Examination of Witnesses (Questions 60-75)
MR KEITH
DUGMORE, PROFESSOR
DAVID RHIND
AND PROFESSOR
DAVID MARTIN
16 JANUARY 2008
Q60 Jim Cousins: You will have seen
the recommendations of the inter-departmental task force on migration
statistics which was set up as a result of an initiative by the
Governor of the Bank of England. Are those recommendations being
actually implemented and what are the key recommendations that
we should look to for being the heart of the argument?
Professor Rhind: The Statistics
Commission's view on the report to which you have referred was
that these were broadly in the right sorts of areas, and we urged
government bodies to get on and do something about that. There
have been discussions and some work between ONS, the Home Office
and various other parties. I do not know, I have to confess because
I do not think it is public, that there has been a great deal
of progress as of yet.
Mr Dugmore: As a user of the data,
I have the impression that ONS have been putting a lot of effort
into wringing as much as they can out of the International Passenger
Survey, but it is a small survey of not very many people, whereas
with some of the administrative sources—national insurance
number records and so on—there is a lot more potential there
that has yet to be mined.
Q61 Jim Cousins: Professor Martin,
perhaps I could follow that very point by asking you this: in
your opening statement to the Committee you said that small samples
were a problem. The adaptation that the Office for National Statistics
has made about the use of the International Passenger Survey,
which is our principal source of international migration information,
has been to adjust it for the Labour Force Survey. What is your
view of that, as someone who is concerned about small samples,
and also of course that is correcting the imperfections of one
small sample by an even smaller sample founded on a different
basis?
Professor Martin: This is some
of the work on experimental statistics that has just been referred
to. The difficulty here is that you are pushing the methodology
to the limit but fundamentally not collecting much extra concrete
information. We are unlikely to be satisfied with the migration
data that we get from that kind of route simply because we are
using very small samples. Also we are asking people about their
intentions when they arrive into the country with the International
Passenger Survey, and so a long time interval then may ensue,
they may not go where they said, they may change their minds once
they have arrived in the country as to their intentions to stay
or not. That mechanism will always be limited, even if you increase
the sample size of it or you cross-reference it elsewhere. It
is the route which starts to capture those people once they have
arrived and connects them to other sources of information—and
yet again we come back to the administrative records—that
is really going to be the only route unless we systematically
record all movements of all people (which we do not have a mechanism
for doing) which is going to tell us about what people have actually
done. That is often very different to what they state when they
arrive and adjustment for other surveys is not really going to
solve that problem. What we have got is an improvement, as you
say it is an adjustment, but it is still on a fundamentally limited
basis because we do not have a mechanism for collecting the thing
we really want to know.
Q62 Jim Cousins: The Office for National
Statistics have themselves set out fairly clearly the kind of
people who are not contacted at all in the Labour Force Survey,
or who, even if they are contacted, do not give information. Do
you think that that is a proper summary of the situation and what
is your view of any moves that are being made to improve the quality
of the Labour Force Survey?
Professor Martin: I think it is
probably a fair assessment of the survey, but it is very difficult
to take a labour force survey and to use it for a purpose which
is not what it fundamentally has been designed for, so again we
are using methodology where it is as far as you can go with the
sources we have got. The fundamental problem that the ONS have
is that we do not have a data source for the thing that everybody
is interested in, so however far you go with those adjustments,
there will be many types of people and categories of people that
you are not going to capture in that way; they are simply not
recorded.
Q63 Jim Cousins: Professor Martin,
we have had evidence, as my colleague Mr Love has pointed out,
from local authorities like Westminster. Do you think the problems
and difficulties we are talking about here—the hard-to-count
population, the impact of international migration, the impact
of internal migration—are general problems in our gathering
of statistics or that they are particularly acute in a relatively
small number of places?
Professor Martin: I am afraid
the answer is both because of the nature of the geography and
demographics of the country, in that those general difficulties
tend to come out, to a great degree, in specific types of neighbourhood.
There are many local authorities where this is not the biggest
issue, but when we talk about our international migrants, we have
communities which are receiving large proportions of people who
come into the country and that demonstrate a big proportion of
the population churn. It is in those places that all the inter-complexity
of the general difficulties of the system come out, which is why
in several places we have said the answer is not really related
whole system-wide; it is some places receive the impacts of all
these things differentially.
Professor Rhind: That said, after
the accession of a number of countries into the EU recently, there
is quite a lot of evidence that the destinations of some migrants
is much more widespread than has been the case previously: people
going to Lincolnshire for fruit-picking and so on; quite a number
of people going to various parts of Scotland, which had not previously
experienced much by way of migration.
Professor Martin: It is not static
but it is locally specific in terms of the impact.
Mr Dugmore I think the ONS's job has
been made more difficult by the accession because at one time
one would have looked to international migrants coming to London
and the big cities particularly but now they are scattered all
over the country. Going back to the point about trying to measure
through surveys, one thing that really struck me was the use of
the International Passenger Survey to pick up on short-term migrants,
which gave an estimate of 16,000 in Greater London a couple of
years ago, and yet in the same year there were 230,000 people
registered for national insurance numbers. Admittedly, quite a
few of those then moved on again, but the contrast between the
two numbers was quite remarkable.
Chairman: We have some final questions
on the role of the new Statistics Board. I want to preface those
by congratulating on behalf of the Sub-Committee Professor Rhind
on his appointment to the Board. I certainly welcome the continuity
that gives us between the Commission and the Board. Mark Todd?
Q64 Mr Todd: The Royal Statistical
Society Study Group indicated that the new Board should "seek
to clarify and improve communications between the ONS and users
... "—and then there is a rather cryptic (to an outsider)
phrase—"rather than adding additional layers of consultation
whose relationship to operational decisions is unclear to users."
It is a very polite English communication of dissatisfaction.
How do you think the new Board should address this criticism?
And perhaps you could add some light to exactly what is meant.
Professor Rhind: I am not sure
I can add any light.
Q65 Mr Todd: Mr Dugmore probably
can say something about the origin of the criticism and then you
can tell us what is going to be done about it
Professor Martin: Perhaps I should
say there was a sense in the consultation, and I think you have
got to bear in mind also that this is a situation where people
are becoming more familiar with the role of the Board as this
inquiry progresses, that ONS and the statistical agencies in Scotland
and Northern Ireland have engaged in a huge amount of user consultation,
but it tends to be quite topic-focused and fragmented to specific
issues—consultation about a set of questions or consultation
about a particular type of output and things of this sort.
Q66 Mr Todd: So very mechanical rather
than about the principles of the use of data?
Professor Martin: Absolutely,
and there was a concern that when users were beginning to work
out how the Board might operate, that this may potentially offer
yet another channel through which there would be the opportunity
to respond on all kinds of technical issues and think about things
which were not actually high level and strategic, whereas in fact
perhaps the best outcome would be for the Board to be able to
provide a more rational steer to the way in which the consultation
processes themselves might be organised such that users know exactly
where to go, and which kinds of issues, and perhaps to have some
input in making sure that there is a completely joined-up approach.
Q67 Mr Todd: If I understand it right,
users want to be involved in issues of quality and useability
of data rather than necessarily exactly, "How do you think
this question should be asked?"
Mr Dugmore: There are different
markets here, different levels, and in some cases users really
do want to get to the nitty-gritty, and there is a limited number
of those who will absorb 100-page consultations and so on, but
I think the ones who tend to be overlooked are those who have
some short and sharp views but have not got the strength to get
involved in a really long consultation. I think that as well as
going into detail, the new Board ought to provide mechanisms for
people, "What are your ten most important Census questions",
half an hour, to try to get in connection with people who are
occasional users of data but do not spend all their waking hours
plunging into it.
Q68 Mr Todd: Or use data at a high
level?
Mr Dugmore: Exactly, yes.
Q69 Mr Todd: Professor Rhind, your
reactions to those thoughts?
Professor Rhind: Thank you, Chairman,
for your congratulations but I am afraid I cannot speak on behalf
of the Board at this stage. No matter, I am sure this will be
a big issue for the Board. The Statistics Commission has done
a number of pieces of work which bear on this. For example, we
did a study of what official statistics are actually used for.
Some of them are obvious. It turned out to be surprisingly difficult
to chase that down, especially in the commercial sector, but elsewhere
as well. We did a study of the useability of existing official
statistics by doing mystery shopper exercises. We talked to a
lot of users of one sort or another who, surprisingly to me, quite
often said they did not necessarily need any more statistics,
they were overwhelmed by all of this, and what they wanted was
some help in understanding what was there already and support
and explanation of what was there. The fact that those are relatively
new findings suggests that the communication between the user
community and the statistics producers community is in need of
some improvement, and I hope very much that the Statistics Board
will take that very seriously.
Q70 Mr Todd: You referred guardedly
earlier to this country having perhaps a less satisfactory mechanism
for agreeing data-sharing than other countries. Would that be
a priority the Board ought to address, because I think you have
discovered in this process that sharing administrative data and
interpreting it is a critical part of improving the quality of
the data that you use?
Professor Rhind: Absolutely. Tomorrow,
sir, the Statistics Commission will publish its final report,
number 38 or something of the sort, which is an attempt to summarise
the experience and lessons that we have had over the last eight
years, with recommendations being made to the Statistics Board.
One of the strong recommendations in that is exactly what you
have said; we must encourage people through various mechanisms
to share data, because we can get much greater added value from
that where the data are all linked together. There is of course
a better legislative mechanism now under the Statistics and Registration
Service Act using secondary legislation. I was very struck how
in the United States a similar situation pertains. Each combination
of data coming from different government ministries has to be
discussed and agreed and mechanisms laid out for it, but once
it is done once, then it is taken as a fairly routine situation,
and I hope very much the same thing can happen in the UK.
Q71 Mr Todd: But you are being very
optimistic there. The culture of the UK, which I think you have
hinted at a few times in answering Mr Cousins's questions about
lack of progress in addressing immigration data for example, is
that people look after their own data and manage it for their
own purposes, without a mind to how it may be broadly shared and
the protocols under which that might be done.
Professor Rhind: I think that
is true in many cases, and of course the recent disasters in terms
of loss of individual data have not helped, nonsensical as those
were, and with any decent processes they should never have occurred.
Q72 Mr Todd: I was going to say that
should be irrelevant.
Professor Rhind: That should be
irrelevant.
Mr Todd: But it can be seen as confusing
the issue.
Q73 Mr Love: You indicated earlier
on in a reply that we had a very traditional method of measuring
the population through the Census. Some of the submissions we
have received suggest that some countries are moving towards a
more regular census, perhaps with lesser information on it, and
some are moving entirely away from a census. Tell me what you
think should happen after 2011?
Professor Rhind: To expand on
what I said earlier, I think we should move towards—and
I completely take the point that this is not throwing a switch
and going from one to another—a situation where administrative
data sources form the bulk of the information and we can produce
from that small area data, supplemented by surveys as appropriate,
to find the things that you cannot easily get out of administrative
data sources.
Q74 Mr Love: Do you disagree, Professor
Martin?
Professor Martin: I would agree
absolutely with the first part. My concern is the issue of how
one throws the switch, looking at the experience of other countries.
I believe the Americans in 2010 will conduct a very thin census
which is an attempt at least to count everybody and calibrate
the results without doing a detailed survey. I would say that
perhaps at the moment there is still an argument for doing that
on at least one more occasion to validate the way in which we
have got going with the administrative records, but I think there
is a rightness about using those records as the major basis which
we move to, and I would certainly agree with the major thrust
of that.
Q75 Mr Love: Mr Dugmore, would the
users of the data find it acceptable that we give a more accurate
measure of the numbers but perhaps less detail than is contained
in our current traditional Census? Would that be unhelpful to
you?
Mr Dugmore: If you look across
government, many of the topics of interest are covered in the
administrative files—if you look at Revenue and Customs
and so on and so forth—so there may be some losses there,
but having the regularity and the speed would be real pluses.
I think that it does need to be pushed. The risk is that it is
just seen as a passive thing that in potential it could be done
but nobody will get round to doing it because they get no plaudits
for it and they might be taking some risks in doing new things.
Chairman: We are going to leave it there.
On behalf of the Sub-Committee, can I thank the three of you for
coming along this afternoon and for getting our inquiry off to
such a good start.
| 
