That sure raises a lot of issues, Humphrey. I don't envy you a census with
such a poor selection of variables in either version, much less a decennial
one!
Anyway, there are a number of questions that look like they could easily be
dispensed with outright, e.g. national identity. Others can be dropped if
another question is asked differently. For example, if you ask a question
like 'When did you first come to live permanently in the UK?', you get the
date of arrival and don't need to bother about intention to stay.
If the language question requires respondents to assess their ability in
each language, it is probably the wrong question. Assuming that the reason
for asking it is to ascertain needs for translations of brochures on
government services, or interpreters in hospitals, etc., there is certainly
no point in anyone determining how many people believe they remember the
high school French they've never found a use for. What we collect is 'Main
language other than English spoken at home'. It is probably the most
useful of the documented language variables: 'Languages spoken at home',
'First language spoken', and 'Main language spoken at home', for the usual
purposes.
I'm surprised that the income question only resulted in 2.9% refusals,
although you don't say how many returned the form with the income data
missing? Some testing has been done here and found that how you ask the
question makes a lot of difference to response rates, as well as accuracy,
etc. I think it's pretty important to have clear instructions to the
respondent reminding them that government benefits, interest, etc. count as
income. They also ought to know that one off windfalls - inheritances,
gifts, gambling winnings (except for professional gamblers) - are not
income (assuming the ONS uses standard concepts). If salary packaging or
the like exists in your hemisphere, it needs consistent treatment and
appropriate instructions. If in kind receipts count as income, it needs to
be clear how cash values are to be imputed.......
Income is not a great indicator of 'economic wellbeing' for a number of
reasons. What we call 'Unincorporated business income', for example, is
collected net of costs, while wage income is not... Even if income from
all sources were truly compatible, standard of living is affected by access
to savings to deplete or credit, etc. I assume income is to be collected
on a person basis if at all. Personal income is not useful as an indicator
of personal economic wellbeing, apart from lone person households. Nor is
household income, given households of different sizes. Nor per capita
household income, because different household members absorb different
amounts of resources. Equivalised household income is a little better, but
the equivalising factors can be arbitrary.
That said, with a social analyst hat on, I'd be crushed not to have some
kind of income data from the census. Sure, the census is a blunt
instrument and the income data are not as good as you'd get from an income
survey where respondents are asked to check their records in advance and
trained interviewers can probe fairly deeply. But as I was saying earlier,
only census data allow disaggregation to small geographical areas and cross
tabulation of multiple variables without compromising accuracy through high
levels of sampling error.
In solidarity,
Harry
Humphrey Southall
Humphrey.Southall@POR
T.AC.UK To
Sent by: email list [log in to unmask]
for Radical cc
Statistics
<[log in to unmask] Subject
.UK> Income questions and other questions in the 2011 Census
Protective Mark
07/11/2007 11:45 AM
Please respond to
Humphrey Southall
<Humphrey.Southall@PO
RT.AC.UK>
Having spent three hours today at an ONS advisory committee meeting
on the 2011 census, coming back to much of the discussion on the
Radstats list was a bit surreal. I would rather report back on the
meeting than discuss some of the postings.
The meeting was the first since the big census test in May, and
probably the last before the questions are finalised ... or so one of
the papers for the meeting said. This was later clarified with a
statement that ONS have already decided what questions they think
should be on the form, but (a) ONS does not take the final decision,
and (b) there is a decision still to be taken on the size of the
form, which has large cost implications. As usual, part of the form
will be questions about the whole household, mainly about the
dwelling, but the larger part will be a series of questions to be
answered about each individual in the household (for up to six
household members in the standard form).
The issue is whether each household member gets three pages or four,
and these are the topics which will be asked about under each option:
3
pages 4 pages
Name
Yes Yes
Sex
Yes Yes
Date of birth
Yes Yes
Marital/civil partnership (new) status Yes
Yes
Student status Yes
Yes
Term-time address Yes
Yes
Second residence (new) Yes
Yes
Country of birth Yes
Yes
Address one year ago Yes
Yes
Month/year of entry into UK (new) Yes
Yes
Intention to stay in UK (new) No
Yes
Citizenship (new) Yes
Yes
Ethnicity Yes
Yes
National identity (new)
Yes Yes
Religion Yes
Yes
Welsh language proficiency (Wales only) Yes
Yes
Language (new) No
Yes
Health status
Yes Yes
Long-term illness/disability Yes
Yes
Carers No
Yes
Qualifications No
Yes
Economic activity status Yes
Yes
NS-SEC (self-employed, occupation, Yes
Yes
supervisor status, ever worked) Yes
Yes
Industry/name of employer No
Yes
Workplace address Yes
Yes
Transport to work Yes
Yes
Using the longer form is not seen as affecting response rates, so the
issue is cost: four pages per person would cost an extra Pnds 22m.,
which ONS does not have but which they are asking other government
departments to contribute to. The best guidance that they could give
us was that it is genuinely uncertain whether the money will be found.
I do have more detailed information on the actual proposed questions,
but too long-winded to repeat here. There is nothing "secret" about
the documents that advisory group members get sent, but JISCmail does
not want us attaching long documents to list postings, and
re-publishing the documents on another web site would breach ONS
copyright, etc. They say they will try to get them on their own
website before advisory meetings rather than after, to help us sound
opinions more widely.
The two issues I want to raise are:
(1) Neither option includes an income question, even though an income
question was included in some versions of the May census test
questionnaires, and we had been sent a separate paper discussing the
pros and cons which said no final decisions had been taken. This is
when the bit about "ONS's decision" versus "the final decision" came
out. In the past, much of the opposition to an income question came
from within ONS itself, and was based on concern about the impact on
response rates. The test measured this, and showed that households
given an income question were 2.9% less likely to return it than
those gettings forms without the question -- and the impact on
response rates was fairly consistent across different categories of
areas defined in terms of "difficulty of enumeration". Of course,
replying to the census test is entirely voluntary while the census
is, in principle, compulsory; but in practice ONS see this as very
hard to enforce.
That information about response rates was interesting, and shows
there is a real "price" to be paid for an income question -- but the
main reason we were given for not including an income question in
2011 was that the central government departments and local government
bodies consulted saw it as a lower priority than the topics that were
being included, even in the four page version. Today's meeting was a
joint meeting of the academic and business advisory groups, and
interestingly both constituencies were very unhappy about this. IF
ANYONE WANTS AN INCOME QUESTION IN 2011, YOU NEED TO LOBBY HARD NOW.
(2) Two questions which have been asked for much of the last century
get dropped in the three page version, on educational qualifications
and on industry/business of employer. These are both very important
in getting an understanding of disadvantage -- it was pointed out
that knowing an "occupation", such as "engineer", is pretty useless
unless you also know what kind of employer they have -- car repairer
or engineering consultancy. Two other questions asked in 2001,
although with less of an earlier history, are also dropped even from
the four page version: number of employees, and hours worked.
Part of the reason there is a lack of space is the series of
questions on "identity" and immigration. One worry is the obvious
political drivers behind these questions, but asking about
citizenship may have a larger impact on response rates in some parts
of the population than an income question (the actual questions is
about what passports are people entitled to hold, and is only asked
of people born outside the UK; the question about "national
identity" is really about how people feel; and the question about
"intention to stay in the UK" is obviously about intent. Past
censuses have stuck to pretty basic factual issues, and there is
already evidence that many people, and not just special cases, find
it hard to answer these questions.
DO WE START A CAMPAIGN FOR "FOUR PAGES, NOT THREE"?
One downside of the longer questionnaire is that we will ALL have to
decide whether the French we learnt at school made us "fluent". The
introduction of civil partnerships has made the marital status
question much more complicated; one of the options now is
"Separated, but still legally in a civil partnership". Lastly, I
have asked if there is any statement available on ensuring
contractors follow the rules on confidentiality.
Best wishes,
Humphrey Southall
******************************************************
Please note that if you press the 'Reply' button your
message will go only to the sender of this message.
If you want to reply to the whole list, use your mailer's
'Reply-to-All' button to send your message automatically
to [log in to unmask]
Disclaimer: The messages sent to this list are the views of the sender and
cannot be assumed to be representative of the range of views held by
subscribers to the Radical Statistics Group. To find out more about Radical
Statistics and its aims and activities and read current and past issues of
our newsletter you are invited to visit our web site www.radstats.org.uk.
*******************************************************
------------------------------------------------------------------------------------------------
Free publications and statistics available on www.abs.gov.au
******************************************************
Please note that if you press the 'Reply' button your
message will go only to the sender of this message.
If you want to reply to the whole list, use your mailer's
'Reply-to-All' button to send your message automatically
to [log in to unmask]
Disclaimer: The messages sent to this list are the views of the sender and cannot be assumed to be representative of the range of views held by subscribers to the Radical Statistics Group. To find out more about Radical Statistics and its aims and activities and read current and past issues of our newsletter you are invited to visit our web site www.radstats.org.uk.
*******************************************************
|