An area classification is used to group together geographical areas

2001 area classification of Super Output Areas and Data Zones
Guidance notes for the 2001 Area Classification
of Super Output Areas and Data Zones
February 2008
Page 1
2001 area classification of Super Output Areas and Data Zones
What is the Super Output Area and Data Zone area classification showing?
This document describes the locations and attributes of each of the supergroups which
make up the Super Output Area and data zone area classification.
An area classification is used to group together geographical areas according to key
characteristics common to the population in that grouping. These groupings are called
clusters and are derived using data from the 2001 Census. Area classifications for the
UK for local authorities, wards, output areas and health areas were created following
the Census. Area classifications have a range of uses in both the commercial and
public sectors and generally these uses aim to identify particular groups of people who
are most likely to exhibit particular behaviours.
ONS has published an area classification of Lower Layer Super Output Areas
(LSOAs, in England and Wales), Super Output Areas (in Northern Ireland) and data
zones (in Scotland) to add to this suite of area classifications 1 . Super Output Areas
(SOAs) and data zones (DZs) are a relatively new geography designed to improve the
reporting of small area statistics in the UK. The beginner’s guide to UK geography
provides a useful description of these geographies and others used in the UK.
The Super Output Area and data zone (SOA/DZ) area classification, like the other
area classifications published by ONS, has been constructed by creating a hierarchy of
clusters, which together typify the characteristics of an area. There are three layers of
the classification which make up the hierarchy. These are:
• Supergroup – the first layer
• Group – the second layer
• Subgroup – the third layer
There are seven supergroups, 20 groups and 53 subgroups which make up the
classification. The table below shows how each of the supergroups are divided into
groups and subgroups:
1
Scottish data zones have been classified to provide UK consistency, however the methodology does
not work as well in Scotland. Within Scotland, the Scottish urban rural classification or Scottish Index
of Multiple Deprivation should be used as the primary small area classification
Page 2
2001 area classification of Super Output Areas and Data Zones
Table 1: Division of supergroups, groups and subgroups
Super Supergroup Name
Group Group Name
ID
ID
1
Countryside
1.1
Countryside communities
1
Countryside
1.2
Rural economies
1
Countryside
1.3
Farming and forestry
2
Professional city life
2.1
Educational centres
2
Professional city life
2.2
Young city professionals
2
Professional city life
2.3
Mature city professionals
3
Urban fringe
3.1
Urban commuter
3
Urban fringe
3.2
Affluent urban commuter
4
White collar urban
4.1
Well off mature households
4
White collar urban
4.2
Young urban families
4
White collar urban
4.3
Mature urban households
5
Multicultural city life
5.1
Multicultural inner city
5
Multicultural city life
5.2
Multicultural urban
5
Multicultural city life
5.3
Multicultural suburbia
6
Disadvantaged urban communities 6.1
Struggling urban families
6
Disadvantaged urban communities 6.2
Blue collar urban families
7
Miscellaneous built up areas
7.1
Suburbia
7
Miscellaneous built up areas
7.2
Resorts and retirement
7
Miscellaneous built up areas
7.3
Urban terracing
7
Miscellaneous built up areas
7.4
Small town communities
The names given to the supergroups, groups and subgroups are labels, rather than
definitive descriptions of all members of the clusters. These names are average labels
defined by examining the average features of each cluster. As a result it is possible
that isolated SOA/DZs will not be accurately reflected by the name of their cluster.
Figure 1 shows how each supergroup is spread throughout the UK. The countryside
supergroup covers the most surface area of the UK, with other supergroups such as
professional city life and multicultural city life located in major cities such as London,
Birmingham, Leeds, Edinburgh and Cardiff. Members of the urban fringe supergroup
are concentrated on the outskirts of major towns and cities, particularly in the south of
England and members of the white collar urban and miscellaneous built up areas
supergroups are located throughout the UK.
Page 3
Subgroups
a,b,c
a,b
a,b,c,d
a,b
a,b
a,b,c,d
a,b
a,b
a,b,c
a,b
a,b,c
a,b,c
a,b
a,b,c
a,b
a,b
a,b,c,d
a,b
a,b,c,d
a,b
2001 area classification of Super Output Areas and Data Zones
Figure 1: Map of the Super Output Area and data zone area classification at
supergroup level
Page 4
2001 area classification of Super Output Areas and Data Zones
Supergroup 1: Countryside
This supergroup comprises three groups:
• Countryside communities (three subgroups)
• Rural economies (two subgroups)
• Farming and forestry (four subgroups)
Despite the map in Figure 1 showing that the countryside supergroup dominates the
UK, only 14 per cent of the population lived in an area classed as a countryside
supergroup in 2001. More people from South West England lived in this supergroup
than other regions, although this is the most common supergroup in which to live in
Northern Ireland.
Countryside communities is the most common group in Northern Ireland and rural
economies is the most common group in South West England. More people in the
East of England and the East Midlands live in the rural economies group than any
other group. More people live in the farming and forestry group in the East of
England than in any other region of the UK.
Figure 2 shows the topography of Carlisle 007E, a typical 2 LSOA of the countryside
supergroup.
Figure 2: Carlisle 007E
Carlisle 007E is a sparsely populated area covering part of Wetheral and the hamlet of
Cotehill. A much higher number of people live in detached houses, work from home
or work in agriculture and fishing than the national average. In addition, far fewer
2
The most typical Super Output Area or data zone in each supergroup is that which has the smallest
Euclidean distance to the cluster centroid. Please see the methodology document here which describes
how the area classification was constructed.
Page 5
2001 area classification of Super Output Areas and Data Zones
people use public transport to get to work and considerably fewer people live in flats
than nationally.
Page 6
2001 area classification of Super Output Areas and Data Zones
Supergroup 2: Professional city life
This supergroup comprises three groups:
• Educational centres (two subgroups)
• Young city professionals (two subgroups)
• Mature city professionals (four subgroups)
As the name implies, SOAs and DZs in this supergroup can be found in most UK
cities, with the largest concentrations in London, Birmingham and Manchester. Nearly
eight per cent of the UK population lived in this supergroup in 2001, making this the
supergroup which contained the fewest people. As might be expected, more people
live in the young and mature city professional groups in London than in any other
region of the UK. However Scotland is the most common area in which to find people
who live in the educational centres group, particularly around Glasgow, Edinburgh
and Aberdeen.
Figure 3 shows a typical LSOA in the professional city life supergroup, Kingston
upon Thames 009A.
Figure 3: Kingston upon Thames 009A
This LSOA covers several buildings of Kingston University in London as well as
Surrey County Council’s County Hall. Like this LSOA, the professional city life
supergroup has a far higher population density, percentage of students, proportion of
people who use public transport to get to work and proportion of people with a higher
education qualification, amongst other characteristics, than the national average.
Conversely, while flats are very popular in this supergroup, the proportion of detached
houses is far lower than nationally, as is the proportion of people who work in routine
or semi-routine occupations.
Page 7
2001 area classification of Super Output Areas and Data Zones
Supergroup 3: Urban fringe
This supergroup comprises two groups:
• Urban commuter (two subgroups)
• Affluent urban commuter (two subgroups)
16 per cent of the UK population lived in the urban fringe supergroup in 2001. As the
map in Figure 1 testifies, more people who live in this supergroup live in South East
England than in any other part of the UK. Likewise, this is the most common
supergroup in South East England, and can be found, as the name implies, on the
edges of towns and cities such as London. More people in the urban commuter group
lived in North East England than in any other region of the UK and South East
England is the area in which to find people who live in the affluent urban commuter
group.
Figure 4 shows a map of Birmingham 003D, a typical LSOA of the urban fringe
supergroup.
Figure 4: Birmingham 003D
Birmingham 003D is a suburban, residential LSOA in Sutton Coldfield near
Birmingham. Typical of this supergroup, the proportion of terraced housing and flats
is far below the national average, as is the proportion of households without central
heating and the proportion of households who rent their home from the council or
housing association. Conversely the proportion of detached houses is much higher
than nationally.
Page 8
2001 area classification of Super Output Areas and Data Zones
Supergroup 4: White collar urban
This supergroup comprises three groups:
• Well off mature households (three subgroups)
• Young urban families (two subgroups)
• Mature urban families (three subgroups)
At 19 per cent of the UK population, more people lived in this supergroup in 2001
than in any other and SOAs and DZs in this supergroup are spread throughout the UK,
from London to the Outer Hebrides. This is the most common supergroup in the East
Midlands, Yorkshire and The Humber and Wales. However more people who lived in
this supergroup lived in South East England than anywhere else, solely because more
people lived in South East England than in any other region of the UK, regardless of
supergroup.
People who were in the well off mature households group were more likely to live in
North West England, those in the young urban families group were more likely to live
in South East England and those in the mature urban families group were more likely
to live in the East Midlands. Well off mature households is the most common group in
Yorkshire and The Humber.
Charnwood 005B is a typical LSOA in this supergroup (Figure 5).
Figure 5: Charnwood 005B
Charnwood 005B is on the edge of the village of Shepshed, near Loughborough.
There is little to distinguish the white collar urban supergroup from the rest of the UK
and no characteristics used in the creation of this area classification are either far
above or far below the national average for this supergroup. The age structure is very
similar to the UK average age structure as is average household size and the types of
occupation undertaken by those living in the supergroup.
Page 9
2001 area classification of Super Output Areas and Data Zones
Supergroup 5: Multicultural city life
This supergroup comprises three groups:
• Multicultural inner city (three subgroups)
• Multicultural urban (two subgroups)
• Multicultural suburbia (three subgroups)
Nine per cent of the UK population lived in this supergroup in 2001 and SOAs and
DZs in this supergroup can be found in large cities in the UK, particularly in London,
Birmingham and Manchester. Unsurprisingly the majority of people in this
supergroup can be found in London and it is the most dominant supergroup in the
city, particularly in north east, and parts of south and west London. Because of this
strong London connection, it is also the region in the UK with the most people in each
of its groups, although the West Midlands contains more people than London in two
of the eight subgroups in the supergroup.
A map of Enfield 035B, a typical LSOA in this supergroup is shown in Figure 6.
Figure 6: Enfield 035B
As the name of the supergroup implies, this supergroup contains much higher
proportions of people of Black Caribbean, Black African, Indian, Pakistani and
Bangladeshi ethnicity 3 than the UK average. Population density is high, as is the
number of people per room, the proportion of unemployed and the proportion of
households who rent their home from the council or housing association. The majority
of dwellings in the multicultural city life supergroup are either flats or terraced houses
and the proportion of detached houses is particularly low. Similarly far fewer
households have two or more cars and far more people use public transport to get to
work than nationally.
3
Ethnicity refers to the National Statistics classification used in the 2001 Census
Page 10
2001 area classification of Super Output Areas and Data Zones
Supergroup 6: Disadvantaged urban communities
This supergroup comprises two groups:
• Struggling urban families (two subgroups)
• Blue collar urban families (two subgroups)
16 per cent of the UK population lived in this supergroup in 2001 and members of this
supergroup can usually be found on the outskirts of towns and cities in the UK. This
is the most common supergroup in Scotland, particularly in the central belt between
Glasgow and Edinburgh. More people lived in this supergroup in North East England
than in any other supergroup and the struggling urban families group was the most
common in North West England. People who lived in North East England or Wales
were more likely to have been found in the blue collar urban families group than in
any other, although the largest number of people in this group lived in Scotland.
A typical LSOA in this supergroup is Salford 004C. Figure 7 shows a map of the area.
Figure 7: Salford 004C
Salford 004C is an area on the outskirts of the town of Walkden, between Bolton and
Salford in North West England and contains a lot of semi-detached and terraced
houses.
There are several characteristics which distinguish this supergroup from the national
average. The proportion of lone parent households, households who rent their homes
from the council or a housing association, the percentage of unemployed and the
percentage of people with a limiting long term illness are all much higher than the UK
average. Conversely the proportion of people with a higher education qualification,
the proportion of households with two or more cars and the detached houses are all far
lower in this supergroup than nationally.
Page 11
2001 area classification of Super Output Areas and Data Zones
Supergroup 7: Miscellaneous built up areas
This supergroup comprises four groups:
• Suburbia (four subgroups)
• Resorts and retirement (two subgroups)
• Urban terracing (four subgroups)
• Small town communities (two subgroups)
18 per cent of people lived in the miscellaneous built up areas supergroup in 2001. Its
members can be found on the outskirts of cities and towns, and in villages throughout
the UK. This is the supergroup in which most people lived in the West Midlands, East
and North West of England. However more people who lived in this supergroup lived
in South East England than anywhere else, solely because more people lived in South
East England than in any other region of the UK, regardless of supergroup. South East
England also had the largest concentration of people in all groups with the exception
of urban terracing, where North West England had the largest concentration.
Figure 8 shows a map of Ashford 004E, a typical LSOA in this supergroup.
Figures 8: Ashford 004E
Ashford 004E is an LSOA in central Ashford in Kent which is partly residential and
partly industrial. It also contains a veterinary hospital and a primary school and most
houses are either semi-detached or terraced. No variables used in the classification are
either far above or far below the national average. In fact, the age structure,
percentage of people not born in the UK, percentage of long-term unemployed and
percentage of people who use public transport to get to work are all very similar to the
UK average.
Page 12
2001 area classification of Super Output Areas and Data Zones
Both this and the white collar urban supergroup are similar in that they have no
variables either far below or far above the UK average. However there are distinctions
which can be made between them. The members of the white collar urban supergroup
are often located in more rural areas than those in the miscellaneous built up areas
supergroup. In the white collar urban supergroup population density is lower, as is the
percentage of people who live in flats and terraced houses, while the proportion of
people who live in detached houses and the proportion of households with two or
more cars is greater than in the miscellaneous built up areas supergroup. In addition
the percentage of people who were not born in the UK is greater in the miscellaneous
built up areas supergroup than in the white collar urban supergroup.
Summary: What is the Super Output Area and Data Zone area classification showing?
This document has demonstrated the characteristics and locations of each supergroup
in the Super Output Area and data zone area classification. The classification can of
course be investigated further by exploring the differences and similarities, not only at
the supergroup level, but also at the group and subgroup levels. It can also be used in
conjunction with other data sources to explore how specific characteristics differ
throughout the area classification.
Please see the methodology document here which describes how the Super Output
Area and Data Zone area classification was constructed and click here for more
information about other ONS area classifications.
Page 13
2001 area classification of Super Output Areas and Data Zones
How is this area classification different to other ONS small area classifications?
This section describes how this Super Output Area and data zone (SOA/DZ) area
classification is different from the ward and output area classifications produced by
ONS, and suggests when each should be used.
Comparison with the output area classification
Output areas (OAs) are the smallest geography for which Census data has been
released and were built from adjacent unit postcodes. In England, Wales and Northern
Ireland the OAs fit within 2003 statistical wards, except in the few cases where wards
fell below 100 residents (the minimum OA population size), in which case wards
were incorporated into larger OAs. In Scotland OAs are related to 2001 wards but do
not necessarily fit inside ward boundaries where confidentiality issues made it more
appropriate to straddle. The minimum population size in Scotland is 50 residents.
Output areas were merged to form SOAs and DZs taking into account measures of
population size, mutual proximity and social similarity. Table 1 shows the number of
OAs and SOAs or DZs and average population size in each of the constituent
countries of the UK at the time of the 2001 Census.
Table 1: Number of OAs, SOAs and DZs with average population size
Country
Number Average OA Number of Average SOA/DZ
of OAs
population
SOAs/DZs population
England
165,665
297
32,482
1,513
Wales
9,769
297
1,896
1,531
Scotland
42,604
119
6,505
778
Northern Ireland
5,022
336
890
1,894
United Kingdom
223,060
264
41,773
1,407
The SOA and DZ area classification has been constructed using the same
methodology as was used for the OA area classification. There are small differences
between the methods, such as the use of unscammed data (i.e. data which has not been
altered for confidentiality reasons) for the SOA and DZ area classification. In
addition, the OA area classification omitted a variable on the long term unemployed
and included a variable on all part-time workers rather than splitting part-time
workers by gender.
The data for the 43 variables were standardised so that each variable would have an
equal weight in the classification. The k-means clustering method, which creates
clusters which are as distinct from each other as possible, was then used to group the
data into between two and 12 supergroups. The eventual number of supergroups,
seven, was chosen because of the strong similarity of members within each
supergroup and because the number of members in the supergroups was fairly evenlysized.
Once the number of supergroups had been decided upon, each of the supergroups was
then broken down to create the number of groups. This was done by repeating the
procedure used for the first level, but this time performed separately for each
supergroup. This time two, three or four groups were considered within each
supergroup, and following this, the procedure was again replicated at the group level
to obtain the number of subgroups. This link describes how the SOA/DZ area
classification was constructed in more detail.
Page 14
2001 area classification of Super Output Areas and Data Zones
The names given to the supergroups, groups and subgroups are labels, rather than
definitive descriptions of all members of the clusters. These names are average labels
defined by examining the average features of each cluster. As a result it is possible
that isolated SOA/DZs will not be accurately reflected by the name of their cluster.
The SOA/DZ area classification masks some of the diversity in its constituent OAs,
but both geographies cover the UK and its people, and have been created using similar
methods. As a result there are comparisons which can be made between the two area
classifications.
We can compare the two area classifications by allocating each SOA/DZ to the OA
cluster to which it is most closely related. This leads to Table 2 at the supergroup
level.
Table 2: Relationship between OA and SOA/DZ area classification supergroups
SOA/DZ supergroups
OA supergroup code:
OA supergroup name 4
Strongest relationship
Countryside
3
Countryside
Professional city life
2
City living
Urban fringe
4
Prospering suburbs
White collar urban
6
Typical traits
Multicultural city life
7
Multicultural
Disadvantaged urban
1/5
Blue collar communities /
communities
Constrained by
circumstances
Miscellaneous built up areas
6
Typical traits
Table 2 shows strong relationships between some supergroups such as the countryside
supergroups and the multicultural city life and multicultural supergroups. Conversely
some relationships are less strong. For example, the SOA/DZ supergroup
disadvantaged urban communities appears to be split between two OA supergroups:
blue collar communities and constrained by circumstances.
The OAs distinguish between the blue collar communities and constrained by
circumstances OA supergroups. The difference between them was masked at the
SOA/DZ level by the disadvantaged urban communities supergroup, but drilling down
in this supergroup to its groups reveals the split: those SOAs and DZs in the blue
collar communities OA supergroup frequently fall into the blue collar urban group in
the SOA/DZ classification, whereas those SOAs and DZs in the constrained by
circumstances OA supergroup are often in the struggling urban group in the SOA/DZ
area classification.
Examples of small area classifications in London and Chichester
Mapping the area classifications over an area such as London clarifies the extent of
the agreement between them. Figures 1 to 3 show the area classifications for London,
by ward, SOA and OA respectively. To aid comparison between the different
classifications, similar colours have been used for similar supergroups.
4
These names were not published by ONS. They are the names given to the supergroups by the Output
Area Classification User Group at http://www.areaclassification.org.uk
Page 15
2001 area classification of Super Output Areas and Data Zones
Figure 1: Ward area classification in London
Figure 2: SOA area classification in London
Page 16
2001 area classification of Super Output Areas and Data Zones
Figure 3: OA area classification in London
The ward area classification map of London in Figure 1 provides an overview of the
types of areas within London but masks a lot of the diversity which can be seen at the
lower geographic levels in Figures 2 and 3. The SOA area classification describes the
capital well and provides sufficient detail of the locations of all supergroups without
being overwhelmed by the geographic detail at OA level in Figure 3. Variation in a
small area since 2001 (when data for this area classification was collected), such as a
new housing estate, is more likely to affect the OA than the SOA/DZ or ward area
classifications where such local factors are averaged over a larger area. For this reason
it is best to choose an area classification for a geography which is as large as possible
for the type of analysis required, rather than building up a larger area from a smaller
one such as OAs.
Conversely, if the area to be analysed is relatively small, it may be preferable to use
the OA area classification. Figures 4 to 6 show the area classifications at ward, SOA
and OA level for the town of Chichester on the south coast of England.
Page 17
2001 area classification of Super Output Areas and Data Zones
Figure 4: Ward area classification in Chichester
Figure 5: SOA area classification in Chichester
Page 18
2001 area classification of Super Output Areas and Data Zones
Figure 6: OA area classification in Chichester
Figures 4 and 5 mask the diversity within Chichester. A large part of northern
Chichester is part of the coastal and countryside supergroup in the ward area
classification whereas the rest of the town is either in the built up areas or student
communities supergroups. The SOA area classification distinguishes between
Chichester itself and the countryside surrounding it, although most of the town is in
the miscellaneous built up areas supergroup. It is only at OA level that different areas
within Chichester are visible. All supergroups are represented on the map in Figure 6,
including the city living and multicultural supergroups which were not visible in
either the SOA/DZ or ward area classifications.
Comparison with the ward area classification
In England, Wales and Northern Ireland SOAs have been created to nest within
wards, whereas DZs in Scotland have not been built with this purpose in mind. As a
result in England, Wales and Northern Ireland most SOAs have a one-to-many
relationship with wards whereas in Scotland DZ and ward boundaries cross-cut each
other severely. This often means that a ward can be examined further by viewing the
classification of its constituent SOAs or DZs. However there are some parts of the
UK, particularly in the City of London, the Isles of Scilly and around Ballycastle in
Northern Ireland, where wards contain only a few people and have had to be merged
with neighbouring wards in order to form an SOA.
Figure 7 shows the areas which are both an SOA (or DZ) and a ward. These are
concentrated in rural areas, particularly in Wales, Devon, Cornwall, north east
England and Northern Ireland. It is interesting to note that there are very few such
areas in Scotland, except on a few islands. 71 per cent of such SOAs or DZs are in the
countryside supergroup and 13 per cent in the white collar urban supergroup.
Page 19
2001 area classification of Super Output Areas and Data Zones
Figure 7: Areas where ward and SOA or DZ boundaries are the same
Page 20
2001 area classification of Super Output Areas and Data Zones
The resident population in an electoral ward varied significantly from around 100
people in some wards in the City of London to over 35,000 people in Small Heath in
Birmingham in 2001. For the purposes of the ward area classification, wards with a
population of fewer than 1,000 people were merged with a neighbouring ward.
Conversely SOAs and DZs are much more stable in terms of population size, varying
from nearly 500 people in a DZ around Strichen near Peterhead in eastern Scotland to
over 6,500 people in an SOA which covers several Oxford University colleges. Table
3 shows the number of SOAs or DZs and wards and average population size in each
of the constituent countries of the UK at the time of the 2001 Census.
Table 3: Number of SOAs/DZs and wards with average population size
Country
Number Average ward
Number of
Average SOA/DZ
of wards population
SOAs/DZs
population
England
7,932
6,282
32,482
1,513
Wales
868
3,345
1,896
1,531
Scotland
1,176
4,304
6,505
778
Northern Ireland
577
2,921
890
1,894
United Kingdom
10,553
5,571
41,773
1,407
A similar method was used to compare the SOA/DZ and ward area classifications as
the OA and SOA/DZ area classifications above. However there were no conclusive
results from this comparison because of several factors. Each classification only takes
account of the data at the geographical level to which it is applied, and as discussed
below, the methods used for the two sets of area classifications differ significantly. In
addition, unlike the data in the ward area classification, the data in the SOA/DZ area
classification were transformed to a log (logarithmic) scale before the data were
standardised. This was done so that the problem of outliers with large values was
greatly reduced and was not an issue at ward level because the increased size of the
geography ironed out such anomalies.
There are a number of different methods which could be used to create an area
classification, two of which are Ward’s method and k-means method. Both of these
techniques have been used in different ONS area classifications. Ward’s method is
used in the local authority area classification, although it only works well when the
number of areas to group is under around 1,000. The k-means method was used for
the ward area classification to group the 10,553 wards into 1,000 clusters, after which
Ward’s method was used to create the final classification. This method was not used
for the SOA and DZ area classification as grouping many more areas (41,773) into
1,000 clusters tends to create groupings with widely differing numbers of members, in
effect separating the outliers from the rest of the population. As a result the same
technique was adopted for the SOA and DZ area classification as the OA
classification (using only the k-means method). More about the specific methods
associated with different ONS area classifications can be found here.
Summary: How is this area classification different to other ONS small area
classifications?
Due to the same methodology being used for the OA as the SOA/DZ area
classifications, strong similarities between the two are observed at the supergroup
level. Such a strong relationship is not evident when comparing the SOA/DZ and
ward area classification due to the different method used for the ward area
classification, despite both classifications covering the people of the UK.
Page 21
2001 area classification of Super Output Areas and Data Zones
The choice of small area classification to be used depends on the question being asked
of any analysis and the size of the area being analysed, whether that is the whole
country, a region or a particular town or city. For further advice and help with this or
other ONS area classifications please contact [email protected].
Further links
Please see here for more information about area classifications, including the Super
Output Area and Data Zone area classification.
Page 22