My understanding is that a census is not subject to sampling error for
obvious reasons. Since there is no sampling error, I believe that non
response has less impact, not just because each non response represents a
smaller proportion of observations, but also because self selection in a
sample survey distorts the randomness of the sample, which is not an issue
with the census.
Furthermore, as sampling error is not a factor (although confidentiality
may be), it is possible to release data on small geographical areas. Such
data are alleged to be of use in determining constituency boundaries,
planning infrastructure development and services, etc. I presume there
would be an outcry from the marketting and psephological community if they
were deprived of the kind of data that allows them to do their analysis and
projections. Around here, Census data are also used as a benchmark used to
recalculate the weights applied in sample surveys.
John Whittington
[log in to unmask]
.UK To
Sent by: email list [log in to unmask]
for Radical cc
Statistics
<[log in to unmask] Subject
.UK> Re: More about Lockheed Martin and the census [SEC=UNCLASSIFIED]
Protective Mark
02/11/2007 08:06 PM
Please respond to
John Whittington
<[log in to unmask]
O.UK>
At 07:26 02/11/2007 +0000, R.Thomas wrote (in small part):
It appears to be increasingly difficult to maintain the dual function
of the Census - as a basis for estimating the total population and as
a source of detailed information on the population. The Government
has already been obliged to use administrative sources for estimates
of the geographical distribution of the population. Surely that
trend will continue?
Surely the Census will increasingly be seen as a large scale social
survey? Wasn't the 1966 Census conducted on a 10% sample basis?
Might not such a solution be on the cards for 2011?
I must say that, perhaps because I'm naieve, I've never really understood
why the statistical community 'supports' (implictly and/or explicitly)
attempts at a 100% Census. If, in virtually any other context, it was
suggested to a statistician that one should conduct a !00% census in order
to estimate a count and/or characterise (detailed information about) a
population (whether of people, trees, fish or whatever) of 60+ million, I
suspect that (s)he would say that the proposal was crazy, since high
precision estimates could be obtained from a sample survey of a tiny
fraction of that population.
Is it not the case that the original concept of a Census (in Biblical times
before) arose because the need was not so much to count and/or
characterise, but actually to _identify_ all the individuals in the
population in order that they could 'be sent tax bills' - which is in some
senses the very sort of thing that many of the "don't want to be counted"
people fear about modern-day Censuses. In statistical terms, I would have
thought that for any purposes _other_ than that, a relatively small sample
survey (certainly no more than 10%, quite possibly less) would be quite
adequate.
However, whether one conducts a census or a survey, those fears about
'usage of the data' are still going to cause some people (presumably
roughly the same proption in samples and in the whole population) to
attempt to 'hide'. The statistical solution to that would presumably be to
offer total confidentiality/immunity to people in relation to anything that
wrote on the form (the easy part to implement) and to convince them that no
use would ever be made of that personal information (next-to-impossible!).
In other words, people would have to be convinced that they could answer
'yes' to questions like "Are you an illegal immigrant", "Are you a
terrorist", "Have you ever abused any children", "Are you evading any
taxes" without any fear of 'having their collar felt', or even of being
brought to the attention of any 'relevant authorities'! That approach is
therefore probably little more than a pipe dream. Ironically, in view of
how this discussion started, it might well be easier to reassure people if
the census/survey was outsourced to a commercial organisation, who
'promised' not to pass on any individual-identifiable data to government.
Kindest Regards,
John
----------------------------------------------------------------
Dr John Whittington, Voice: +44 (0) 1296 730225
Mediscience Services Fax: +44 (0) 1296 738893
Twyford Manor, Twyford, E-mail: [log in to unmask]
Buckingham MK18 4EL, UK
----------------------------------------------------------------
****************************************************** 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.
*******************************************************
|