The following article by Robert Heyward,from OSLO News, circulated to local
authority Ordnance Survey liaison officers, may be of interest.
Geography for 2001 Census and Neighbourhood Statistics
(Article for Ordnance Survey Liaison Officers trailed in OSLO News December
2001)
Summary
2001 Census results for small areas are scheduled to be available during January
to June 2003. They will be based on a new geography of 'census output areas'.
Census output areas are also likely to the finest geographic grain for the
collection of neighbourhood statistics. What can you expect ? How might you
prepare ? What is different from 1991 ?
First Sight - One Number
The first results based on the 2001 Census will be the annual 'mid-year
population estimates', to be published by the Office for National Statistics
(ONS) in August 2002. These will give estimates of population in age group for
mid-2001, consistent with the combined findings of the Census taken in April
2001, and the Census Coverage Survey (CCS). The Census and the CCS together
will generate a single output known as the One-Number Census (ONC). During
2002, One Number Census estimates for each local authority district will have
been assessed and quality assured by ONS; and then adjusted to the midyear date
and embodied in the published population estimates. The population base for
Census output will refer to the April 2001 date of the Census and will thus
differ slightly from the mid-year estimate.
Key Statistics for Local Authorities.
The first 'census -badged' output is expected in December 2002; in the form of a
report to Parliament, which will give values of about 500 key statistics for
each local authority district plus regional and national aggregations. All
census tables will be drawn from a database adjusted for non-response by the One
Number Census process.
Initial Statistics for Wards and Census Output Areas
Soon after December 2002, values of the same 500 key statistics will be released
for around 10,000 wards and 200,000 census output areas. Unlike 1991, a
specifically constructed geography will be created for the finest geographic
grain of census output, known as census output areas (OAs). Each census output
area will typically cover about 125 households, with about 200 to 250 residents.
In 2001 the term 'enumeration district' (ED) will have no relevance for census
output. 'Enumeration district' now refers only to a patch covered by one person
in census fieldwork, which varied in size according to the expected ease of
enumeration.
Boundaries of Census Output Areas
A map 'backcloth' or 'underlay' and digitised boundaries will be essential for
making sense of a new geography with over 200,000 zones. Both will be available
along with the Ward and Census Area Statistics through the initiative of the
Census Access Project.
How is the extent of a Census Output Area determined ?
Census Output Areas are constructed around the data recorded in the census.
Census Output Areas will chiefly comprise groups of unit postcodes (such as SW1P
3XX, 3XY, etc.). But OAs will 'nest' within wards, not necessarily within
higher levels of postal geography. Some unit postcodes may be split to respect
ward (and parish) boundaries. Every OA must contain more than a minimum number
of 40 households and 100 residents, so some very small parishes will need to be
amalgamated. Unit postcodes, to be grouped into an OA, must be adjacent to one
another, and together as far as possible will approach the target numbers of 125
households and 200 to 250 residents, and be compact in shape. The postcodes
within each OA will as far as possible be alike in terms of mixture of tenure
and accommodation type, and in 'countryside areas' will try to group urban and
rural parts separately.
The process for creating Census Output Areas is outlined in pages 9 to 11 of an
ONS publication 'Standard Area Statistics Discussion Paper' (SAS-DP-2001)
obtainable via http://www.statistics.gov.uk/census2001/sasdiscussionpaperuk.asp
This describes how the OA production system automatically creates postcode until
polygons around the co-ordinate references of addresses of households and
communal establishments enumerated in the Census, then groups the polygons
through a process of iterative recombination to form OAs which optimise
objective criteria.
ONS are planning a workshop on census output geography, to be repeated at 3
venues in late November and December. Each meeting will allow the opportunity
to learn more and contribute to discussion about the methodologies and issues
regarding the creation of the output areas, the supply of products and services
in an environment of changing and overlapping geographies, and plans for the
future. They will include sessions covering: the new output geography;
application to neighbourhood statistics; dealing with postcode and co-ordinate
references; issues with output areas; output for changed and overlapping
geographies; census geography products and applications; extending the use of
automatic zoning. Details will appear via
http://www.statistics.gov.uk/census2001/georoadshows.asp; or by email request to
[log in to unmask] .
Geography and the Census Access Project
The Census Access project has the dual objective of supplying an integrated set
of statistics, geography, metadata and functionality in user friendly forms
under terms which allow free-in-effect and unrestricted use. All costs are
being met up front by the Government's Invest to Save Budget in combination with
partners, including the LGA, representing the major user sectors.
The Census Access Project will supply Census statistics, the geographical
framework in digital form and some tools for mapping and visualising this
output. Separate products like 1991 'ED-Line' will not be needed to display
small-area boundaries. Much of the Census Access Project will eventually be
delivered through the Neighbourhood Statistics web-site, but the output will
also be available in free-standing form, particularly in the first phase of
delivery. However, each element should be provided in a form suitable for
independent use. The digitised boundaries for OAs will be available in formats
(such as ESRI files) that can be loaded into other mapping packages.
A number of base geographic products are proposed (see pages 14 of SAS-DP-2001)
including digitised boundaries and spot references, such as centroids for OAs;
lookup tables, between unit postcodes and OAs - with headcounts; measurements of
hectares for OAs and associated figures for density or sparsity of population.
Further geographic products may be provided, if there is sufficient demand from
users. These include a more generalised ('lite') set of digitised boundaries,
eg for less demanding display purposes; and boundaries of populated land.
The backcloth / underlay provided by ONS (from Ordnance Survey sources) on the
Neighbourhood Statistics web-site may be a raster product. However, as local
authorities should soon have vector versions of OS products such as Internet
Street Mapping, then they may be able to display census data for their locality
more easily against a vector-generated backcloth. ONS open dissemination policy
(and HMSO click use licences) will allow the re-use and publication of Census
material without charge (see pages 7 and 13 of SAS-DP-2001).
The free availability of boundaries for census output areas through the census
access project places OS products like CodePoint with Polygons in a different
light. OS Codepoint and CodePoint with Polygons will help in generating
locational references and displaying postcode-based information. However
organisations wanting to display census based material should find enough in the
OA-boundaries product from ONS. Further, coding address or postcode material to
OA-zones should be facilitated by other tools provided by ONS to data suppliers
(see Neighbourhood Statistics - below). The co-operation achieved between ONS
and OS is important in bringing about these easier facilities for local
organisations.
Main Statistics for Wards and Census Output Areas
In the first half of 2003, 'Standard Tables' for wards and above, and Census
Area Statistics for all areas will be released. Unlike 1991, tables will be
issued in broad topic tranches; not county by county. There will be over 100
Standard Tables for each ward; and a broadly similar number of CAS tables for
each census output area. There will be fewer items in each CAS table: around
7,000 items in the whole set of tables for each census output area, compared to
over 25,000 items for each ward. In order to prevent the potential
identification of individuals, information will be presented for broader
categories in CAS tables, and finer categories in Standard Tables. For example
one Standard Table has 25 categories for occupation; while the similar CAS table
has just 9.
This is only a brief outline of the bulk of the census output: there are many
other innovations in the type of information presented, compared to 1991.
Further all statistics available at CAS or ward level will also available for
higher levels of aggregation, such as district or ward. Throughout, all the
figures will be estimates rather than 'counts' so as to include the
characteristics of people or households estimated as missing from the main
Census fieldwork. The One Number Census generates these characteristics by a
process known as 'imputation'.
Census Output Areas and Neighbourhood Statistics
ONS proposals for collecting Neighbourhood Statistics will be based around
census output areas. Data suppliers will be asked to summarise specified
indicators to produce a count or a rate for each OA. Coding address or
postcode material to OA-zones should be facilitated by other tools provided by
ONS to data suppliers such as local authorities under the umbrella of the
Neighbourhood Statistics Project, which was given substantial resource from the
capital modernisation fund in July 2001. Further details may appear via
http://www.statistics.gov.uk/neighbourhood/aboutness.asp
Local Studies: Identifying 'Neighbourhoods'
Looking ahead a year or two, what zones will be used for presenting small area
information? In the 1990s electoral wards were the most widely used zones for
summaries / presentation within districts. Electoral wards have distinctive
features - not least the tendency to change each time a district is reviewed by
the Electoral Commission. Wards also differ considerably in size.
Among shire districts, wards sizes range mostly from 2,000 to 7,000 residents.
Among metropolitan districts almost all wards are above 7,000 residents and many
are over 15,000. It is unlikely that large wards contain just one
neighbourhood; and conversely many neighbourhoods may straddle parts of more
than one ward.
However census output areas are too small for the presentation of information
about neighbourhoods, but can be used as bricks to build neighbourhoods and
other zones of interest. A district with 100,000 residents will contain about
400 census output areas. Fine for data collection and assembly, but too many for
ease of presentation; and many census output areas will have much in common with
adjacent areas.
In the coming decade there is an opportunity to devise a more sensitive and
robust set of boundaries for neighbourhoods, based on amalgams of locally chosen
groupings of census output areas. See Bradford Community Statistics Project -
http://www.bcsp-web.org "Maps and Stats" for an example of a district with
neighbourhoods as one of several layers of zones for presenting local
information. This project (working with material from 1991 Census and other
sources) also includes a method of estimating data for small areas selected by
users.
Local organisations working within the framework of local strategic partnership
might consider proposals and then agree boundaries for neighbourhoods to assist
the joint planning of local services. If these neighbourhoods are built as an
amalgam of census output areas, then all the resources of the census and
neighbourhood statistics will be available. In particular, local strategic
partnerships may want to obtain more detailed analyses for neighbourhoods
(similar in content to the standard tables for wards). These will be subject to
ONS disclosure rules, which may prevent separate analyses for 'rival'
definitions of neighbourhoods, which overlap only slightly. So local agreements
will be important in gaining access to these analyses.
Moving Ahead
One of the opportunities for local authorities working with Census data, due
from December 2002 onwards, will be to propose and create appropriate local
geographies for working with partner organisations. The geographic products
provided by the Census Access Project will assist substantially in this.
If you have any comments or questions stimulated by this article, please contact
Robert Heyward, email [log in to unmask]
Scotland
Some of the detail in this article is specific to England and Wales. Readers in
Scotland will need to be aware of variations: for example, see
http://www.gro-scotland.gov.uk/grosweb/grosweb.nsf/pages/censushm and
the Scottish Census Results On-Line (SCROL) Pilot via
http://www.scrolpilot.org.uk .
**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.
This footnote also confirms that this email message has been swept by
MIMEsweeper for the presence of computer viruses.
www.mimesweeper.com
**********************************************************************
|