The future of output geography - Office for National Statistics

The future of output geography
- geography policy for NeSS
Robert Heyward
ONS
Census Output
Areas
• recap
• progress
• what’s new?
Neighbourhood Statistics
Geography Policy
Super Output Areas
Census Output Areas
• recap
• what’s new ?
• outcome
• availability
Output Areas - recap
• New for 2001 Census (E&W), a geography
specifically for low-level census outputs.
• approximately 175,000 OAs
in England & Wales
• align with administrative boundaries
(Wards, Civil Parishes)
• built from unit postcodes on Census day.
(April 29th 2001); split for wards etc
Output Areas - recap
• strong homogeneity of tenure and
dwelling type within each OA
• size thresholds and targets
• synthetic but snapped where possible to road
centrelines and administrative boundaries
• Scotland - similar but different approach
• Northern Ireland, postcodes > same algorithm
Output Areas - What’s new?
• improved ‘shape’
– now minimising distance
between population centroids.
• improved boundaries
– to mean high water
– generalised (low resolution) sets
• eg for thematic display
• target size based on households
– with a minimum size for
both population and households.
Output Areas: Achieved Size
70000
60000
50000
• Hhlds
40000
30000
20000
10000
0
40 - 49 50 - 59 60 - 69 70 - 79 80 - 89 90 - 99
100 109
110 119
120 129
130 139
140 149
150 159
160 169
170 –
179
180 189
190 199
200+
Household range
40000
30000
• Pop
20000
10000
0
100 124
125 149
150 174
175 199
200 224
225 249
250 274
275 299
300 324
325 349
Population r ange
350 374
375 399
400 424
425 449
450 474
475 499
500+
Output Areas - Outcome
• Achieved size within narrow range
– 95% OAs between 100 and 400 population
– only 1% above 500 population (eg university
accommodation) (example figures for one city)
• boundaries much shaped by
postcode and ward geography:
– most OAs made up of 'whole' postcodes
– OAs at ward / parish boundary contain
mixture of 'whole' and split postcodes
Output Areas - Availability
• Vector boundaries for OAs, (like the data),
available ‘free’ to ALL sectors
• Simple ‘click-use’ terms
• ONS have paid all OS license fees
for a period of 10 years
• Limitations on commercial repackaging possible but talk to OrdSvy or Census CS
• Factsheet available
Neighbourhood Statistics
Geography Policy
this is what it means for users …..
Geography Policy
• Emphasis on stability
– improving our ability to measure change over time
• Use of standard geographic units
– Grid reference
– Output areas
– Super Output areas
– Districts
• Statistics built from these blocks or by estimation for other areas
Geography policy
• Standard
geographic
units
Districts
Super OAs
Output Areas
• ‘Frozen’ bases for
data collection &
output
Grid reference
Geography Policy for NeSS
• Output Areas (OA)
–
–
–
–
–
–
–
Small - provide focus
Stable - help with time series
Stable - ease integration
Stable - can act as a building brick
Homogeneous - reflect the real world
Help with disclosure
Independent
Geography Policy for NeSS
• Super Output Areas (SOAs)
– fill the gap in existing geographies
between district and census output area
– comparability is easier as size is standardised,
compared to present variation in size of wards
– helps to address disclosure control issues
as OAs could be too disclosive
– enables ready comparison over time:
more stable than wards
Geography Policy for NeSS
• SOAs - Hierarchy of Intermediate Layers
Layer
Size of Zone
•
•
•
•
•
Output Area
Lower
Middle
Upper
District
-
250 persons (average)
1,000 (minimum)
5,000 (min)
20,000 (min)
25,000 (and above)
Geography policy
• Standard
geographic
units
Districts
Super OA 3
Super OA 2
Super OA 1
• ‘Frozen’ bases for
data collection &
output
Output Areas
Grid reference
• Group of
Output
Areas
• Forming a
lower level
SOA
• Group of
Output
Areas
• Forming a
lower level
SOA
• Group of
lower level
SOAs
• Forming a
middle level
SOA
• Group of
middle level
SOAs
• Forming an
upper level
SOA
• Upper level
SOAs
• Nest within
the District
Super OAs : Design Issues
• Size : variation within layer
– largest may be two (or three) times minimum size
– target size : relative weighting of size and other factors
– target size in persons or households ?
• Relationship with Wards
– Lower layer - Nest :
• data continuity : Census Standard Tables
– Middle layer - Overlap :
• more scope for homogeneity : connectedness
Geography policy
Districts
• Standard
geographic
units
Super OA 3
Super OA 2
Wards
?
Super OA 1
• What relationship ?
Output Areas
• What layer ?
Grid reference
If constrained to wards ….
• variation in size means
many wards will be one or two SOAs
• we get all the disadvantages of wards
(not our ideal geography)
If NOT constrained to wards
• - more complex disclosure issues
BUT ….
Even if we do constrain to wards the
relationship only holds at the start …..
• Because SOAs are frozen and the wards
within which they fit are not
The complicated relationship with wards
• Individual
Output Areas
nest within
wards
The complicated relationship with wards
• Individual
Output Areas
nest within
wards
• The lower
layer could
also be
constrained to
wards
The complicated relationship with wards
• But that does
not mean that
ALL layers
have to be
constrained
• Here the next
layer is built
from lower
blocks but
breaks wards
The complicated relationship with wards
• This would be
because the
higher level
area is more
homogeneous
than one built
from wards
Similar
characteristics
SOAs : Design Issues (contd)
• “Shape” / Connectedness :
– down valleys NOT across mountains
– OS “streets” : OA boundaries : novel
• Homogeneity
– tenure, type of dwelling : Additional Factors ?
– ONS classification of Output Areas
• Relative Weights / Balance
– greater weight to homogeneity / connectedness
– wider range of size within each layer
Super OAs : Process
– Census OA
– Lower
– Middle
– Upper
– District
-
Given : Census 2001
Zoning software
Software / Appeal
Suggestion / Software
Given
Super OAs : Provisional Timing
• Feb to April
•
•
•
•
•
May & June June & July
August
Autumn 2003 Winter 2004 -
Feedback on Proposal
Feasibility Study
Review and Decision
Generate by Software
Distribute
Appeals/Suggestions
Publish
Autumn Process - Sequence
• Local Agencies consider and agree
• (via Local Strategic Partnerships ?)
– potential modification to middle layer
– suggested composition of upper layer
– names for zones in all layers (if desired)
• separate target dates for each stage
• lower layer not open to modification
Looking Ahead
• Collect and hold
– core geography
• Publish via web site
– core geography and
– user defined geographies, either
• ad-hoc
• popular, prepared
Geography policy
Districts
• Standard
geographic
units
Super OA 3
Super OA 2
Wards
?
Super OA 1
• ‘Frozen’ bases for
data collection &
output
Output Areas
Grid reference
• Outputs built
from building
blocks
An output built by adding
up data collected for the
frozen block geography Output of
EXACT
count-OAs,
SOAs, LADs
Choose
dataset(s)
Choose
year(s)
Choose
area
An output built by
estimating for an area
that isn’t an exact fit to
the blocks
Output of
BEST FIT
ESTIMATE
for area
Geography Policy - Dissemination
– Data sets to 2001
referenced to 1998 ward and LA boundary
– Data from 2001 to 2003
to be built from counts for wards or output
areas (*) and LA, boundaries at 31/12/02
– Data from 2004 held for OA or SuperOA layer, & district
best fit to current (*) output geographies
(* but)
• Data referring to for 2001 to 2003
– For each data set, either wards or
SuperOAs may be estimated, if necessary
to prevent ‘dual geography’ disclosure
• Data for 2004 onwards
– Counts for OAs, SuperOAs and district
– Estimates for ‘current’ ward geography :
boundaries by SI at end of previous year
Geography Policy for NeSS
• Issues
– will not collect or publish data on
exact ward boundaries after 31/12/02
– outputs will be provided to changing
boundaries by ‘best fit’ techniques
– additional estimation techniques
need to be developed
– assessing impact of estimation and
disclosure control / rounding on data quality
Examples of estimation - Census
• “mainstream” geography : exact boundaries
– Census Ward, Parish, District, Govt Office Region
– Wards as known by 31.12.02 or anticipated
• other ‘standard’ geographies : estimated
– Parliamentary Constituencies etc
• estimated by aggregating CAS data
from best fit of Output Areas
– National Parks, exceptionally,
• independently aggregated from grid referenced records
Future Electoral Wards
• NeSS : Prepared user defined geography
– estimated from OAs and SuperOAs
– estimates pre-calculated for ready access
• Synchronised date
– PAT-18 initiative; now National Statistics policy
– As determined by Statutory Instruments laid
before 31 December in previous year.
• Eg data for 2003 based on boundaries in SIs before
31.12.2002, (whether next elections in 2003 or later)
Summary
Geography policy
Districts
• Standard
geographic
units
Super OA 3
Super OA 2
Wards
?
Super OA 1
• ‘Frozen’ bases for
data collection &
output
Output Areas
Grid reference
• Outputs built
from building
blocks
An output built by adding
up data collected for the
frozen block geography Output of
EXACT
count-OAs,
SOAs, LADs
Choose
dataset(s)
Choose
year(s)
Choose
area
An output built by
estimating for an area
that isn’t an exact fit to
the blocks
Output of
BEST FIT
ESTIMATE
for area
Further information and Contacts
• Super Output Areas
- Proposal via
http://www.neighbourhood.statistics.gov.uk/Information_About_Ness.asp
– Feedback
• by end March, preferably
• end April, latest
– [email protected]
– Robert Heyward 020 7533 6115
• Geography Policy
– [email protected]