2001 Geography: the good, the bad, the ugly

1
2001 Geography the good, the bad, the ugly?

• David Martin
• University of Southampton and ESRC/JISC Census
  Programme

2
Overview

• What did users want?
• What happened
• The good
• The bad (the ugly)
• What has happened since
• And a few other things
• Ways forward

3
What did users want?

• Postcode and statutory geographies
• Uniformity of population sizes (all above
  threshold)
• Control over shape (observe settlement pattern
  and topographic features)
• Internal homogeneity of population
• Compatibility with previous census geographies!

4
What happened

• Entirely new methodology
• Automatic generation of postcode polygons by
  Thiessen polygons around Address-Point, clipped
  to features
• Automated zone design by iterative combination of
  postcode polygons until best combination of
  statistical constraints

5
Good points

• Postcode and statutory geographies
• Uniformity of population sizes (all above
  threshold)
• Limited control over shape and homogeneity
• Nationally consistent geography
• Free digital boundaries and lookup tables that
  match the OA data on time and to spec!

6
Source scrol.gov.uk
Source nisra.gov.uk
Source national-statistics.gov.uk
7
The bad (the ugly)

• Mismatch with real (or OS?) topographic features
• Some bizarre OA shapes
• Some postcodes still split
• Complete change from 1991
• Differences across UK
• Lack of local input/validation

8
Source neighbourhood.statistics.gov.uk
9
What has happened since

• Embedding within SOAs and use for NeSS
• New datasets e.g. ID2004, but OAs/SOAs are not
  neighbourhoods
• NeSS not in original design, but cannot be
  ignored
• Community split over merits of wards and
  stability vs. currency

10
And a few other things

• Mean High Water (as 1991), not Extent of the
  Realm (as 2001)
• Business OAs
• Use MasterMap rather than synthetic polygons
• Sort out a top-quality address infrastructure!

11
Central Newcastle
Severn Estuary
12
Ways forward

• Retain all the principal advantages, while
  winning others
• Ward-based geographies will never solve our
  problems (for more than a couple of years)
• Geography and disclosure control issues cannot be
  tackled separately

13
Conclusion

• Geography isnt the census
• But when we get it right, it makes the picture
  clearer

2001 Census Programme
14
Abstract

• 2001 geography the good, the bad, the ugly?
• David Martin
• University of Southampton and ESRC/JISC Census
  Programme
• 2001 output geography in England, Wales and
  Northern Ireland was produced by an entirely
  different methodology to that used for previous
  censuses. Automated zone design procedures were
  applied to artificially generated postcode
  polygon boundaries in order to create a
  nationally consistent output geography. This
  approach led to the best ever match between
  census and postcode geographies while respecting
  ward and parish boundaries, ensured that all
  output areas contained above-threshold
  populations and provided a degree of control over
  social homogeneity and geographical shape. The
  boundaries were integrated with the postcode and
  higher-area lookup tables and were produced to
  specification and on time. In many respects this
  represents a remarkable series of successes.
  However, the absence of definitive postcode
  boundaries and limited input of topographic map
  features nevertheless led to many output areas
  having highly irregular shapes. Some postcodes
  are split across ward and parish boundaries so a
  perfect match between census and postcode
  geographies was not possible, nor do they always
  respect social divisions apparent to those who
  know an area well. The complete change of
  methodology made comparison with 1991 enumeration
  districts very difficult and the automated
  process provided no opportunity for local
  consultation on boundary placement. Subsequent
  to the census, the output area boundaries have
  been further aggregated to create the super
  output areas used by the Neighbourhood Statistics
  Service and many users have been encouraged by
  the promise of boundary stability, yet important
  decisions are still to be made regarding the
  maintenance strategy for the 2001 output areas.
  Key to 2011 census output will be the very
  challenging need to address the principal
  shortcomings of the 2001 geography without losing
  its major advantages.