The appropriate sample size for a survey would depend on the geographical scale at which you are looking for analysis.
Different government departments have different takes on this issue. The question is whether there are neighbourhood effects and the scale at which they might occur. For example, if there is a neighbourhood effect based on the view that 'no-one works round here', is the 'here' a region, a NUTS 2 area, a local authority, a MSOA, a ward, a LSOA, or an Output Area. I suspect that most people who look into that question would suggest that neighbourhood effects might be identified at LSOA or output area but are rather less likely to be identified at higher scales.
If you think there are significant neighbourhood effects at very local scales then a sample survey would need to be so large that it is almost a census.
On the other hand, if you do not believe in neighbourhood effects, then a much smaller sample might be sufficient for your needs. Whether a researcher thinks neighbourhood effects might be significant depends in part on disciplinary background....geographers have a slight bias in favour, and economists may regard unobserved heterogeneity as a problem.
---------------------------------------------------------
Paul Bivand
Head of Analysis and Statistics
Direct Line: 020 7840 8335
Inclusion
3rd floor, 89 Albert Embankment, London SE1 7TP
Tel: 020 7582 7221
Fax: 020 7582 6391
Inclusion website: www.cesi.org.uk
The contents of this e-mail and any attachment(s) are intended solely for the use of the addressee. If you are not the intended recipient please return the e-mail to the sender and delete from your mailbox.
The contents of this e-mail and any attachment(s) are intended solely for the use of the addressee.
-----Original Message-----
From: email list for Radical Statistics [mailto:[log in to unmask]] On Behalf Of John Whittington
Sent: 05 November 2007 12:05
To: [log in to unmask]
Subject: Re: More about Lockheed Martin and the census [SEC=UNCLASSIFIED]
At 11:08 05/11/2007 +0000, John Bibby wrote (in part):
One advantage of a full count is that it s easy to understand. Also, I guess the sampling fraction would have to be pretty low (<40%) before you see any cost-gains. The problem is not in the aggregates, but in the small cross-classifications.
I realise that, but the total population is so large that I suspect that a sampling fraction considerably under 40% (as I said, I was have guessed 10% or less) would provide estimates of more than adequate precision, even for the 'small cross-classifications'.
In deciding what degree of precision one should aim for, it is worth bearing in mind that we are taking a 'once-per-decade snapshot' of a relatively rapidly evolving situation. To my mind, that suggests that the cost of demanding excessively precise estimates of those snapshots would be very difficult to justify.
Has anyone seen cost-=benefit analysis applied to this area i.e. What is the cost of this information? What are the benefits ? (and what are the marginal benefits for marginal changes in size or reliability?)
I certainly don't know. I presume that the cost side of the equation has been looked at in some depth. However, even once one has decided on what regards as 'the benefits', I would presume that many of them are things which are very difficult to 'value' quantitatively.
Kind 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. *******************************************************
******************************************************
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.
*******************************************************
|