Census Data (Hansard, 25 January 2002)

§ Ruth Kelly

The information requested falls within the responsibility of the National Statistician. I have asked him to reply.

Letter from Len Cook to Miss Anne McIntosh, dated 25 January 2002: As National Statistician and Registrar General for England and Wales I have been asked to reply to your recent questions to the Chancellor of the Exchequer asking:

(i) what estimate has been made of the impact that raising of the minimum threshold for the publication of census data from 50 persons/20 households to 100 persons/40 parishes (sic) will have on the planning of services in rural areas (30193);
(ii) what representation he has received concerning the availability of 2001 Census data at parish level (30156);
(iii) what representations he has received on the impact of the raised threshold in the 2001 Census on planning services in rural areas (30155):
(iv) if he will consider reverting to the 1991 Census threshold of '50 persons per household' (sic) (30157); and
(v) what measures the Office for National Statistics will take to assist authorities responsible for local planning where separate census data for individual parishes has been lost by the amalgamation of figures as a result of the increased minimum threshold (30194).

1151W I will respond generally while covering the issues raised in each of the questions. Following a recent ONS Review of the measures to protect the confidentiality of personal information provided in the 2001 Census, I have announced some additional measures. These include a decision that standard statistical output (amounting to some 10,000 counts) will not be released for small areas below a certain population threshold, this being 100 resident persons and 40 resident households. The design of output areas from the 2001 Census will, however, recognise the need to identify parishes separately in the 2001 Census, but the number of parishes that fall below this threshold will not be known until the processing of data is completed later this year. From information from the 1991 Census, however, it is estimated that some 1,230 parishes in England would have fallen below the 2001 Census threshold. Nevertheless, the average size of output areas in the 2001 Census will be significantly smaller than in the 1991 Census, facilitating greater geographical flexibility overall in using Census output. The resulting increased amount of detail to be made available from the 2001 Census will assist in the better planning of local services. I have listened carefully to the views expressed throughout extensive consultation on output requirements from the Census. The Review was fully aware of, and considered carefully, all the requirements for output, including those for parishes, that users had articulated during the course of wide and extensive consultation with all local authorities and other users. Following the initial announcement last November of my decision to introduce these additional disclosure control measures, I received a number of representations including six from local government bodies expressing particular concerns about the impact of the raised threshold on the availability of 2001 Census data for parishes. In addition, I have recently answered two Parliamentary Questions from the right hon. David Curry MP (nos. 28863 and 28862) on the subject. I am, of course, keen to ensure that users will obtain the maximum benefit from the use of 2001 Census data. To this end, I invited representatives from the census-user community to a meeting on 13 December 2001 at which we were able to discuss further the specific issues that users had raised including the particular issue about sub-threshold parishes. At that meeting I explained that as Registrar General and National Statistician, I have an obligation not to reveal information collected in confidence in the Census about individual people and households. Indeed, I have given public assurances that I will not do so. In presenting increasingly detailed results from the Census, protecting personal information is of paramount importance. Maintaining the confidentiality of individual data underpins the trust that exists between data suppliers and any agency that acts as custodian of information about them. At ONS we are fortunate that businesses and the public have confidence that their information is securely held and that we do not release any data that could identify an individual. It is essential that this trust be maintained. The 2001 Census Discussion Paper on Standard Area Statistics (September 2001) noted that the confidentiality threshold being proposed at that time was 50 persons and 20 resident households, but that disclosure control measures were under review as part of an ONS-wide programme of methods and quality. The recent Review was conducted to explore whether or not the measures that were in place would maintain this trust. The Review considered whether or not the measures initially proposed protected against both real and perceived disclosure. In particular it considered:

(a) the level of statistical detail in the proposed tables;
(b) the level of geography for which the tables would be produced;
(c) the increased availability of the data; and
(d) the proliferation of other datasets against which Census data could be matched.

On the basis of this Review I decided that an additional measure of rounding all cells to either zero or a multiple of 3, should be introduced for all tables to be produced for England and Wales, and that the population threshold for Output Areas be raised to 100 resident persons and 40 resident households for the release of Census Area Statistics. These measures, together with a data modification measure ('record swapping') already announced, will protect against not only the real risk of disclosure but also, and equally as important, the risk perceived, by the very public about whom the information refers, of disclosure that may arise in output table cells containing counts of '1'. Without these additional measures the risk of perceived disclosure would have been substantially higher. Once formed, any such public perceptions would be difficult to dislodge and would, I believe, damage the integrity of all census output. In the circumstances I do not propose reverting to the threshold adopted for the release of small area statistics in the 1991 Census of 50 persons and 16 resident households. We agreed at the meeting on 13 December, however, that the ONS and those present representing the user community would work closely together to minimise the impact that the rounding and increased threshold will have on the data and its uses. In particular, the ONS will consider further what summary output for 'sub-threshold parishes'—in the form of simple statistical profiles, for example—may be safely produced from the 2001 Census, in addition to some basic headcounts that have already been agreed. A note on that meeting has been posted on the ONS website, and a paper giving further background into the decision and the method, and addressing the consequences and issues raised at the meeting and elsewhere, will be produced shortly.