Examination of Witnesses (Questions 20-39)
MS KAREN
DUNNELL AND
MR MIKE
HUGHES
19 MARCH 2008
Q20 Dr Turner: To what extent does
the UK Statistical Service sample surveys rather than conducting
national registers, as most countries do—national databases?
Ms Dunnell: For some purposes
we need registers, and we have registers. For example, we have
in ONS, and maintain for use across Government, a register of
businesses in the UK, which we use to produce statistics and also
of course to use as a sampling frame for all the business surveys
that feed into the national accounts and balance of payments and
things like that. Very often, because one of the great advances
in the 20th century was the development of sampling, the statistical
theories about sampling, it is much more cost-effective to carry
out a sample, take the sample and collect the information from
them, and then make an estimate. You could then say quite precisely
within which boundaries your estimate is likely to fall. For many,
many purposes, this is what we do, because it is cost-effective
and quicker, and it is often higher quality because you are collecting
the information for a statistical purpose, whereas with quite
a lot of registers and administrative registers, for example,
the information is collected for administrative purposes and may
not be of sufficient quality for the kind of statistics that we
produce.
Q21 Dr Turner: Where do consultants
fit in to the government's statistical work; to what extent do
you employ them; and, when you do employ them, who determines
the methodology and reviews their work?
Ms Dunnell: We do not use consultants
very much for the actual statistical work. Where we do use consultants
is for quite a lot of the IT side of things. We are often trying
to develop things when we have not got the right skills. On the
other hand, we do have contracts with academics quite often to
do things like peer reviews, as Mike has mentioned, but also to
carry out a particular type of analysis or to help us carry out
development work. For example, we are working at the moment on
planning a new major survey of disability in the community, and
we are using an academic on a contract to help us develop that,
partly because he used to work for us and has a lot of expertise
in this field, but partly because we do not have that particular
skill ourselves. Also, of course, we work extensively with organisations
like ESRC and will often put out to contract pieces of work, for
example, in relation to the census, when we usually get a university
contract in collaboration with the ESRC to provide services to
academic users on the census. Similarly, we have a contract with
Southampton University to run an MSc, which many people in the
GSS go on. So there are a wide variety of those things. We also
use contracts for some of our data collection on occasion.
Q22 Dr Turner: It occurs to me that
you obviously must make fairly heavy use of IT.
Ms Dunnell: Yes.
Q23 Dr Turner: You know, of course,
the disastrous history of governments of all colours in promoting
large national IT schemes. Do you get involved by the Government,
or do you have any input to Government considerations of large-scale
IT schemes, because they must be relevant to you as presumably
you would want to access information from them from time to time;
so are you consulted?
Ms Dunnell: Yes. In fact, every
Government department has something called a chief information
officer, which is a bit of a misnomer, but basically head of IT,
as I call them. The head of the whole Government—there is
a head of profession, rather like myself—who works in the
Cabinet Office, and he runs a council of all the heads across
Government, in the same way that I run my head of profession department
meetings and so on for statistics; so we are very plugged in to
that. In fact, one of the things that the CIO council has done
is to create a shared service for basic IT infrastructure in Government
departments, and we were one of the pilots for that; and indeed
just signed up to it. Our basic infrastructure is actually moving
to this service called Flex at the end of this month. We do get
very involved in that. It is for providing all our basic telephony
and equipment and storage and security and so on. The thing that
we keep very much to ourselves, and will continue to do, is all
the work we do on statistical computing, which involves this worldwide
sharing of good practice and expertise and development about how
to best take that forward. On things like data access, we are
very plugged in, and I am a member of Gus O'Donnell's security
advisory team, which is advising the whole of government—and
Mike is my substitute on it—about how to do much, much better
with data security. Of course, we have all had to do audits of
our security arrangements and so on. Given that keeping personal
data confidential is absolutely at the fundamental core of our
business, we are quite active in advising and helping to get the
standards across Government right.
Q24 Dr Turner: You have not lost
any disks?
Ms Dunnell: We have not, no.
Dr Gibson: How would they know?
Q25 Mr Boswell: While we are thinking
about IT, presumably this means that you or collectors working
for you make increasing use of direct inputting of electronic
information on site. Perhaps you can comment on that, and, if
so, does that lead to improvements in accuracy as well as timeliness,
and also on the security side is that equally satisfactory?
Ms Dunnell: Doing data collection
using electronic methods has many, many advantages, one of which
is that as you enter the data you can do all kinds of checks on
it and the signposting to the next question is all automatic;
so you tend to get much better quality data. We have very, very
strict security arrangements for the transferring of data into
the office. One of the reasons why we are going quite slowly with
Internet collection is because of the security side of things;
we have to be absolutely sure that data coming down the line cannot
be interfered with or looked at in any way.
Q26 Mr Cawsey: I would like to ask
about how Government policy should be more evidence-based, and
the role that you can play in that. You are probably aware of
Michael Blastland and Andrew Dilnot's piece on the tiger that
is not, where they argue that the tendency to ignore the need
for statistical verification is only now beginning to change in
the UK. Do you agree that this verification has not until now
played a part in the formulation of Government policy?
Ms Dunnell: I do not necessarily
agree with them, no, because the Government's statistical service
produces an enormous range of statistics, which are used across
the piece every day in Government departments, local authorities
and all kinds of bodies for planning, evaluation and all kinds
of parts of the policy process. I am a great admirer of Andrew
Dilnot, I have to say, and he has picked out some examples from
a very, very wide range of statistics to make some very important
points. I think that some of the stuff he has written in there
about variability and so on will get everybody to read it because
it is very, very important. However, I think it is very, very
obvious now that statistics are absolutely vital for government
policy at all stages. We see many examples, and most recently
in the budget speech in the work that has just come from the prime
Minister's Delivery Unit, looking at the future of policy. That
is a statistical picture about what is going on in society and
the economy and is at the forefront of thinking about policy development.
This does not mean, however, that this is perfect in every case
and in every department, because a lot of policy-making is highly
political and often happens at great speed. Our role, and one
of my ambitions, is to make sure that we can improve the impact
and influence of statistics and statisticians in all sorts of
Government policy-making at both national and local level.
Q27 Mr Cawsey: What I am interested
in is a bit chicken-and-egg really. Is it that the Minister or
the department will say to you, "We have got this whiz of
an idea for a policy, so give us all the statistics to back it
up"; or do you go with a load of statistics and say, "This
policy needs to be changed"?
Ms Dunnell: Do you want to try
that one, Mike!
Mr Hughes: I think the syndrome
that you outline probably did prevail fifteen or twenty years
ago, but, as Karen has said, we try extremely hard to get involved
in the policy process as early as possible.
Q28 Mr Cawsey: Do you ever create
it, is what I am asking; do you ever start the process and say,
"Look at these statistics; we are highlighting ... ."
Ms Dunnell: Yes.
Mr Hughes: Yes. When I was head
of profession in transport, we were producing a range of statistics
for the board and actually chose to challenge some of the things
that were happening on transport policy at the time. Is it right
to be spending so much money on rail safety when probably there
are five people killed a year, and yet eighty people a day die
on the roads? It is that kind of dichotomy.
Q29 Graham Stringer: It did not change
the policy, did it?
Mr Hughes: No, it did not, I would
agree with you; but at the end of the day we were putting those
points up to the top of the office.
Q30 Mr Boswell: I am often saying
privately that I do not think social trends arise and hit us in
the face until it is too late. Are you really in the forefront,
in your advice to Ministers, in being able to spot social trends
that previously perhaps had not been anticipated, and then actually
drawing their attention to it hopefully getting them to do something
about it before it is too late?
Ms Dunnell: I think it is a very
important part of our role. For example, something that the ONS
is responsible for on behalf of all other Government departments
is the monitoring and estimating the size and structure of population.
For example, the aging of the population is a very important phenomenon,
which needs to be taken into account in every bit of policy that
happens both nationally and locally. I think that actually statisticians
have done a great deal to make sure that everybody does not forget
about the aging of the population, because it is very, very important.
That is a very good example. Similarly, on issues like the situation
with family status, the whole evolving, monitoring system for
looking at, for example, the big changes that we have had in marriage
and cohabitation patterns came very much from the statisticians
attempting to measure a rapidly changing situation, and then making
those new statistics part of what everybody now accepts as something
that we update all the time. There are similar things on the economy,
which is much more difficult in some ways because it changes quite
rapidly; but again we feel very strongly that it is up to us to
draw attention to things that are happening in the economy, which
people may not be putting enough emphasis on.
Q31 Dr Gibson: Do you have an input
into the growth percentages of the economy?
Ms Dunnell: Yes—well, we
produce all that every quarter, yes—all of the national
accounts, yes.
Q32 Mr Cawsey: If a statistician
finds a piece of information and passes it on to the Government
and then finds it is ignored, what can they do to correct that?
Ms Dunnell: That is very similar
to the thing that Mike said: if you are a statistician in a department,
or in my case in ONS, and you feel that policy is not making enough
or taking enough account of some change or something which you
think is quite important, then it is one of our professional responsibilities
to keep putting that piece of information in place.
Q33 Mr Cawsey: But if you keep putting
it into a red box, but it never gets any further, would you put
it into a red to eventually?
Ms Dunnell: We try to get all
of our statistics, to be honest, into every kind of newspaper—that
goes back to my answers to some of your earlier questions: in
my view, the more that everybody in our community understands
and appreciates statistics and their usefulness, the better. We
would not do a special press release, particularly aimed at any
particular newspaper; but we would take every opportunity, within
our normal publication process, to draw attention to those changes
that we feel are important.
Q34 Mr Cawsey: What if it was the
other way round? What if a journalist approached you and said:
"The Government has just announced this, this and this; what
is the statistical evidence for it?" and you knew there was
none, would you say that there is none?
Ms Dunnell: I am trying to think
of an example. We give all kinds of advice to people all the time
over the telephone. Personally I do not think—and I do not
think most statisticians in the Government's Statistical Service,
would make a judgment like that because it would be quite dangerous.
What they would try to do is say, "Here is the statistical
picture of something as we see it" and present the thing
afresh.
Q35 Dr Harris: Dangerous to health,
dangerous to career—what do you mean by "dangerous"—to
give an honest, open view?
Ms Dunnell: Because we try not
to comment particularly, and one of the statistician's basic codes
of practice is not to comment about things political. Sometimes,
if you have made a comment about a particular policy, it can be
interpreted as political. Our task is to be objective, full of
integrity, and apolitical.
Q36 Graham Stringer: Can we come
back to the point about the large amount of money that the previous
Deputy Prime Minister wanted to spend on transportation warning
systems all over the rail system. It was very high profile. John
Prescott was very keen on it. How, as a real example, did you
deal with the information you had which said, "spend this
money on road improvement schemes or whatever and it would save
more lives"? How did you deal with that?
Mr Hughes: It is going back quite
a way now, so I have to think about it more in the abstract than
remembering the detail.
Graham Stringer: I think the detail would
be much more fun.
Q37 Dr Gibson: Could you send us
the detail?
Mr Hughes: The way that we were
dealing with these sorts of things was to put together a regular
briefing note, a package of material called "transport trends"
that would go out regularly to the management board for them to
look at. In that particular case the decision was very much swayed
by political imperatives and that was why it ended up the way
it did.
Q38 Graham Stringer: I think what
Ian was asking about—in that situation where the statistics
pointed in a different policy direction, would you have gone to
the Sun or the Mirror and said: "Here are the
statistics; look them up yourself; this policy will cost lives"?
Mr Hughes: We are civil servants;
we work for Ministers in departments, and that is the process
by which we work. I do not think it would comply with the kinds
of principles that Karen has just outlined that any of us would
do that sort of thing. That information is not buried in the depths
of departments. That is analysis that anybody outside can equally
do.
Dr Gibson: You must forgive us if we
are slightly jaundiced about getting information! We spend a lot
of our time trying to get information from people in Government
departments.
Q39 Dr Iddon: I want to move on to
targets. Do you have any role in the formulation and measurement
of targets the Government sets?
Ms Dunnell: Again, it is quite
varied between departments whether or not statisticians get involved.
One of the things that we are trying quite hard to do now is to
ensure that statisticians do get involved and that they get involved
at the early stages of the formulation of targets. One of the
particular problems of course is that people set targets that
are quite difficult to measure, and then quite a lot of short-term
and sometimes not very well thought-through things happen in order
to measure the progress of targets, and these are the kinds of
things where, if statisticians do get involved early on, they
can be avoided. It is very important.
| 
