Memorandum by the Greater London Authority
The Greater London Authority has, since its
creation in April 2000, been involved in consultations on the
2001 Census. The staff involved on behalf of the GLA are the same
staff who were involved in those consultation before the GLA was
created, at that time representing the London Research Centre
and the London Boroughs. These staff are nationally recognised
experts in Census matters and are frequently called upon for advice
by Census users in local authorities throughout the country. Independently
of this staff were also employed as part of the Census field force.
1. The adequacy of the preparatory work and consultation
undertaken prior to the Census
1.1 There was plenty of consultation prior
to the Census, much more so than for the previous Census, and
it was generally welcomed. Nevertheless, it has not always been
clear how much influence the consultations have had on subsequent
decisions and how comments made by large groups of users have
been weighted against comments made by one individual who may
only use Census data sparingly.
1.2 Our view of the consultation process
has also been influenced very strongly by recent changes in policy
by ONS which make a mockery of all the time and effort put into
the consultations by users. The policy changes (detailed in Section
3 below) mean that many of the opinions and needs of local authorities
expressed throughout the consultation process would have been
totally different if based on ONS policy as it is now.
2. The conduct of the Census
2.1 Many areas of the Census operation were
good and worked as well as in the previous Census. Among these
we would include the staff recruitment, which required a bit more
effort in London but was eventually successful. Publicity was
generally good and many people were expecting the enumerator when
2.2 Some areas of the Census operation were
disastrousshortages of household and continuation forms
were a complete surprise to field staff and caused enormous problems.
Posting back of Census forms has been hailed a success by ONS
but it did not seem this way to the field staff. Problems with
the postal service in parts of London made the enumerators' jobs
extremely difficult, if not impossible to carry out according
to instructions. The follow up by enumerators to addresses was
carried out very sketchily, if at all, in some areas because they
had to wait so long before they could be sure there was no response.
Forms were taking two weeks to get through the postal system in
parts of London, and this meant that the very keen workforce had
to be held back and the time allowed for follow up drastically
2.3 Forms were not checked properly by the
field staff, not through any fault of their own, but because they
were instructed only to check three questionsage, sex and
marital status. An additional problem with checking the forms
was the estimate by ONS that a form could be checked in 1 minute.
This was not the case in London where there were many language
problems and difficulties in contacting people, and it took more
than one minute to check forms which were filled in by respondents
whose first language was not English.
2.4 Outsourcing of parts of the operation
appears to have caused major problemsin particular the
payments to field staff were a complete embarrassment to all concerned
and caused a lot of ill feeling.
3. The utility of the planned outputs for
prospective users(Will the data be fit for purpose)
3.1 It is not possible to comment definitively
on table layouts and proposed outputs that are not finalised and
which are still under discussion. It had looked as if the proposals
might be acceptable. However, ONS have just announced that "All
counts in all tables for England and Wales will be rounded either
to zero or to numbers divisible by three". This makes all
the data unfit for purpose and users are extremely angry that
they have been misled by ONS for some considerable time. We have
previously had categorical assurances from ONS that any adjustments
necessary to preserve confidentiality would be made to the underlying
database and not to individual cells within tables. If ONS carry
out their latest proposals tables will not be internally consistent
and different counts will be produced for the same variable depending
on which table is used to produce it. As a result, the whole One
Number Census project becomes complete nonsense.
3.2 A further very late change in policy
is that the minimum thresholds for release of data for output
areas has been raised considerably, and this has very important
implications for the earlier decisions on the average target size
for output areas. If the thresholds had been clear from the start,
and not increased at the last minute, then very different decisions
would have been made about the preferred size of output areas.
We would expect such radical changes in policy to result in a
complete rethink in a number of areasespecially the target
size for output areas which must be increased.
3.3 Doubts have been raised in two other
areas that are vitally important to Census users. One is in relation
to the boundaries which are to be provided with the data. It had
been expected that we would receive vector boundaries, as suggested
in the consultation documents but we now hear that unusable raster
boundaries may be provided instead. These are no use whatsoever
for thematic mapping and would result in considerable extra expense
and delays. This is unacceptable. The other area is related to
boundary changes which are due to be implemented before the Census
data is published. We want ONS to adopt the April 2002 boundaries
as standard for the data, as do users in other parts of the country
also affected by boundary changes. The boundaries will be provided
in digital form to ONS by Ordnance Survey, but we have still been
unable to get any commitment to use them out of ONS.
4. Lessons for the next Census
4.1 We had hoped that lessons from the 1991
Census would have influenced the 2001 Census but are very disappointed
that some very important lessons have been ignored. One of the
major concerns raised by local authorities throughout consultations
on the 2001 Census output has been the internal consistency of
the datasets. Basically we want to get the same answer to any
query whichever table is interrogated. For example we would like
the number of people aged under 5 in any area to be consistent
and not differ between tables in the output. We want the figure
for the number of under 5s in any ward to be the same as the number
of under 5s in that same ward reached by adding up the corresponding
figures for all the smaller constituent output areas. As a result
of this we have no confidence in the ability of ONS to carry forward
the lessons that should have been learnt from previous censuses.
4.2 Very late changes in ONS policy have
meant that our views and comments made over a number of years
have been totally disregarded. We have no confidence in the decision
making process that has led to this and would have to think very
seriously in the future about feeding back information to ONS
which is then totally disregarded.
5.1 We ask the Committee to seek assurances
from ONS that any adjustments necessary to preserve confidentiality
will be made to the database and not to individual tables.
5.2 We ask the Committee to urge ONS to
ensure that the Census data is fit for purpose, in particular
by recognising the administrative boundaries that are current
at the time of the release of the data.
5.3 We ask the Committee to instruct ONS,
in the light of the major changes to minimum thresholds at this
very late stage, to consult further with users about the implications
of this change so that user priorities under these altered circumstances
can be determined.
5.4 We ask the Committee to urge ONS to
come to an agreement with Ordnance Survey about provision of digital
vector boundaries with the Census data, and to ensure that local
authorities are provided with the boundaries in a format that
is fit for purpose and acceptable to them.