Examination of Witnesses (Questions 57-128)
MS JIL
MATHESON AND
MR GLEN
WATSON
19 NOVEMBER 2009
Ms Matheson: If I could start
by describing the process by which we got to the recommendations
which we have, which was four years of very detailed consultation
with a range of users: users in central government, local government,
voluntary sector, business—we have heard about some of that—plus
extensive testing with thousands of members of the public. The
balance that we have to make depends on a whole set of criteria
about what is the need for this information—not just what
is the desire, but what is the need for it—what is unique
about the census that means information that is collected from
people at the same point in time, so providing information about
small areas and about the national picture, are there any alternative
sources, could the information be provided in any other ways,
how good is the quality of the information that people can provide.
That is on the demand side. It is important to balance that with
what is the ability and acceptability of that with members of
the public. All of the parts of the proposed questionnaire have
been tested, they have been discussed with users and also tested
with members of the public. What we have come up with is the professional
judgment on where that balance lies so that it provides information
that is needed by everybody in society but also matches the ability
and willingness of the public to provide that information. There
were some tough judgments. We could have had a form much longer
than the one that is proposed. You will have heard that certain
people are disappointed, and that is because it matters. There
have been requests for additional questions which we are not recommending.
Some people have said they would like the questionnaire shortened,
but there is no agreement on how you would shorten it. What we
have got before us is the basis of four years' work and our professional
judgment on the right balance.
Q57 Chairman: That is all very interesting
but not quite the answer to the question, which was how open are
you to further change?
Ms Matheson: If you are asking
about the process, this is our professional judgment on what the
best balance is. It is for Parliament to decide on what the final
content will be between now and whenever the Order and then the
regulations are passed. This is our advice and our recommendations.
Q58 Chairman: When you produced this,
as you did recently, and you read the headlines about "A
Snooper's Charter" and all that, you must have groaned, because
you did all these years of testing and research and then some
idiot politician comes out with remarks like that.
Ms Matheson: I did groan. Because
there is a decade's worth of decisions that are made by all sectors
that impact on everybody then it is important that we get the
population estimate right, that is the core of what the census
is about, that we maximise response, we make it easy for people
to fill in and we explain the information they give us is secure
and safe, it is held confidential for 100 years and has been since
the census was begun in 1801. Being able to explain that to people
is the essence of the census operation, which is why in response
I did write to newspapers to say I thought the reporting of that
was unfortunate because we have to get those messages over to
people: it matters; the information is confidential; and it is
safe and secure.
Q59 Chairman: If we leave that response
aside, it is the case, is it not, given the last session, it is
these areas of sexual orientation, religion and income which are
the ones that have been most testing for you?
Ms Matheson: Each of them separately
has been tested almost to destruction. We have been absolutely
determined to have a whole series of tests, which have included
postal tests, which are voluntary, looking at the difference in
response rates whether you include or exclude an income question,
for example. Plus face-to-face interviews with large numbers of
people, talking to them about the form and what their attitude
is to particular questions. We can pick up on the individual ones
and what we found out about those. That has been the process that
we have been through.
Q60 Chairman: The voices that we
were just listening to are authoritative figures, they know about
censuses, questions and data. They are voices that need to be
attended to, are they not?
Ms Matheson: Indeed, and they
are very much part of the process that we have been going through.
We have had extensive conversations with the people you have spoken
to, and many others. We get expertise from academics and users,
and also from census offices in other countries. There is a network
of census offices so that we can learn from each other, draw experience
from each other and make sure that our developments are in line
with experience that we can draw from elsewhere.
Q61 Chairman: I am sure colleagues
will want to ask about the different areas we asked about just
now. Glen, do you want to add anything before we start?
Mr Watson: No, not at this stage.
Thank you.
Q62 Kelvin Hopkins: If I may pursue
the question about religion once again. I have concerns that the
question is not formulated in the right way and that for many
people the question would be uncomfortable or difficult to answer.
You have had representations from the British Humanist Association
and some of their suggestions raise their own complexities and
difficulties. As I said to the previous group of interviewees,
what about if you just put in front of that, "If you have
a religion, what is it?" That would overcome many of the
problems. It seems to me to be a compromise that would go some
way towards satisfying the criticisms.
Ms Matheson: Glen can probably
give you more detail about this. One of the things that we did
do in the last year was to test an alternative question. It was
not that one. It was an even more open question, which was simply
to say, "Which of these groups best describes you" because
of the difficulties you were discussing earlier that if you used
the word "religion" or "faith" or "belief"
it confuses. We did test another version which was even more neutral.
What we found from that was even that more neutral question was
not one that significant proportions of people felt able to answer,
they did not quite know how to answer it. It did not make much
difference either to the proportions who said they were Christian
or Muslim or the major groups. On the basis of that, despite the
discussion that I know has been had previously that the question
there is potentially leading, we found it was also one that people
understood. We have not found an alternative which performs better
and also does what an awful lot of users have said they want,
which is continuity, time series. Being able to say how things
have changed over time is one of the core elements of the census.
I know from previous experience that whenever we present census
results, one of the first questions people ask is, "And how
has it changed?" There would have to be very good reason
and an alternative tested that was better and we were able to
understand what it was measuring in a better way, and we have
not found that which is why we are recommending it staying on
the grounds of continuity.
Q63 Kelvin Hopkins: As someone with
an interest in statistics it is important to have consistent time
series so that they become meaningful. The BHA has said it overstates
religiosity. Taking my suggestion further, if you had a series
of religions and "other", and then "no response",
the "no response" would be the easy way out for those
people who do not want to say very much about their beliefs, but
you would get a very positive response from all those people who
did have religious beliefs. A simple question: "If you have
a religion, what is it?", a series of religions, "other"
and then "no response". That would seem to me to be
a way of overcoming the problem.
Ms Matheson: We have not tested
that but can I give you an instant reaction, which is that I am
always wary of questions where you are interpreting a blank, ie
if you do not have a religion you do not fill anything in. At
the moment, the very first option is to say "no religion".
That is the very first item that appears in the list. If somebody
simply does not fill the question in I am not quite sure how we
would interpret that.
Q64 Kelvin Hopkins: Your predecessor,
Karen Dunnell, seemed to suggest that the question was still open
on religion and even at this stage there might be a possibility
of adjustment, and I just hope that my suggestions might be considered.
There is the other question about people who have religious heritage,
not just Jews or Sikhs but I know many people who are proud of
their Catholic heritage, for example, an Irish Catholic identity,
but are not actually religious. If you included in the ethnicity,
"Ethnicity or cultural heritage", you could shift that
complication out of the religious area and put it into the cultural
heritage/ ethnicity area. Would that not solve the problem?
Ms Matheson: I am going to ask
Glen. Because of the issues that have been raised he has been
responsible for the detailed testing that has been done over the
last year.
Mr Watson: Can I first of all
comment on what Jil was saying about the testing of the different
religion wording. We have tested six or seven different possibilities
over the last three or four years and the version that we tested
that was completely open-ended, "Which of these best describes
you?", showed less than a two percentage point difference
between those saying "Christian" under that question
and under the "What is your religion?" question and
showed no difference at all in the proportion of people saying
"No religion". That was enough to satisfy me and our
people that, although grammatically it might look like a leading
question, in terms of the way people answer and interpret it,
it is not and there is no evidence to support that. That satisfied
us that it was the right construct given that we wanted to have
continuity with 2001. On the Sikh, Jew and ethnic and cultural
heritage question, we do keep the Sikh and Jew as tick boxes within
the religion question and our research has shown that we get better
counts of people who consider themselves Jewish or Sikhs in this
way. For example, when we introduced the Sikh tick box into the
ethnic group question with our testing it caused confusion, particularly
for some Indian people who were thinking, "I'm Sikh in religious
terms but I'm Indian in ethnicity terms" and we found the
count of Sikhs went down. Our judgment is that it is a better
way of getting a good count of Sikhs and Jews to keep these things
within the religion question.
Q65 Kelvin Hopkins: Where I live
the last census threw up a population of 183,000, something like
that, and the local authority with other measures came up with
a figure for the population of the town of well over 200,000,
ten per cent more at least. It is very worrying that there are
so many people not being identified. I know we have a high proportion
of ethnic minority people from abroad and so on, so that is going
to cause a problem, but if that was repeated across the country
we would seriously be getting our population very badly wrong.
Would you like to comment on that?
Ms Matheson: This is at the heart
of what we are trying to do in the population estimates on which
local authorities in part are basing their expectations, and this
is the point Professor Rees was making earlier about expectations.
The population estimates that are produced every year are part
of that expectation and then there is the census estimate which
provides the benchmark. Getting the census estimate right in 2011
is the test that we have applied to all the bits that have come
in to us: can people answer and is it going to help us make sure
that we get the population estimate as good as it can be? Part
of that is also matching the work that is going on elsewhere,
which we are not here to talk about, about improving the population
estimates so that the gap in reality in expectation is not there.
Also, working with local authorities in advance of the census
so that they understand what the census estimate is. If I can
just go on to talk about that very briefly. One of the issues
about the population estimate which I think is not widely understood
is that the population estimate is based on the definition of
the "usual resident", people who are living in an area
or in the country for 12 months or more. That is a UN definition.
Of course, there are people living at any one point in time in
an area with lots of turnover, lots of churn, people coming and
going. That does not mean people are not there, just that they
were not counted in the mid-year estimate. Explaining what that
is and trying, as we have done for the first time this year, to
produce estimates of short-term population, both estimates of
churn, how many people come and go from an area, and what are
the estimates of the number of people who are here for less than
12 months as being part of the work on population estimates, the
census provides an important benchmark.
Mr Watson: Quite a few of the
new questions proposed in 2011 are precisely to try and get this
count of the population right and to help us reconcile differences
in perception between the census results and what local authorities
tell us by reference to other sources. This is partly why we are
asking about second residencies and making sure that we collect
information about visitors overnight. This is why we are asking
about the short-term migration questions that have already been
discussed. It is to allow us to be in a much better position to
reconcile these different views of the world and to explain those
differences.
Q66 Chairman: Just on second residencies,
what precisely is the question going to be?
Mr Watson: The question is, "Do
you spend more than 30 days in another address during the year"
and the second question, "If so, what type of address is
that? Is that a place where you go and work? Is it an Armed Forces'
base? Is it a student's term-time address?" et cetera.
Q67 Chairman: If you go and stay
with granny for 30 days or more a year, that confuses that with
someone who has got a villa in Spain, does it not, which is what
we want to know about?
Mr Watson: It is helpful for many
purposes, for local authorities and other planners to have an
idea of how many people are staying in their patch rather than
usually resident there. They are using services, using street
collection, local transport routes, health services, et cetera.
Q68 Chairman: We do not know at the
moment how many people do have a genuinely second home and we
do not know how many of those are abroad. That is big stuff in
terms of changes happening in society. I do not see that this
30-day question will help. Why not ask the question directly,
"Do you own a second property? Is it in this country or is
it abroad?"
Mr Watson: It is not just necessarily
about owning or renting, it might be about temporarily staying
there. The reason we picked 30 days was because we wanted to separate
out those very short-term visits where somebody is just going
on holiday, somebody is staying with a friend for a couple of
weeks. We want to pick out more substantive stays at other times
of the year.
Q69 Mr Walker: Is it 30 continuous
days?
Ms Matheson: No. I have got the
question here.
Q70 Mr Walker: What happens if you
go and spend two weeks with a friend at Christmas and two and
a half weeks in the summer, does that get caught?
Mr Watson: If it was over 30 days
potentially, yes, if people follow the instructions correctly.
Q71 Chairman: I would urge you to
look at this again. I have discovered it has been frustrating
in wanting to know various things—not to get data—on
how many people own a second property either in this country or
abroad. That has been a huge social change and not to be able
to log it through a census and I do not think you are going to
do it, as Charles' question reveals, through this 30-day question,
why not just ask them?
Mr Watson: Because there has not
been a particularly strong user demand for that information. What
there has been is a strong user demand for us to be able to make
sure that we have a method of counting the population on different
bases. We have the usual resident population base, which is what
the primary census results will be based upon, but using this
visitor information, second residence information, it will be
possible for us to construct estimates of how many people are
there during the working week, how many people are there potentially
at different times of the year. Analysts and policymakers will
have quite a lot more information that they can draw on.
Q72 Mr Walker: Quite a lot of people
go on holiday in August. Why do we need a survey to tell us that?
Mr Watson: We do not, and that
is not what this is trying to do.
Q73 Mr Walker: Can I ask some questions?
There has been an exchange of views between Nicholas Hurd MP and
Sir Michael Scholar regarding the question about the number of
bedrooms. According to Sir Michael Scholar that is "to help
local councils establish whether and where accommodation in their
areas is overcrowded". I used to be a local councillor in
Wandsworth and I am well aware of Broxbourne Council, I represent
Broxbourne, and they have a very able planning department, very
able building control teams who have a pretty sharp idea as to
the amount of accommodation in their area. In fact, most councils
have a pretty sharp idea about this. Can you just persuade me
that there is an absolute overriding demand for this information?
What I have heard from yourselves and other witnesses is this
stock answer, "That information would be very interesting
to some people". You can argue that on any question. Give
us an overriding reason why this must be included in the census
when—I am sorry, this is a long question—the former
Chairman of the Statistics Commission in a letter to Nick Hurd
said: "I was Chairman of the Statistics Commission at the
time of the 2001 census and I saw in operation a ratchet mechanism
by which every ten years the census form becomes more and more
complicated and less fit for purpose". There we have it,
gosh!
Ms Matheson: I will start with
the first part, bedrooms. I said right at the beginning that part
of the test for this is not do people want it, it is do they need
it. Part of what we have been doing is testing with all of those
who said, "Yes, we'd like to have", and asking, "Tell
us how that is going to be used? What decisions are impacted by
that information?" On the bedrooms, I do not know whether
it was Wandsworth and Broxbourne but the response came back from
local authorities about the bedroom standard, which is the measure
of overcrowding. What they may have is an indication of some kind,
and I suspect it may be variable across the country, about the
stock of housing. What they also need is to understand about the
population who are regularly living in different kinds of housing.
It is an aggregate level. You may know that in Wandsworth there
are a certain number of bedrooms and a certain number of people,
but what you do not know is how those things go together in particular
buildings. It is about where are the pockets, and again this was
the test for the census, "Do you need to know it at a small
area level?" The case was very strongly made by CLG and local
government that in order for them to be able to understand and
act on overcrowding, as measured by the bedroom standard, knowing
about bedrooms was an essential part of it.
Q74 Mr Walker: What do you say to
the view of the former chairman of the Statistics Commission[2]
that the census form becomes "more and more complicated and
less fit for purpose"? Indeed, he goes on to say: "The
only people who are not consulted are the ordinary men and women
who are required to complete the form". Then he says that
there is a danger of frivolous answers while other people would
just put the form behind the clock on the mantelpiece and not
be bothered with it at all.
Ms Matheson: I disagree with him
for several reasons. One is the argument about the need, not just
the need but the use. The importance at the heart of this is the
population estimate and the fact that we have to reflect all societal
and demographic changes that are happening in order to be able
to capture the population. Kids do spend time with more than one
parent, people do have more than one address and there is much
more mobility, so being able to capture that is essential. What
is also essential is that we make it easy for people to fill in.
One of the changes in the proposals for 2011 is a redesign of
the form. You were discussing earlier the increase from three
to four pages per person. That is partly because we have simplified
the layout. You will remember from when you completed your 2001
form that on each page you were asked to fill in three columns
of questions. Professor Rees was talking earlier about people
not answering some questions and that was because we have some
evidence that particularly questions in that middle column got
missed out, it was hard work for people to fill it in. What we
have done is taken the opportunity to redesign it using research
evidence from academics and elsewhere, and our own testing procedures,
to say, "Let's make this easy for people to fill in".
The layout has changed, the ordering has changed, and the internet
will help with that as well, of course. That is the reason for
some of the increase in length. It is about getting the population
right and making it easy for people to fill in.
Q75 Mr Walker: Final questions on
the bedroom. If you decided you did not want to fill in the bedroom
question, for example, you felt uncomfortable with it, could you
just leave it blank?
Ms Matheson: You have an obligation
to fill in the census form. The only question that is voluntary
is religion.
Q76 Mr Walker: You feel uncomfortable
about filling in the number of bedrooms, so on the day you fill
in the form you move out the bed, you stick a kitchen table in
there and call it your dining room, then you have only got a three-bedroom
house or a four-bedroom house or a two-bedroom house. You cannot
legislate against that, can you?
Ms Matheson: No, you cannot.
Q77 Mr Walker: It is actually how
people determine the use of their rooms that counts.
Ms Matheson: It is.
Q78 Mr Walker: For anybody who felt
uncomfortable about doing that, move in an air-conditioning unit
and call it a utility room for a day or move in a table and call
it a dining room for a day.
Ms Matheson: Or call it a study
rather than a bedroom. You cannot legislate for that, which is
why there is testing. The other point that was in that letter
that it has not been tested with members of the public is simply
not right, it has been tested with thousands.
Q79 Mr Walker: If politicians wanted
to assuage the concerns of their constituents, what we should
do is say, "Look, if you are concerned about these questions
there are a number of methods you can deploy on the day you answer
them that allows you not to answer them in reality". We could
say that to our constituents, we could help them overcome their
concerns without them feeling if they did not answer these questions
honestly they would get an enormous fine from you or whoever does
the fining.
Ms Matheson: Glen is desperate
to get in here.
Mr Watson: One of the things that
we do in our testing is we ask people whether or not they feel
comfortable or have any objections to answering these questions.
That has been tested for all of these. People are not uncomfortable
with this. The newspapers choose to make something of it, but
generally the population are quite happy to say how many bedrooms
there are in their property.
Q80 Mr Walker: I can see a website,
"How to dodge the census without actually being in jeopardy
of getting a fine" springing up. I can see some civil rights
group getting frightfully excited about this.
Mr Watson: I can imagine all sorts
of websites springing up between now and 2011 around the census.
You might well be right. One of our challenges is to try and get
the right public message out first of all that this information
is valuable, secondly that there is an obligation to fill in the
form and, thirdly, that it is confidential.
Q81 Mr Walker: You might be losing
that battle. The problem is there is huge distrust about government
at the moment, and it will always be there for a variety of reasons
which I will not rehearse again, and unfortunately you are seen
as part of government and yet another part of government taking
a look under people's duvets and inside their bedrooms. You cannot
escape that because perception is reality. If The Telegraph
calls me a crook, I am a crook, there is nothing I can do about
that. If The Telegraphs says you are snooping, you are
snooping and, guess what, there is probably nothing you can do
about that. The battle has probably already been lost.
Ms Matheson: I am nowhere near
as pessimistic as that. The battle is not lost.
Chairman: Just check the results for
Broxbourne when they come in!
Q82 David Heyes: I would like to
go back to income. We have heard from leading academics, from
representatives of the business and commercial sectors and we
know that local authorities, I guess government departments, would
want to see income data collected as part of the census. Certainly
MPs would want it. I think I would like to have better quality
income information as part of my argument that I make for resources
for the most deprived parts of my constituency. There is that
huge weight of opinion that says you should do it. Why have you
resisted it again?
Ms Matheson: I will start and,
again, Glen can provide some of the detail. I recognise the case
that has been made. There is no doubt a lot of people have said
that they would like to have an income question on the census
and a lot of people have said they would like other questions
as well, but income in particular, so we recognise that need.
I mentioned earlier that among the criteria for deciding how much
of a priority this is, and it does come down to priorities, is
are there alternative sources and how good is the information
that could be collected. Again, there was extensive testing on
an income question. There was a difference in response rates.
I think it was, and Glen can correct me, about a three-percentage
point difference in response on the postal test when we had a
split sample, households sent a form with an income question and
households sent a form without an income question. Given what
I said about what the core aim of the census is and what our key
determination is, which is the population estimate, then anything
that is a three percentage point difference immediately raises
some concerns. In addition to that, there has been testing with
interviewing and talking to members of the public about how they
felt following the 2007 test and what do they think about each
of the questions. Of those who objected to a testing, income was
the single biggest complaint that they had. Of those who objected
to any question at all, more than half of them said it was the
income question that was causing them difficulties. There is a
third element which is about the kind of work that we have been
doing since 2001—it was not included in 2001 where there
was also a demand—which is about producing the kind of thing
Professor Rees was talking about, small area estimates modelled
and so on, and looking to see what administrative data either
in DWP or in HMRC could be made available. It was all of those
things combined, plus some concern about the quality of the income
information given that this is a householder who is responsible
for collecting information. Imagine lots of students living together
and having to go round saying, "What is your income? What
is your income" in order to add it up to give a very broad
income band. All of those things combined convinced us that we
did not want to propose the income question.
Mr Watson: I would add a couple
of things. One is that for those people in the test where there
was an income question, so half of the sample, of those people
who did return their questionnaire there was a higher level of
non-response to that particular income question than there was
for the other questions. People might be willing to give us the
form back but with a lot of blanks on the income question. As
part of our post-2007 test evaluation survey asked the question
again to a number of people. We went back to 400 households and
we asked for a second time, "What is your income?" and
something like a third of those people gave a different answer
the second time round. This is a hard question for people to get
right. This was when there was a sources of income question attached
to it as well giving them a reminder to include all the different
types of income, whether from benefits, pensions, salary, share
dividends, investments, whatever it was. The two questions added
together takes up quite a lot of space. There is a lot of non-response.
There is a lot of objection to it. It is not particularly good
quality information and, as Jil said, the key thing was it suggested
a 2.7 percentage point drop in response. Can we afford to take
that risk? This was in a voluntary survey when only half the people
responded anyway. What would that level of non-response have been
for the 50 % less compliant half of the population?
Q83 David Heyes: I follow the logic
of what you are saying. You would be right to be concerned about
the weight of objections that you get, concerns about the quality
and the availability of alternative sources of information, those
are all good arguments and were exactly the arguments that were
overridden when you included a migration question in the census.
Why do those arguments weigh so heavily against including income
and yet they are not such a problem when it comes to looking at
whether or not to include questions on migration?
Mr Watson: We do not have evidence
that suggests the response rate would be affected by—
Q84 David Heyes: The discussion earlier
suggested that there were very similar objections from the same
community of academics, the business community and so on, whose
arguments you rejected in relation to income.
Mr Watson: There are an awful
lot of opinions on this subject, so what we have to do is try
and complement those opinions with some hard evidence. That is
why we do tests and all of this research. In the research that
we have done with these migration questions there is not evidence
that it would materially affect response rates. For example, we
have done a postal test running to tens of thousands of people,
half of the questions included migration questions and half of
the questions did not, and there was no discernible difference
in overall response rates, whereas for income there was a clear
difference. That is why we have taken a different position.
Q85 David Heyes: Jil mentioned earlier
that she meets regularly with her counterparts internationally
and they do not have the same problem with it. Why is this easy
in Australia, Scotland and various other places around the world?
Ms Matheson: It is fascinating,
is it not?
Q86 David Heyes: It is not a problem
for them. Why is that?
Ms Matheson: It is interesting.
The results that we found about the depression impact on response
rate before the 2011 census, ie the tests we have done this time,
mirror the picture we got before 2001. One of the things we do
know is that there is an anecdote in statistical circles about
what are the questions that it is difficult to ask in different
countries, and certainly in the UK even on our household surveys
income is seen as one of the most difficult areas for us to investigate.
There are lower response rates to income questions on surveys
in the UK than in some other countries. I do not know whether
that answers it but it is part of the picture.
Q87 David Heyes: Just one final question
on this. We have heard several times this morning that there are
potentially good alternative sources of income data available
that would serve this purpose and you used the phrase "could
be made available". From the broader concerns of the ONS
I would have thought this would be extremely important information
for you to have. What action is taking place to substitute for
the fact that you are not collecting it through the census?
Ms Matheson: There are detailed
income questions on household surveys. In fact, there are a couple
of surveys that are dedicated to exploring that in detail, which
is what you need to do to get good quality income information.
That is being used to model. What that does not do is provide
detailed local information.
Q88 David Heyes: You cannot drill
down to the individual district.
Ms Matheson: Exactly. What has
happened with that is we have used the information from higher
geographic areas about incomes from surveys alongside information
from other sources to estimate what the average income is likely
to be in different areas across the country. That is one of the
innovations since 2001 because we did not have the income question
in 2001. Alongside that there has been a big increase in the amount
of administrative data that is available for small areas. For
example, on a neighbourhood statistics website there is information
for small areas about benefit recipients and there is some information
from HMRC, but it is not complete because they do not have a complete
picture, at certainly smaller areas than was there in 2001.
Q89 David Heyes: The way you describe
it feels patchy. It certainly is not doing what you were suggesting
might be a good alternative, which is to acquire the total data
in a different way, maybe from HMRC or wherever. Are there any
movements being made towards achieving that volume data?
Ms Matheson: It certainly is an
objective of National Statistics as a whole. What it will not
do is what Keith Dugmore was saying, being able to understand
how that then relates to some of the other questions that are
in the census, relating at the household level, if you like, what
is the relationship between income and the other characteristics
that are collected. It will only do it at the area level, whatever
the smallest geographic area is that the aggregate data can be
provided for while protecting confidentiality.
Q90 Mr Prentice: Why did they go
ahead in Scotland then? You have given us all the reasons why
it is impossible or very difficult to collect data which is meaningful
in England and Wales, why did they go ahead in Scotland?
Ms Matheson: There is also an
issue of prioritisation. We decided that what we really needed
to do was focus on the population estimate and those questions
in England and Wales we need for population, so there is a certain
element of choice. There is also the question about what is the
evidence that you are basing the judgments about quality on. The
evidence in England and Wales is as we have described; in Scotland
their test was different.
Q91 Mr Prentice: Prioritisation is
a red herring really, it is whether the data that is collected
is accurate and meaningful, and you have told us about a 50 %
response rate when you ask people to give details of their income,
and despite all that the people in Scotland are going to press
ahead.
Ms Matheson: I think that was
a different point that Glen was making actually.
Mr Watson: Scotland are pressing
ahead with a household income question rather than an individual
income question. Amongst a large number of people we have tested
reactions and difficulties with answering a question about household
income as well and the feedback we have got, which Jil has already
alluded to, is that it is difficult to know what others in the
household earn, it is difficult to ask that question.
Q92 Mr Prentice: I understand that
you are talking about household income, but you speak to the people
in Scotland and you will have expressed all the reservations you
have about getting accurate data, so why are they going ahead?
Are they saying to you, "No, we disagree with you, Mr Watson.
We think that we will get accurate data at a household level and
that is why we are asking the question"?
Mr Watson: They think they will
get something meaningful and accurate. Our research for England
and Wales, and I think some of our testing has been on larger
samples of the population, suggests otherwise.
Q93 Mr Prentice: How strange.
Mr Watson: Ultimately we have
to both make our own recommendations to our own separate legislatures.
Q94 Mr Prentice: Is there a problem
about Scotland going one way and England and Wales going the other?
We have two censuses asking different questions. That must create
headaches.
Ms Matheson: In fact, we have
got three census offices because Northern Ireland is a separate
census office as well. One of the things that my predecessor and
now me, with the heads of the census offices in Scotland and Northern
Ireland, do is talk to each other very regularly, and we have
a UK Census Committee with a view to harmonising wherever possible,
so the core of the questions are the same, the census day will
be the same and so on. There are users and, indeed, some international
obligations to provide UK level data.
Q95 Mr Prentice: But harmonisation
is going out the window, is it not, because you have got this
income question that everyone is very exercised about.
Ms Matheson: There are some differences
because of different user needs in different parts of the UK.
That is what happens when you have different census offices, they
are responding to their user needs and we are responding to ours,
and also to our circumstances. For example, some of the questions
about why we are so focused on the population estimate, some of
the questions about mobility, numbers of languages spoken and
all of that is a stronger imperative in England and Wales than
in Scotland and Northern Ireland, for example.
Q96 Mr Prentice: I know Kelvin is
desperate to get in. He is straining at the leash!
Mr Watson: I would just add one
point on your question. The feeling in England and Wales is we
really cannot afford to take the risks with response rates. I
am not saying Scotland can, but Scotland did not suffer the same
difficulties with response rates being too low in some areas and
challenges and studies after the census from many local authorities
challenging the results. Their evidence suggests that they have
not got a problem and are prepared to take the risk, and we just
cannot really afford to take the risk.
Q97 Mr Prentice: On this issue of
testing, you tested in Newham, am I right?
Ms Matheson: No.
Q98 Mr Prentice: You tested some
questions in Newham. I read that somewhere.
Ms Matheson: No.
Mr Watson: We are currently doing
a rehearsal in Newham. The 2007 test was in five local authorities,
but not Newham. The testing we have done in various postal tests,
various cognitive tests, has been countrywide over the last four
years.
Q99 Mr Prentice: In those five areas
is there one with a high non-white population?
Mr Watson: Camden.
Q100 Mr Prentice: What is the percentage
of the non-white population in Camden? Just take a guess. There
are only five, you must know.
Mr Watson: I do not know. I would
be guessing a number.
Q101 Mr Prentice: A gentle guess
then.
Mr Watson: I am not sure I want
to.
Q102 Chairman: I think it is a little
unfair even to ask a statistician.
Ms Matheson: I have to say there
is a risk as National Statistician being asked a question like
that and giving the wrong number.
Chairman: Of course there is.
Q103 Mr Prentice: It was just that
we had a big debate earlier about intended length of stay and
whether it was meaningful to put that question in the census.
I am interested in you saying that all of these questions are
tested to destruction. You take a sample and you test them and
look at the response rate. I just wonder what the people in Camden
said about this question on intended length of stay where you
thought it was worth including.
Mr Watson: If my memory serves
me correctly, the question on intended length of stay was not
in the 2007 test.
Q104 Mr Prentice: So it has never
been tested?
Mr Watson: It was a question that
was added later and it was tested thoroughly in postal tests to
10,000 or 20,000 households at the start of last year after the
2007 tests.
Q105 Mr Prentice: In a postal test,
but in a question like this you really want to test it in an area
with a large non-white population. There is no point sending it
to Tunbridge Wells or somewhere like that.
Mr Watson: We picked an area that
did have a relatively high level of immigration in the last decade
or so.
Q106 Mr Prentice: Which area was
that?
Mr Watson: I think we went for
somewhere in the East Midlands. I think we went for Northampton
where there has been quite a lot of immigration, particularly
from Eastern Europe.
Q107 Mr Prentice: And people filled
in that question, there was no problem?
Mr Watson: There was no difference
in response rates in terms of those who returned the questionnaires
with those questions and those without the questions. There were
some people who returned the questionnaires who did not fill in
that question. We have done cognitive testing as well with a number
of people from different backgrounds, recent immigrants, more
settled immigrants, and generally the feedback from the cognitive
testing was that people would be willing to fill in the questions
and give us that information.
Q108 Chairman: When you were talking
about second homes you said that there was no pressure to put
those questions in, so obviously the pressure to have questions
in is a big contributor to whether they get in or not after you
do the testing. Where did the pressure come from to ask the intention
to stay question?
Ms Matheson: It came from ourselves.
It came from our experience in the 2001 census, and the experience
following the 2001 census in discussing with some local authorities
the difference between what they thought their population was
and our estimate of the population from the census, this difference
between the usual long-term resident and people who were there
relatively short-term. It came from the massive amount of work
that has been done since then on looking at the improvements to
migration statistics. It is a combination of getting the census
population estimate right and making it easy by, "If you
are going to be here for three months or more, tell us".
That is a very simple instruction. It also came from the increasing
desire to produce estimates about short-term migration. There
is an intention to stay question included in surveys which are
used in migration estimation. It is not dreamt up out of "this
would be a good idea".
Q109 Kelvin Hopkins: Following from
Gordon's questions—I am fascinated by this—you said
earlier that Britain was very different from other countries in
their willingness to say what their income is, or household income,
but it seems it is more England and Wales because in Scotland
they are quite prepared to let it hang out and boldly say what
their income is, like Gordon would I am sure, but it is the precious
upright English, like me, who have a problem with this. Is there
a cultural or psychological problem which needs to be addressed?
Ms Matheson: I do not know is
the answer. All I know is what the evidence is.
Q110 Mr Prentice: Britain is changing
very rapidly. Is there a case for a five-year census? There are
countries, like Canada, which have these snapshots every five
years and we have heard people say that so much has changed in
ten years the census gets out of date very rapidly. Is there a
case for a five-year census?
Ms Matheson: Certainly it is changing
very rapidly. One of the projects that we have started is to look
at—we have called it "Beyond 2011"—the 2011
census as vital and a benchmark, but then what we have to do is
say what happens after that and how rapidly are things changing,
what are the areas where we need more regular benchmarks. That
may or may not be through a census, it may be through better use
of administrative data, through big surveys, a whole range of
possibilities.
Q111 Mr Prentice: The Government
estimated that 13,000 people from Eastern Europe would come into
the United Kingdom and the reality was 600,000. If change is going
to be that rapid then you have got to get the picture.
Ms Matheson: The census is not
the only source that we have already, of course. The population
estimates and migration estimates are captured through other sources
intercensally. The reason the census is important is that it provides
that benchmark, that point in time, everybody on the same day,
the same questions. It is then supplemented annually, or more
or less frequently, with other sources. The question is, what
do we need to do for the next decade by way of providing the rapid
information that is needed to measure the changes that are happening.
Q112 Mr Prentice: You think about
these things all the time and you must have a view as to whether
it would be a good idea to have a five year census, like Canada,
or whether you should stick with the ten year census and all this
other information that can be collected from other sources.
Ms Matheson: There is nothing
magical about five years at the moment. What I would like to see
is the 2011 census as the core, high quality benchmark that we
can then supplement with administrative information increasingly,
and surveys, so that you have got a much more rapid view of how
things are changing. Whether as part of that you need some kind
of exercise in the intercensal period after five years, six years
or whatever it is, to benchmark is an open question.
Q113 Chairman: Just tell us why we
cannot have a sexual orientation question?
Ms Matheson: For the reasons I
said at the beginning. Partly this is about balancing the user
need and the user case, relevancy of other sources and public
acceptability. Evidence of the testing is that it does depress
response rates. We have to be able to explain to people why it
is there. This is the essence of it: people need to be able to
understand why the question is there, how the information is going
to be used and that it is going to be kept secure. There were
some concerns in a household form about the householder being
responsible for collecting information on sexual identity from
other household members, or those household members then having
to say, "I don't want to tell you, can I have my own separate
form, please". There were concerns about the quality and
concerns from some of the copies of interviews that it would have
an impact on response, in a nutshell.
Q114 Mr Prentice: You have got a
new question on the type of central heating.
Ms Matheson: Yes.
Q115 Mr Prentice: Why is it necessary
to have a question on the type of central heating?
Ms Matheson: Can you remember
what the case was for the question on what type of central heating?
Mr Watson: The question in previous
censuses has been, "Do you have central heating or not?"
It was no longer a good way of distinguishing between different
levels of poverty, deprivation, housing standards, because 99
point something, a very high percentage of places, now have some
form of central heating. There was an opportunity to ask something
that was more relevant and useful in today's society. Given the
interest in fuel poverty, in sustainability and different forms
of heating, a case was made by central government, and I think
by local government as well, to use this space to ask a question
about the type of central heating.
Q116 Mr Prentice: So whether it is
gas or coal-fired?
Mr Watson: Whether it is gas or
coal-fired, exactly.
Mr Prentice: How strange.
Mr Walker: Can you have electric central
heating? I think you can, can you not?
Chairman: You can.
Q117 Mr Walker: If you have electric
central heating, do you have to get in touch with your supplier
and say, "Do you supply me with electricity from nuclear
power, coal power or gas power?", because all those things
produce electricity?
Mr Watson: Let me tell you the
six options: no central heating; gas; electric, including storage
heaters; oil; solid fuel, for example wood, coal; and other central
heating. That has been worked on in consultation with the central
government policy departments who have requested this information.
Q118 Mr Walker: You do not like getting
rid of questions, do you?
Ms Matheson: We have got rid of
some from 2001.
Q119 Mr Walker: Have you? Gosh. It
seems when they become redundant you find a new way of making
them relevant.
Ms Matheson: There are some that
have gone from 2001. For example, there was a question on—
Q120 Mr Walker: "Do you have
a horse?"
Ms Matheson: Lowest floor level
of accommodation. People who are living in blocks, "What
is the lowest floor level of accommodation?" That has gone
completely. There are some other examples.
Q121 Mr Walker: Are you going to
start asking people what type of car they have, electric or diesel?
Ms Matheson: No plans to do so.
Mr Walker: But we could make a good case
for its inclusion!
Mr Prentice: What percentage of households
in England and Wales do not have central heating? In my constituency
quite a lot of people do not have central heating. What is the
figure in England and Wales?
Q122 Chairman: That is one of those
unfair questions again.
Mr Watson: I would prefer to let
the Committee know afterwards.
Q123 Mr Prentice: It is not unfair
because this is a new question and you would know we were going
to ask you about the new questions in the census. You just explained
you wanted to know about the type of central heating because so
many households now have central heating. My question, I think,
is a fair one, which is how many households in England and Wales
do not have central heating?
Mr Watson: I make no comment on
whether it is a fair question or not, but either way the answer
is the same, I am afraid I do not know and would prefer to write
to the Committee.
Q124 Chairman: That is why I say
it is an unfair question, because you are expected to give a statistical
answer and you do not know necessarily what it is and if you gave
it unreliably you would be mocked. I think you are well protected
there. So we do not end on central heating—nothing wrong
with central heating, we have gas—can I ask you about national
identity. This is another one that is coming in fresh. I do not
know the detail of the question but my slight worry here would
be that if people have got to choose a national identity we know
from other sources that many people see themselves as having multiple
identities, and if people are asked to describe themselves as
Welsh, Scottish, British, many people would want to say, "I'm
Welsh and I'm British" or "I'm Scottish and I'm British".
Is that possible?
Ms Matheson: You can do that.
We have got the advantage of having the dress rehearsal form in
front of us. The question says, "Tick all that apply".
Q125 Chairman: So you can do the
permutations, okay. How catastrophic would it be if politicians
started playing about with this in the next period and started
suggesting that questions should go and some should come?
Ms Matheson: We have got a timetable.
Census day is 27 March 2011. We need the regulations in place
in order that we can print the questionnaire, set up the systems,
finalise the arrangements and have the authority to recruit and
train the 30,000-odd people that we will need in order to carry
the census out. Any changes after spring of next year are going
to introduce both cost and risk and, therefore, are likely to
be very serious. Up until that time it is for Parliament to decide.
Q126 Chairman: If Parliament, for
whatever reason, said, "We would like an income question"
or "We would like a sexual orientation question", you
would just groan, grit your teeth and get on with it?
Ms Matheson: I think there is
a procedural point, which Glen was going to remind me about, which
is that this is a strange Order in that it is partly subject to
affirmative resolution and partly to negative resolution, so the
only way of adding would be to reject the Order and then to start
again. The bit that is prescribed by the 1920 Census Act is the
negative resolution and the bit that is new or additional, or
not covered there, is affirmative[3].
Q127 Chairman: My understanding is
it would be possible for a government, and it can only be a government
that can do it, to put down an amendment.
Ms Matheson: Exactly.
Q128 Chairman: If Government was
prevailed upon that there ought to be an income question or a
sexual orientation question they could put down an amendment and
then you would have to put up with it, would you not?
Ms Matheson: Yes.
Mr Watson: Can I add one point
of clarification. Some people have made the case for a sexual
identity question, for example, to be voluntary in the way that
the religion question is. If that were to be the decision of Parliament
then that would require an amendment to the primary legislation,
the 1920 Census Act would have to be amended.
Chairman: That sounds like a shot across
the bows to me. Helpful information, as we said. That has been
interesting. Thank you for coming along and talking through this
again. Thank you very much indeed.
2 Sir John Kingman Back
3
Ev 27 Back
| 
