Not sure that I am glad to get confirmation that census results that report
0 or 3 actually
mean 0, 1, 2,or 3.
Why publish all these finctional 0s and 3s? Would it not be better to
publish all these results as '3 or less than 3'?
Idon't know haw to make a 'less than or equal ' sign on an email system, but
'3 or less than 3' it can be expressed in two symbols that are fairly
widely understood even if not easy to put into an email message.
Ray Thomas
********************
----- Original Message -----
From: Ludi Simpson
To: [log in to unmask]
Sent: Monday, January 12, 2004 8:03 AM
Subject: Re: census quality
Since punchlines tend to get reproduced, please note that in the census
output ALL 1s and 2s are randomly edited to either 0 or 3. No value of 1 or
2 appears in the census output. So a published zero may be a 0, 1, or 2, and
a published 3 may be a 1, 2, or 3. The shock expressed by recent messages
(where the impact was expressed slightly differently and wrongly) was shared
by all the census users and representatives of users who complained when the
decision was sprung on them in 2001.
Whether it is important in practice depends on the context. The paper
referred to gives advice as to how to measure the impact, which is likely to
be greatest in studies of small areas, such as characteristics of
regeneration zones, or health around hazard sources like chemical factories.
Ludi
---
"The adjustment procedure has not been adopted by the Census
agency in Scotland, where the Registrar General feels that the
risk of disclosing personally identifiable information is not
sufficient to warrant the cost and impact of adjusting the census
data in this way."
******************************************************
******************************************************
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]
*******************************************************
|