Examination of Witnesses (Questions 1-19)
PROFESSOR DAVID
RHIND, MR
RICHARD ALLDRITT
AND MR
MARTIN WEALE
2 NOVEMBER 2005
Q1 Chairman: Professor Rhind, may I welcome
you back to the Sub-Committee and perhaps you would introduce
yourself and your colleagues?
Professor Rhind: Thank you, Chairman.
May I start, even before that, with apologising for our late arrival.
We have been camped outside for a little while, so, sorry about
that. On my left is Martin Weale, a Commission member and on my
right is Richard Alldritt, the Chief Executive of the Statistics
Commission.
Q2 Chairman: Thank you very much. Now,
we have some fairly rapid fire questions for you today because
we, too, have a time constraint from divisions in the House a
little later on. Can I start by asking you about the Framework?
The Framework for National Statistics is, I think, five years
old now. How effective do you think it has been?
Professor Rhind: I think there
is no question that it was an improvement, on the fact that we
did not have anything comparable to that beforehand and it has
certainly paved the way for the code of practice and the Commission
and various other, I think, improvements, but for us it has a
number of weaknesses and we therefore welcome the fact that there
is going to be a review of this and we have got a number of really
rather crucial points which we would like to see addressed in
the process of reviewing this and I can go through those, if that
would be helpful.
Q3 Chairman: I think we will come
on to those in a moment but it is useful to have your general
summary first. One of the things you have discussed in previous
appearances before the Sub-Committee is the need to secure and
enhance public confidence in official statistics. The latest ONS
survey showed that only 17% of those surveyed thought that official
figures are produced without political interference, and only
14% thought the figures were deployed honestly. Why do you think
that is? What do you think we can do to improve that?
Professor Rhind: First of all,
it is very recent that we have had any idea of how good and/or
bad the situation was and I think I am right in saying that the
UK is the first country in the world which actually tried to look
at how good or how bad it is. So we are beginning the process
but we have a long way to go. Our work and that of the Office
for National Statistics suggests that there are multiple causal
factors which merely shift over time and we are in the throes
of putting together—a plan may be too grand a phrase for
it—a way of actually trying to address the issue. It is
clear, for example, that those who are closer to official statistics—those
who are expert commentators, Members of Parliament and indeed
various others—do not believe that most of the statistics
themselves are actually a serious problem. In many cases, it is
the way they are interpreted. If you ask the general public, the
general public's view is much worse than the experts', if you
like, in that particular area. So we have multiple targets that
we have to address. Part of it is a perception issue, no question
about that, and that of course involves the media in a number
of important respects. So we have four or five different areas
in which we have to tackle the issue of public trust. Fixing one
of them is not, I think, a solution. On top of that, I would say
that it is not a five minute job, either—we were recently
in Canada and I was very struck how, there, it took something
like a twenty year period for the Statistics Canada people to
recover from some serious issues that they had in the 1970s and
1980s. So we have the embryonics of a plan which we are discussing
with National Statistics people and others, but it has to cover
a number of different points.
Q4 Chairman: My colleagues may wish
to pick that up. Finally from me, the Treasury Committee generally
and this Sub-Committee has from time to time looked at this whole
issue of re-classification, or the decision taken by the ONS in
some of the more controversial areas such as public finances and
we have been quite struck by the differing periods of time taken
by the ONS when it is faced with a classification or a re-classification
decision. Have you looked at that particular area? Do you think
there is any evidence that the time taken to decide on classification
actually depends on the direction of the proposed change, if I
can put it like that?
Professor Rhind: We have certainly
looked at a number of particular cases. In some cases, they have
turned out not to be re-classifications but correction of errors
or other things, but may I pass over to my colleague, Martin Weale?
Mr Weale: Yes. On the issue of
re-classification, obviously different sorts of classifications
are of different types. What was initially presented by the ONS
as a re-classification of the treatment of roads in the national
accounts, they subsequently changed their minds and said it was
a correction for double counting. Now, taking that, we have no
reason to doubt that the second description was the correct one,
and incidentally the initial favourable impact on the Government's
budget—favourable from the Government's point of view, was
subsequently halved by a revision in the national accounts later
in the year although that received much less publicity—here
I am quoting Len Cook on that change. So I think that the difficulty
in saying that there is a bias one way or the other, that favourable
changes go through quickly, unfavourable changes are perhaps delayed,
is that some changes, some things, are much easier to address
than others. I know, I am quite sure we would all be shocked if,
having found that there was a mistake, that there was double counting,
it was not corrected as quickly as possible. I think on the sort
of particular issue that is in some people's minds, the treatment
of the Private Finance Initiative and whether elements of liability
associated with that should be treated as a Government debt or
not, that is something where the ONS seems to be taking its time,
but of course, at the same time, the sort of political issue,
whether or how much it brings the Government's borrowing closer
to the upper limit of the fiscal rule 40% of GDP, that is also
a slow moving issue; it is quite a way off 40% of GDP at the moment,
and if borrowing goes on it will get closer, but certainly I think
it would be wrong to say—we have no reason to think and
the arithmetic would not support it, either—that slow progress
of that was done with the aim of stopping crossing the debt limit.
Q5 Chairman: Sure. I was really wondering
about decisions that seem to get postponed or run on for ever,
whether there was some way the Commission would note in its Annual
Report that there are classifications still awaiting, if you like,
that there should be some end date or at least some notification
from you that there are outstanding matters and there was a period
beyond which you would expect them not to be left outstanding,
if you like.
Professor Rhind: As a matter of
principle, if it is within our locus we should be monitoring those
and trying to ensure that undue delay does not occur. This may
be an expression of failure, but I do not recall any at this particular
moment.
Mr Alldritt: I cannot recall any
being raised with us directly; if there were, there are steps
we could take to investigate and would do. It is quite possible
that an issue of that kind has been raised, but I cannot seem
to recall it at the moment.
Professor Rhind: Can we offer
you a note on that?
Chairman: Sure. That would be very helpful.[1]
Q6 Peter Viggers: The Framework for National
Statistics when published in June 2000 said that it would be reviewed
in five years or so, which would take us through to the summer
of this year. What progress has been made in reviewing the Framework,
to your knowledge?
Professor Rhind: We are awaiting
the Government's statement of the launch of this and the details
of the terms of reference. We have had some discussions with HM
Treasury and, indeed, with the Office for National Statistics.
We are, as I say, still waiting for the formal announcement of
who is going to do this and how it is going to be done, and what
it is for. On 16 November the Commission has its annual Open Meeting
and the theme this year is the Framework for National Statistics,
at which we will have the National Statistician and various other
people speaking and the Treasury has welcomed that as an input
into the review but we very much hope and wish that this gets
on and progresses very rapidly.
Q7 Peter Viggers: I would submit
that you are, whilst a very small body, very strong in specialist
knowledge and I would hope that you will be centrally involved
in the review.
Professor Rhind: Thank you. We
certainly believe we have something to contribute and we have
already written to the Financial Secretary to the Treasury with
a number of points that we believe the Review has to consider
and are matters in the public domain so we can talk about it,
if that would be helpful.
Q8 Peter Viggers: Would I be fair
in pitching in rather higher than that and saying that your mission
statement might be summarised as saying that you would like to
see an accurate and appropriate Framework for National Statistics
and that very much centres your objectives?
Professor Rhind: Absolutely. We
think some of the things in that are not too difficult to achieve—we
have all learnt, I think, from the experience of the last five
years—a number of things are improvable. There are some
big issues in there which will be more difficult, but some progress
can certainly be made relatively easily.
Q9 Peter Viggers: Well, speaking
personally, I would like to strengthen your arm in those discussions.
Turning to the code of practice, what revisions would you like
to see made to the code of practice?
Mr Alldritt: I think there are
a number of changes we would like to see to the code of practice.
The code is quite a complicated document; it comes in 13 booklets
and it covers a great range of things. There are only a small
number of those which are very specific about what action is acceptable
and what action is unacceptable. There are a lot of areas, for
example, on how to manage the quality of statistics and the consultation
processes that should take place in deciding what statistics to
collect and in relation to what independent commentary with statistics
looks like, where it is very much left at the discretion of government
departments producing the statistics to interpret the code. We
would like to see that ambiguity and the exceptions and exemptions
taken out of the code, so there is quite a lot of change we would
like to see.
Professor Rhind: If I may, Chairman,
just add to that—the 13 different sections of the code—the
code itself and 12 protocols—almost all the complaints we
have had relate to one of those 12 areas and that is, in terms
of release practices, information being released early, or inappropriately.
You can either take the view that the rest of the code is exemplary
and no-one has a problem with it or, alternatively, the view that
I think we take, which is that significant portions of the rest
of the code are so ambiguous and aspirational that it is very
difficult to pin down whether the code is being followed or not.
Our experience of visiting a number of government departments
over the last few years is that the interpretations of the existing
code made by different government departments, really does vary
a great deal and almost all of them—in fact, all of them,
I think—believe that their interpretation is perfectly fair
and reasonable. That suggests to us that we need to re-visit the
code and make it a clearer and tighter document.
Q10 Peter Viggers: So would it be
a fair paraphrase of that to say that some departments are selective
and early in the use of some statistics?
Professor Rhind: I am sorry; I
am not trying to avoid the question, but there is a very interesting
issue which at the moment I do not believe anyone knows how to
handle, which is that the distinction between statistics collected
by surveys, which is the traditional method, and for which the
code was originally written, and statistics which are increasingly
flowing out of administrative and bureaucratic systems and release
practices in regard to things that you are collecting every week,
every day, are a very difficult and not thought-through challenge
yet. So forgive my being complicated, but there are these two
different sorts of statistics and the code really relates to one
of them and we need, all of us, to think how we can cope with
the other sort of statistics.
Q11 Peter Viggers: You have in the
past been prepared, under cross-examination, to agree that certain
departments are more culpable than others. Are you prepared to
name government departments which you find are frequently among
the worst offenders?
Professor Rhind: At the risk of
sounding complacent, I have to say that the problems we have had
with clear breaches of the code this year are substantially fewer
and we have reported a number of cases in our Annual Report, I
think annex D from memory. You will see that those are fairly
modest. We have investigated all of the ones which seemed to be
breaches and in some cases there were, we think, genuine mistakes
made and we are pretty confident that we are on a path of improving
that, so the obvious issues are less serious than they have been
and we have put them out in there; the non-obvious issues are
about this big problem about how on earth do you know if you are
transgressing a code which is ambiguous and, in some cases, simply
aspirational. So I think there are problems, but the trouble is,
we cannot pin them down until we actually set the guidelines a
bit more clearly.
Q12 Peter Viggers: Your small body
is there because it is objective but, if it is to retain a reputation
for objectivity, it must occasionally be judgmental.
Professor Rhind: Absolutely. And
there are a number of government departments that have not always
totally agreed with our observations on what they have done in
times past.
Q13 Peter Viggers: To what extent
are the worst offending departments named and shamed? What sanctions
do you have?
Professor Rhind: Our standard
process, is to do some kind of study, get some evidence, often
in response to a complaint or a query and then to go to the department
and say, "On the evidence we have available, this is what
has happened, it appears to us to breach the code or some other
criterion; do you have any observations on that?". If that
department then says, "Oh, dear, clearly we got that wrong
and we will write it down to you and say that this was a mistake
and it will not happen again" then I think that is a reasonably
satisfactory situation as long as we monitor that it does not
happen again. If a department were not to take us terribly seriously,
we have two sanctions; the first is that our operations are largely
in the public domain and our correspondence is in the public domain
and our reports are in the public domain, and the media take quite
an interest in this, so that is helpful—usually helpful—in
terms of publicity, and the second sanction, Sir, is to come back
to this Committee; through your Chairman you have kindly offered
to assist us in times past and we would not hesitate to do that
if that seemed the necessary step to take.
Q14 Peter Viggers: The latter step,
really, I assume you have not taken that.
Professor Rhind: We have not taken
the Armageddon option yet, Sir!
Q15 Peter Viggers: How many bodies
are hung out to dry, or how many heads are on spikes?
Professor Rhind: We have had,
as I said, I think, earlier, rather fewer problems of the discernible
kind this year and I think there are 15 cases of some issues in
the Annual Report, of which five or six are where we believe there
were breaches of the code. We followed each of them up and were
happy with the result, so on that basis we did not feel it was
necessary to come to you.
Q16 Peter Viggers: How significant
is the influence of ministers over statisticians in government
departments? To what extent is the principle that statistics should
be free from political influence compromised? Is it possible to
answer that general question?
Professor Rhind: We think that,
first of all, there is a widespread perception issue that ministers
interfere in everything and that that certainly is manifested
in the response that the Office for National Statistics got to
their Omnibus survey where the very low percentage of people believing
that the figures were not fiddled came out. I think it is clear
that ministers do have some strong influences on statistics and
examples of that are, first of all, that ministers decide which
statistics are to be classified as national statistics and thereby
to be created following a process which is related to the code
of practice and supposed to meet the code of practice. More than
half—it is rather difficult to be precise—of British
Government statistics are not produced by the Office for National
Statistics. In other words, they are produced in whatever manner
the department deems to be the right manner to create those and,
as I say, the ministers decide which ones are to be so labelled
and to meet the code of practice. Ministers also decide a number
of issues about release practices within the code of practice
and they also, in some senses, decide which statistics are to
be created, or be collected. There is no question that ministers
do have some real impact on this and we have always taken the
view, for example, that ministers should not decide on which statistics
are to be produced in accordance with the code of practice.
Q17 Peter Viggers: If the National
Statistician were granted greater authority over government department
statistics, would this help to neutralise ministerial influence?
Professor Rhind: Certainly we
think it would be a helpful step forward if the National Statistician
had some greater authority over statistics created across the
piece. As I said a moment ago, I think probably more than half
of official statistics are created outside of the Office for National
Statistics, which is where the National Statistician has direct
line management responsibility. She has influence in various forms
across statisticians elsewhere as Head of the Government Statistical
Service. She certainly has some influence in the development of
staff and so on, but her real authority is pretty modest.
Q18 Peter Viggers: Often the statistics
are all correct—I am thinking of crime statistics, a well
known case. Every time the crime statistics come out, Opposition
parties say crime is up, pointing at one figure and Ministers
say that crime is down, pointing at another. Does anybody have
any role in blowing the whistle on this dispute and saying, "The
whole piece is not being reflected"?
Professor Rhind: That is a very
timely question. As it happens, we were having presentations by
Home Office statisticians this morning about explaining the nature
of the different sources of evidence and what they mean, and so
on. I think generally we believe that the quality of British statistics
as a group, insofar as you can talk about all of them, is probably
getting a bit better. In some respects, actually, they are as
good as anywhere else in the world, but what we are most worried
about at the moment is exactly the issues raised; in other words,
reasonable interpretations out of the statistics. We have seen
a number of examples where what has been said, for example in
the media, bears no relationship to what came out in statistical
releases and descriptions of the statistics. I think there is
a very major issue about whom you can trust to interpret statistics
and how do you get simple, clear but reasonable messages out of
it.
Mr Alldritt: At present we are
in the early stages of a review of crime statistics, which by
Christmas we expect to have pinned down about five or six issues
that we will want to look at in greater depth. It seems highly
probable that one of those is going to be exactly this issue of
how to extract messages or interpretations from the statistics
convincingly, in a way that will command wide confidence. We will
be expecting to make recommendations in this area.
Professor Rhind: May I turn to
my colleague, Commissioner Weale.
Mr Weale: Could I just add another
point on that note? This seems to me to demonstrate very clearly
the importance of having a means of disseminating statistics which
is not only independent but also visibly independent—that
the phenomenon you have described over crime statistics and the
way they are presented in different ways is, in my mind, associated
with the fact that statistics at large are not trusted by the
public and I think that would be helped by having more visible
independence in their presentation and interpretation.
Q19 Peter Viggers: So rather than
presenting a large well from which people can select their own
facts, you go some way to sifting this and producing some kind
of summary?
Mr Weale: Well, you do that, but
it needs to be visibly independent of the political process.
1 Ev 11. Back
| 
