Examination of Witnesses (Questions 20-39)
MR KEITH
DUGMORE, PROFESSOR
DAVID RHIND
AND PROFESSOR
DAVID MARTIN
16 JANUARY 2008
Q20 Mr Dunne: Professor Rhind, you
have touched on some of this in your responses to John Thurso,
but perhaps, Professor Martin, you could elaborate. You have indicated
that an increased frequency of census would be useful. Are you
advocating a five-year census or a rolling census along the lines
of what Professor Rhind was talking about in some other countries?
Professor Martin: I think the
sense in which that discussion has gone on within the Royal Statistical
Society Census community, the most important thing, is the increased
frequency of the information, and that would be the priority,
because for the decennial Census you know that you need information
which is more up-to-date, and so do commercial users. Certainly
across the sectors that is important. A more frequent census is
a model which some countries have adopted, but that is not to
say that you could not, if you moved to something driven from
an administrative source, get to the same position. I think there
is some caution in feeling that an administrative system can be
switched on very quickly. When we look at the experience of other
countries, there is a tension between the operational activity
of running administrative registers and the statistical use of
those registers, and, therefore, something which perhaps used
2011 or even 2016 to try to cross-calibrate those and thoroughly
understand how we move from one system to the other would the
most important. I do not think the five-yearly census is the magic
answer, but the increased frequency of reliable information is
a priority for all those users.
Q21 Mr Dunne: Indeed, and there is
a distinction between counting the population as to purely the
number of people and the other information that is available from
the Census questionnaire, but presumably the latter could be dealt
with largely on a sampling exercise. You do not have to count
100% of the population to get a broadly representative answer.
So, is there a distinction that should be made between population
location and the information that is available through questionnaire?
Would Mr Dugmore like to answer that?
Mr Dugmore: Yes, I would still
favour collecting a lot of information about individuals, because
certainly in the commercial world there is interest not only in
the number of people but the types of people, classifying in all
sorts of way, and that leads through to market estimations, to
what you are expecting people to buy, and so on and so forth,
and so in that sense I agree very much with David, I think there
is great deal of value to be obtained from the use of administrative
files and producing small area data. The great downside with sample
surveys, of course, is that it gives you very little geographical
detail. Expenditure and food surveys give national and regional
figures, but certainly not parliamentary constituency level figures,
let alone store catchment figures.
Q22 Mr Dunne: Indeed, but a sample
of a thousand people per constituency, say, would give you a very
good indication. The opinion pollsters would argue that a sample
of a thousand is a good representative sample and, if you were
to do that and apply that to the 646 constituencies—that
is counting nearly 1% of the population—you might get similar
answers to counting 100% of the population. You are telling me
that you would not statistically.
Mr Dugmore: I think for obtaining
statistics for constituency level that would be fine, but if you
are wanting information for neighbourhood renewal or store catchments,
local things, then you really do need to get down to either administrative
records or a precognitive census.
Q23 Mr Dunne: But administrative
records do not provide you with a questionnaire level of data.
Mr Dugmore: No, quite true.
Q24 Mr Dunne: Is there anything which
we have learned from other countries in terms of finding a way
to do this more cheaply? You referred to the half a billion pound
cost of a traditional census. Are there lessons that we can apply
from elsewhere that you are aware of?
Professor Rhind: If you go back
to where we were earlier, the costs of producing population figures
on a daily basis in Scandinavia and some other countries is miniscule.
There is a dramatic graph that the Finnish people showed us, I
think, where the costs fell down dramatically from the time that
they moved from having censuses. All that pre-supposes, of course,
that you have these registers in existence, so there is an up-front
cost, but actually in the UK we have a number of administrative
data sources which, as a first approximation, I think, might be
looked at for creating this: hence my suggestion of an experiment
in parallel with the 2011 census.
Q25 Mr Dunne: As we were hearing,
in Scotland they have already achieved a definitive address file.
Did that overcome the barriers to data protection that combining
a file from the ONS and local government electoral records and
Post Office data would allow you to do?
Mr Dugmore: I do not think that
the issue is data protection at all. If we look at the initiatives
to try to produce a definitive national address file over more
than ten years now, the three parties involved, the Post Office,
who to some extent have put it to the side, but Ordnance Survey
and then local government, each have their own files, but when
it came down to merging them into a definitive file, there were
issues of intellectual property, ownership of the file, and not
wanting to co-operate or not wanting to see the other person win,
and so there has been a stand-off there for a long time.
Q26 Mr Dunne: That could be dealt
with through legislation.
Mr Dugmore: I am sure it could.
Q27 Mr Dunne: Would that be your
recommendation? Is that what was required in Scotland or did they
find another way of reaching agreement?
Mr Dugmore: That would be my view.
As I said in my notes, I think that heads need to be banged together.
It seems as though the parties involved will never come to an
agreement and somebody needs to tell them to agree.
Q28 Mr Dunne: Do you agree with that,
Professor Martin?
Professor Martin: Yes, we have
seen many initiatives of different types and names, over a period
of probably pushing towards two decades, to attempt to create
a definitive address register in some form and none of them has
brought us to a point where we have that. It seems ironic that
we recognise that as a major issue from 2001 which it has not
been possible to deal with and it is still a major issue that
will challenge 2011.
Q29 Mr Dunne: Without wanting to
put words into your mouth, would you be recommending to this Committee
that we recommend to the Government that they introduce legislation
for a national address file?
Mr Dugmore: I would certainly
say that. Just to give you one illustration, the ONS census test
last year in Camden, they prepared two address files and then
they found another 7,000 addresses which were not on either. If
you think that the next census is going to rely on post-out to
approach people, then you do need a definitive national file that
people have got faith in.
Q30 Mr Dunne: Can I ask each of the
other witnesses to respond to that question specifically about
legislation and the recommendations from the Committee?
Professor Rhind: I do not think
I could add anything more to what Keith has said.
Professor Martin: I would agree.
I think the only comment which is worth bearing in mind is that
our aspirations towards administrative registers are equally reliant
on getting the address register right, and the Census is challenged
by it in just the same way as the administrative record issues.
Q31 Mr Todd: Was this not a good
example of a quite woeful lack of will to address a really quite
critical national problem? We are talking about this from the
point of view of the Census, but there are a large number of other
reasons why having an accurate, unified national address register
was a desirable goal.
Professor Martin: Yes.
Q32 Mr Todd: If my memory is right,
Professor Rhind, you were probably involved in some of these discussions
at various times.
Professor Rhind: I confess, Sir,
I started the address work inside Ordnance Survey 15 years ago,
or thereabouts, so I do have a history. It is extraordinary, looking
from the outside now, how considerable efforts over at least a
five-year period, but between the different parties, has come
to nothing. The Department for Communities and Local Government
have put a great deal of effort into trying to move towards—
Q33 Mr Todd: Even though two of the
parties, the Ordnance Survey and local government, are, of course,
under their direct remit. The Post Office is not.
Professor Rhind: Ordnance Survey
is a—
Q34 Mr Todd: A trading fund?
Professor Rhind: Yes, a trading
fund, which obviously has some impact, but it is not only an executive
agency but a department on its own in one sense, although its
ministers sit in DCLG.
Q35 Mr Todd: Surely this should not
require legislation. This should be a straightforward matter of
commonsense addressed by ministerial ownership of the problem.
Professor Rhind: In an ideal world.
Q36 Mr Todd: But perhaps it has never
been escalated sufficiently to ministerial attention that has
allocated enough priority.
Professor Rhind: Certainly I know
that a number of ministers have looked at this and made some attempts
to resolve the issue. I do not know why it has not been sorted,
and there are different opinions on that.
Q37 Nick Ainger: Professor Rhind,
given the difficulties with the national address register, is
the decision of ONS to cut the number of enumerators in the 2011
Census from 71,000 in 2001 to somewhere between 45 and 50,000
the right decision?
Professor Rhind: We are not comparing
exactly like for like, because there is a post-out, is there not,
planned in England and Wales for the next Census, and part of
the role of enumerators was to hand out, in the past, those things;
but certainly there is cause for some concern, not least because
I think ONS have readily accepted that the hard to enumerate areas
will require more enumeration time than in 2001, and we know things
are becoming increasingly difficult to enumerate for all the gated
communities, for all the migrant workers who do not want to be
found, and so on, so I think there is some real concern in that.
It is not, however, I suspect, just the numbers. Clearly, there
is an issue of training and quality, and that relates in part
to how much the enumerators are paid and who you can attract.
I think there was an analysis of the 2001 Census where it became
clear that the amount paid to individual enumerators was such
that they did not in every area attract the quality of people
that they needed.
Q38 Nick Ainger: In fact that is
the evidence that we have had from the ONS, that they did have
a problem with recruitment and certainly retention. After training
an enumerator, they then went off and got a different job and
they never carried out the work of an enumerator.
Professor Rhind: There is an interesting
parallel to the United States, where the United States have changed
their census from a big bang event to a smaller big bang event,
with a three-year rolling census thereafter where some parts of
the country are done one year and some parts are done another,
and one of the driving factors was to have people in continuous
employment, who they could rely upon and could train up over a
number of years to do this.
Q39 Nick Ainger: But that is not
what we are facing between now and 2011?
Professor Rhind: No.
| 
