Neighbourhood Statistics

1
Neighbourhood Statistics

• Margaret Frosztega
• Agnès Estibals
• Office for National Statistics

2
Overview

• Policy drivers
• Vision of Neighbourhood Statistics
• How will we get there
• New methodologies
• Geography policy and implementation
• Statistical challenges
• Where we are now
• Where we want to be by 2006

3
Cabinet Offices Social Exclusion Unit (SEU)
Prime Minister asked SEU to report on how to
develop integrated and sustainable approaches to
the problems of the worst housing estates,
including crime, drugs, unemployment, community
breakdown and bad schools etc
Led to National Strategy for Neighbourhood
Renewal
- bridge gap between most deprived neighbourhoods
rest of England
- in worst neighbourhoods, achieve lower
long-term worklessness, less crime, better health
better educational qualifications
4
Policy Action Teams (PAT)PAT18 Better
Information
?????
Goal
How to overcome the barriers to obtaining good
quality small area information and make
recommendations on how to achieve this.

• What information needed why

• Problems in getting better information

• What is happening to deal with these problems

• What needs to change

5
Policy Action Teams (PAT)PAT18 Barriers to
Information

• Supply Side
• Confusion about the law
• Hoarding instincts
• Excessive charging
• Poor information about information
• Data collected for different areas/geographies
• Some data not collected at all
• Demand Side
• No agreement on identifying most important data
• No institution to assess the demand and invest to
  meet it

6
Vision for Neighbourhood Statistics

• PAT 18 Report - Recommendation 9
• Establish Neighbourhood Statistics
• capability of analysis by any geographical area
  ethnicity gender age over time
• accessible at affordable price, preferably free,
  one-stop-shop
• Develop implement a common Geographical
  Referencing framework for the public sector
• Promote good practice in data collection, sharing
  use
• Encourage greater coding of administrative data
  by ethnicity

7
How will we get there

• Maximise use of national administrative data
  sources e.g. social security, tax, housing,
  property prices etc.
• Co-ordinate and harmonise data from other sources
  e.g. Local Authorities, Health Authorities,
  Police Forces etc.
• Develop new methodologies e.g. small area
  estimation, disclosure control, data linkage,
  indicator development etc.
• Develop a new geography for the reporting of
  small area statistics

8
New methodologies

• Develop a stable hierarchy of geographical units
  to collect and hold statistical information
• Research methods to estimate the distribution and
  magnitude of a target variable from one area
  geography to another
• Develop disclosure control standards and tools
• Develop quality measurement framework
• Derive multivariate indicators

9
Geography Policy
Aim

• Support the move to a more stable small area
  geography
• Enable change over time to be tracked at the
  small area level
• Permit users some degree of self-definition of
  areas
• Enable continued dissemination of data on
  electoral wards and other widely-defined areas
  through estimation
• Protect the confidentiality of information

10
Policy into practice

• Development of stable hierarchy of geographical
  units to collect and hold statistical information
• Dissemination using combinations of these units
  plus best-fit estimation methodology
• Similar best-fit methodology to compare
  information over time by recasting data
• Adoption and implementation of this policy
  through large-scale roll out to data owners of
  common tools
• Development within ONS of geographic referencing
  infrastructure to support all National Statistics

11
(No Transcript)
12
Applying Geography policy

• Minimising Boundary change
• Agreement by Boundary Commission to minimise ward
  changes
• Proposal to align future wards with OA boundaries
• Statistical ward defined as the wards notified by
  December of previous year
• Maintenance policy OAs / SOAs
• Handle population shifts by merging or dividing
  single OAs

13
Applying Geography policy

• Issues
• Odd shapes of OAs
• Naming SOAs for ease of use
• Historic analyses using wards
• SOAs are not neighbourhoods
• How good is the quality of estimates for other
  geographies derived from OAs and SOAs
• How quickly can the estimates be delivered
• How to ensure sharing of user defined
  geographies

14
Applying Geography policy

• SOAs based on size, homogeneity and proximity
• lower level nest into 2003 wards
• In England and Wales, computer generated lower
  and middle layer
• Consultation on middle layer
• Upper layer to be proposed by local planners ??
• Difficult to reflect physical geography - rivers,
  major roads, railway lines
• In Scotland, consultative process layer by layer

15
Examples of Geography Policy 1

• Estimating to Geographies not in Standard
  Hierarchy

Result Exact Fit OAs 1-7 OR (SOA1 SOA2) -
SOA8 NB. All OAs next within the users chosen
area
16
Examples of Geography Policy 2

• By aggregation that approximates to user defined
  or system defined area

Result Users chosen area OA1 to 6 OA8 OR
(SOA1 SOA2) - SOA7 NB Result depends upon
users choice for example whether position of
population weighted centroid of Output Area is
inside or outside the boundary of the users
defined area
17
Examples of Geography Policy 3

• Approximate estimate for users chosen area

Result Users chosen area OA1 to 6 part OA7
and part OA8 NB The value for a part OA may
reflect the proportion of base units (eg
postcodes / addressess / households) in the OA
that are within the users defined area
18
Statistical challenges

• Access key administrative data sources, including
  at individual record level, and produce small
  area statistics whilst protecting confidentiality
• Develop small area estimation methods and models
  for assessing the accuracy of synthetic estimates
• Develop a framework for assessing and reporting
  the data quality
• Derive indicators that are comparable over space
  and time to support neighbourhood renewal agenda
• Develop spatial analysis tools

19
Statistical challenges

• Small area estimation
• The estimation method currently tested is based
  on weighted imputation
• The target variable is apportioned down from the
  source geography to a lower level centroid
  geography according to a suitable weighting
  criterion
• The centroids inside each target area are then
  aggregated to generate an estimate of the target
  variable

20
Statistical challenges

• Quality framework
• A series of simple designations will be
  introduced for each dataset
• National Statistics - the information complies
  fully with the National Statistics Code of
  Practice and related protocols
• Experimental Statistics - The information has
  been developed in accordance with National
  Statistics principles but has yet to be fully
  accredited as a National Statistics
• Not National Statistics - the information based
  on administrative data does not comply fully with
  the National Statistics Code of Practice

21
Statistical challenges

• Indicator development
• The objective is now to bring indicators to the
  forefront, which involves
• linkage of datasets at individual record level
• framework for ensuring comparability over time
  e.g. issue of discontinuities in administrative
  datasets from which indicators are derived
• framework for ensuring comparability over space
  e.g. issues of nationally held datasets collected
  under central and local guidance, and locally
  held datasets

22
Where are we now?

• Initial Launch - ward level database
  February 2001ü
• Incremental phase - add to ward level data
  2001 - 2002ü
• GIS implementation phase - geographic
  informationsystem incorporated on web pages
  2002ü
• Census launch phase - 2001 Census results
  dovetailinto NeSS 2003ü
• Full Service phase - new geographical units,
• further datasets, improved user flexibility
  2004-2005

23
Where do we want to be by 2006?

• Increased outreach
• Common environment - small area statistical
  repository for all requirements
• Access to key microdata
• More comprehensive, consistent, topic-based set
  of indicators
• Improved analytical capacity