Select Committee on Treasury First Report

Results of the Census


29. As ONS have noted, the effort and expense of taking a Census are worthwhile only when the results meet needs and are delivered effectively. ONS told us that data users had been consulted on this over a period of years and a number of innovations will be made to the output-production process to improve the products and services available. National and local data will be released concurrently. Virtually all the products will be electronic, with easy-to-use products available free of charge at the point of access.[63]

The response rate

30. ONS estimate that the response rate for the 2001 Census will reach 98 per cent of households; with around 88 per cent of the estimated number of households returning forms by post, a further 7 per cent having their forms collected by hand by the enumerators and of the remainder, possibly 3 per cent being vacant dwellings and second homes. While ONS will not know the precise response rate for individuals until all the forms are processed, they expect it to be as good, if not better, than that achieved in 1991 when the response rate for individuals was 97.8 per cent.[64]

31. ONS said that they had tried to minimise the impact of the fact that certain groups in the population, such as young men and certain ethnic minority groups, are less likely to respond than others. They said that one of the critical lessons they had learned from the 1991 Census was to target their effort and focus their resources on those communities, especially the inner city areas, where they knew they would have problems taking the Census.[65] They also said that changes in society over the last ten years had made things more difficult as there are more properties with entryphones or security systems, more people not at home at any "sensible" time or not at all, and more people living alone.[66] ONS took planned measures designed to improve enumeration in hard to count areas, for example, field-staff workloads (in terms of the number of households to be covered) generally were half the size of those in less difficult parts of the country.[67]

32. ONS told us that the lowest response rates achieved were in the centre of London, but it would not be possible to provide estimates of response by area until the final Census results were available. ONS subsequently provided a note of the 10 areas, shown in the table below, where the postal response rate (the number of forms returned by the Post Office during the enumeration period as a percentage of the previously estimated number of households in each area), were lowest. These ranged from 53 per cent in Westminster (North) to 74 per cent in Wolverhampton, compared to a response rate on this basis of 88 per cent across England and Wales. ONS noted that "these figures need to be treated with extreme caution and cannot be taken as a measure of the success of the Census in these areas, because more forms were obtained by action in the field. ... Areas with lower response rates were the subject of more intensive and longer follow-up than those areas with very high response rates. ..." [68]

Figure 2: Areas with the lowest postal response rates


Postal response rate

Westminster (North)


Kensington and Chelsea (North)


Islington/Hackney (North)


City/Westminster (South)


Tower Hamlets/Hackney (South)


Kensington and Chelsea (South)




Leicestershire East






source: ONS (Ev 56)

33. We asked what effect low response rates would have on Census outputs. ONS told us that the Census Coverage Survey was designed to measure the proportions of non-responding households and people by geographical area. It had been designed to be much larger than in previous Censuses in order to estimate properly differential underenumeration and so enable reliable population estimates by area to be produced. The results will be used to adjust the output from the Census and "the effects of underenumeration on Census output are minimised and the output will be far more consistent than previously."[69]

34. The Market Research Society noted from preliminary information published on the ONS website that the new data collection procedure appears to have had an adverse impact on question response rates. For the majority of questions asked in both 1991 and 2001, the non-response rate had increased from less than 1 per cent in 1991 to 1 to 5 per cent in 2001.[70]

35. We recommend that in their evaluation of the 2001 Census, ONS review the balance of resources devoted to enumeration in the best performing areas and those devoted to the worst and consider what changes may be necessary to the Census in the light of the response rates to individual questions. In view of the comments of the Market Research Society, ONS might also usefully evaluate whether returning forms by post, rather than through enumerators, had any impact on the response rate to particular questions.


36. The first outputs from the 2001 Census (mid-year population estimates for 2001 providing data broken down by age and sex for each local authority area) are due to be published in August 2002. This will be followed by key statistics for areas throughout England and Wales in December 2002 and the main national and local results for local authority standard spending assessments in the first half of 2003.[71]

37. According to the Financial Times, the UK takes considerably longer to publish Census results than Australia, Canada, India, Ireland, New Zealand and the USA.[72] ONS said that they could have produced, as they have done in earlier censuses, a simple count of the number of people in each area earlier. What they plan to produce is the number of people normally resident in an area adjusted for undercounting using the Census Coverage Survey. ONS told us that a significant driver for the timetable was the resource allocation round for local government which takes place in August each year. They had targeted August 2002 for the first results as this was cheaper than condensing the amount of effort into a shorter period. Producing the results a year earlier might have increased processing costs from some £50 million to £65 or £70 million.[73]

38. We raised with ONS the fact that Census results for local authority standard spending assessments would not be available until 2003 and that until then they would be based on data up to 12 years old. Mr Cook told us that in all statistical work there is a trade-off between accuracy and timeliness and availability. In this case the trade-off had been agreed in 1999 when these targets had been set, and people had decided they would rather have a final set of figures in early 2003 rather than a series of numbers earlier which would be subject to some revision.[74]

39. We note that the first results from the 2001 Census will not be available until August 2002 and that the main results will not be available until the first half of 2003, when data on which local authority spending assessments are based will be 12 years old. We recommend that ONS review the trade-off in cost benefit terms of the Census results being available earlier for users and public resource planning against the additional cost of doing so, and publish the results of this exercise. In undertaking this work, ONS should take account of the requirements of all data users, and not just the resource allocation round for local government, which seems at present to be the primary determinant of the timetable.


40. The Market Research Society told us that the welcome improvements in planned outputs following consultations with users had been placed at risk by "an uncharacteristically bad decision" by the National Statistician to introduce additional disclosure control measures. They were concerned that these measures, designed to preserve confidentiality, involving rounding counts and raising the population threshold for output tables, could seriously degrade the quality and usefulness of the Census results.[75] Similar concerns were raised by the Greater London Authority who considered that any adjustments necessary to preserve confidentiality should be made to the database and not to individual tables. It believes ONS should consult further with users about the implications of the major changes to minimum thresholds so that user priorities can be determined.[76]

41. We recognise the conflict between confidentiality and precision in outputs containing small numbers. We recommend that ONS reconsider their decisions on rounding and the minimum threshold for individual output tables in the light of concerns expressed by data users and others in the evidence submitted to the Sub-committee.

63   Ev 9, paras 82-83  Back

64   Ev 6, para 50 Back

65   Q17 Back

66   Q18 Back

67   Ev 7, Q66, paras 68-69 Back

68   Ev 55-56 Back

69   Ev 57 Back

70   Ev 58, section 2.1 Back

71   Ev 9, para 85  Back

72   Financial Times, 18 August 2001 Back

73   Qq13, 69-72 Back

74   Qq86-88 Back

75   Ev 58-59, sections 2.3, 3 Back

76   Ev 51-52 paras, 3.1-3.2, 5.1-5.3  Back

previous page contents next page

House of Commons home page Parliament home page House of Lords home page search page enquiries index

© Parliamentary copyright 2002
Prepared 6 March 2002