Radical
Statistics
A community agenda for statistics?
Ludi Simpson
Is there a 'community agenda' for research? How is it different from central and local government agendas? How can it be achieved?
The Bradford Community Statistics Project (BCSP)
The project makes use of new technological developments to present local area statistics in ways which in a few years time will probably be common, but at present are as far as we know still unique. Community development work to encourage confident use of statistics, and an explicit community agenda for statistics, may remain less common but are an important part of our work. BCSP is funded partly by the European Community and managed jointly by Bradford Council and the voluntary sector's Bradford Resource Centre.
The technical content of the project's on-line searchable statistical database for small areas is described first in this article; it is currently developed as a test site within Bradford Council. It has been driven in the first instance by the data and agendas of local statutory organisations (Council, Health Authority, fire service, for example). Once the system is available during the summer of 2000 to specific community organisations, and then by the end of 2000 on the Internet, issues will arise concerning how the project can deal with conflicting agendas between the many agencies involved. Some of these are discussed at the end of the article.
Here is how the project business plan (part of its bid for resources) describes the scope of its work:
- to enable local communities to have accurate evidence, in the form of tangible officially recognised statistics, for funding bids that will attract inward investment;
- to assist the local consultation process to recognise areas of greatest need, identify local priorities, and develop area strategies;
- to inform the development of local and District-wide regeneration strategies; and
- to enable local communities to say how statistical information sources can better meet their needs in future.
Users will log on via the Internet to a map of Bradford. The map becomes a street map when 'zoomed' to small areas. Moving around will be made easier through the ability to search for a postcode or a street name.
Apart from identifying the sites of public services, GPs, sports centres and the like, the user will be able to select statistics for standard areas such as electoral wards and Bradford's neighbourhoods.
A particular innovation has been to allow the user to define their own area. This is achieved by the user tracing their area on the screen with their computer mouse, while the map is on the screen. The quality of the maps is important, so that users unfamiliar with maps and computers will also be able to specify and edit the area that is important to them.
The server then retrieves profiles of the area for:
- Population: age, sex, ethnic group
- Voting rates: in local elections, and change between years
- Adults receiving welfare benefits, and
- Burglaries.
In each case a change between a previous year and the current year is included, as well as a comparison with the District.
Other data that may be added to profiles in the near future includes:
- Children receiving free school meals
- Health: mortality and use of accident and emergency services
- Car ownership, and
- Green space.
The key to having statistical profiles for an area that has not been pre-defined but is defined by the user at the time of the enquiry, is a boundary-free database of statistical information. In BCSP this is achieved by 'spreading' all the statistical data to postcodes and allocating it to a point representing the centre of the postcode. If the data is already postcoded, as some are, then no spreading is required. But much data - for example on population, unemployment, health, green space, and voting rates - are held for larger areas such as Census enumeration districts, electoral polling districts, or postal sectors.
The spreading to postcodes is achieved by a simple version of what is usually called in the textbooks 'synthetic modelling'. The measured characteristics of the larger area, say car ownership, is given to the postcode, but sensitively with regard to other information that is known for each postcode. In the current system, that information is simply the number of addresses recorded on the OS Addresspoint for the postcode. Thus if there are 20 households without a car in a Census area of 100 addresses, a postcode of 10 addresses would be allocated two households lacking a car (2=20*10/100).
When a user requests a profile for their own area they are given statistics based on the summation of all the postcodes in their area. The user's area has to be much larger than a single postcode (the data would not be reliable for such small areas). The minimum size may be chosen as 500 addresses. Their area would overlap a number of different Census enumeration districts, a number of different electoral polling districts, and so on.
What the estimation and re-aggregation of postcodes data has achieved is two-fold:
- the estimation of a profile that includes data from different Census, postal, and administrative 'geographies'; and
- re-estimation from each dataset to the user's chosen area, in a way that is sensitive to the size of the overlaps between the area and each dataset's own geography.
There are many technical issues for development from the first approach based on these procedures:
- how to ensure reliability of the data as provided, which should include measuring that reliability;
- how to improve the indicators used to 'spread' the data to smaller areas;
- what improvement would be gained by making the points single addresses rather than whole postcodes;
- how to maintain the data and its documentation, and
- how to ensure consistent answers to the same question asked by different users.
However, the project is also an organisational and political challenge.
The financial support for BCSP has come from the EU (and therefore the regional government office that approves EU projects), the local council, the Health Authority, and other agencies. Their hope is that community groups will become better able to bid for the resources that exist within government programmes. Does this inevitably make community activists into community managers for government programmes? May it distract those activists from independently defining priorities of their own area, or will it give them the statistical ammunition they need to do so?
The statistical data, its presentation for small areas, and the community development support to understand and access the data, will all allow a rich understanding of local conditions by many more people than at present. But for someone asking 'Why are the conditions as they are?', the answers will rarely be found in local area statistics. They are more likely to be found in economic and political studies of government and business: a local political economy of the sort the Community Development Project (CDP) addressed in the 1970s with great success. Will the project manage to encourage such a broad view of the research needs of communities?
The project is currently protected by current political priorities of community government, partnership between organisations, and open government. These are priorities adopted by both local and central government. Their agreement to share information publicly lies at the heart of the project. But even while these political priorities last - and they may not - there may be tensions that pull in other directions. Councillors may become nervous when they receive well argued cases on the basis of statistical evidence that they have some control over (through finance) but not the time to analyse fully themselves. A willingness to accept critical use of open information may have to be tested. There may also be resistance from local communities, if information on their relative prosperity or deprivation is plainly attached to accessible maps and seen as stigmatisation.
Some have suggested that the statistics would be useful for commercial market research. Will the providers of data see an opportunity to make a sale of data that then prevents them remaining in the public domain?
The project must be concerned about the quality of the data it provides, especially when it is accessible publicly on the Internet. The use of a test site by the community workers should throw up the most obvious unsuitable aspects of the data. An emphasis on quality should remain a concern of all those working with it. The project makes each provider of data responsible for documenting and validating the data provided; this will probably not be the only safeguard required.
One aim of the project is to develop a community, campaign or voluntary sector agenda for statistics that will be relevant to local and national statistics, including the Census. At present the voluntary sector does not have the resources nor the pooled good practice to influence statistical output as do other users.
The project will find it hard to always see research seen as neutral information and analysis. Its information will be useful as intelligence for real actors wanting different and opposing outcomes in local planning decisions of all kinds. One of the main challenges for the project will be to identify the different agendas involved for different actors, and to ensure that the community's agendas are formulated and heard clearly.
Ludi Simpson
City of Bradford Metropolitan District Council
E-mail: ludi.simpson@bradford.gov.uk
