This statement reports the ONS decision on the disclosure control method to be used for 2001 Census output, following the user consultation carried out in May 2002. Advisory Group Paper AG0203, which formed the basis for the consultation, provides details and a comparison of the two options:
rounding of all counts to base three
small cell adjustment.
ONS has now considered all responses and has decided to adopt option 2, small cell adjustment.
Over 80 responses were received in response to the consultation. The majority of responses expressed a preference for the second option. Whilst expressing these preferences, users also expressed disappointment at the decision to introduce either of these additional measures. ONS recognises this disappointment. The decision to introduce additional measures was a difficult one but one that was necessary to meet the legal confidentiality commitments we have made to the public both in the Census White Paper and on the Census form itself. The decision also reflects the fact that protecting the confidentiality of details about individual people becomes less simple with each Census, as the amount of accessible and publicly available information about individuals has increased. There is also more information that can be matched statistically with the Census, and electoral rolls are now more widely used in electronic form. Alongside this, for the 2001 Census, we are releasing a larger range of small area statistics, notably because we no longer obtain any key measures from just 10 per cent of the population. This information will be more accessible than before and will be available over the Internet.
The decision clearly demonstrates the integrity and the high professional standards that ONS employs in the treatment of confidential information. This is vital if we are to maintain the high levels of response, cooperation and trust we receive from all our data suppliers (including the public) which in turn can only have a positive impact on the quality of the statistics we produce. ONS is the major custodian of statistical and administrative records about people across England and Wales, and much data covers those of the whole United Kingdom. We are trusted to do this because we can guarantee that records we have will only produce statistics that do not identify individuals.
Much has also been made of the effect of the variation added to aggregated census returns through the One Number Census adjustments, question non-response, other respondent errors and migration since the census. However, a large proportion of Census returns are of extremely high quality and for people who completed these forms the arguments that suggest disclosure risks are low will be of least relevance. ONS has a responsibility to protect everyone's data.
Supporters of option 2 preferred option 2 because it affected less cells and ensured that tables would be internally additive. They felt that the main disadvantage of option 1 was that all cells would be rounded and that tables would not be internally additive. The benefit that option 1 had in providing greater flexibility of outputs was considered not great enough to offset the disadvantages.
There was a small number of users that preferred option 1. Users who preferred this option expressed a desire for greater flexibility in the output and were concerned that the impact on a particular count could be large in a small number of cases.
ONS does recognise the concern that adjusted data may appear to be more difficult to analyse and manipulate than unadjusted data and in particular the impact that adjusting small cells will have on origin destination statistics. However, relationships between variables will not be impacted. Strong relationships will still show up in the data and no new relationships will be created. We are prepared to work with users to ensure that Census data is used to its full potential, that users understand the quality of Census data and that it is interpreted correctly.
Decision: Option 2
In making its decision, ONS has listened to the concerns of a wide range of users. It has recognised the needs expressed by the majority of users to adjust small counts only and to ensure that individual tables are internally additive. ONS recognised the concerns raised by supporters of option 1 but felt that option 2 met the needs of the vast majority of users and that the impact on data quality would be minimal.
It is worth re-iterating that ONS will only provide very limited information about the methodology for option 2. There will also be an impact on the level of flexibility that ONS may be able to offer in the future.
Providing advice to Users of Census Data
There is clearly concern over how to make appropriate judgements from adjusted Census data. During the autumn ONS will hold a series of seminars and workshops to prepare users for the release of Census Data and provide advice on how it should be analysed and interpreted to its full potential. It is likely that the seminars will form the basis for further documentation that will be released to all Census users before the main results are published.
Details of the seminars will be published on the National Statistics website and a link will be sent to Census Advisory Group members when these dates have been finalised.