mid-year population estimates. They were used to give an idea of the range within which the Census population count could be expected to fall, the alternative data sources. They were not used as control totals or to give an indication that the Census results were in error. Full details of how the diagnostic ranges were calculated can be found in the ONC Quality Assurance strategy paper, available in pdf format.

The quality assurance process found that the One Number Census estimates were lower than the diagnostic ranges for the males aged between 25 and 39 in about 75 per cent of local authorities. It is the consistency of this population profile, as well as the robust methodology, that gives confidence in its accuracy. It was present across England, Wales, Scotland and Northern Ireland and across three quarters of local authorities, whether they had high or low response rates to the Census and the CCS. The difference between these Census estimates and the previous mid-year population estimates is thought to be principally due to two factors: difficulties in estimating migration, particularly the increase in outward migration; and over-estimation of the adjustments required for these age-groups in the 1991 Census (more information in the commentary, published with the first release of Census results.

The specific comparator data used to calculate the diagnostic ranges are available. These include the 2000 mid-year population estimates, patient records, pension data, child benefit data, school census information and approximate 2001 mid-year population estimates. Many of these sources were 'cleaned' and adjusted to allow for known errors, and so will not be the same as 'raw' figures which may be available from elsewhere. Full details of these processes, and how the diagnostic ranges were calculated, can be found in the ONC Quality Assurance strategy paper, available in pdf format.

Access ONC quality assurance for local authorities
This information is available for all local authorities in England, and for all Unitary Authorities in Wales

This page last updated 01 November 2005