Phil Lee <[EMAIL PROTECTED]> writes:
> In his discussion Lambert says "most of the increase was caused by a change in
> the way crimes were counted." Homicides should not be affected by counting
> rule changes.
Homicides make up a very small fraction of violent crimes. Changes in
the number of homicides will have virtually no effect on the total
violent crime rate.
> A common technique of advocacy "science" is
> to pick two years for results and say "the change between the two years proves
> my point." Lambert's choice is 97 and 2001/02 for violent crimes.
Not my choice. I was responding to Malcolmm's claim that
"And in the four years from 1997 to 2001, the rate of violent crime
[in England] more than doubled."
> Of most interest is Lambert's assertion that surveys are more accurate
> ("surveys like BCS give a much more accurate estimate of the total number of
> crimes than police reports"). I would like to see that assertion documented,
> but will point to the "2001 British Crime Survey" at:
> http://www.homeoffice.gov.uk/rds/pdfs/hosb1801.pdf (see page 3, Acrobat page
> 17) where a comparison between the BCS and Police Recorded Crimes is made.
> Nowhere do I see a claim for greater accuracy for the BCS. I do see one
> difference is the BCS does not measure "Crimes where a victim is no longer
> available for interview." Obviously, that includes the violent crime of
> murder which I thinks is a serious accuracy defect.
Page 6:
"The BCS measures both reported and unreported crime. As such the BCS
provides a measure of trends in crime that is not affected by changes
in public reporting to the police or police recording."
Generally random sampling provides much more accurate results than
trying to count the entire population. Police figures in England may
be better, but Maltz's examination of the quality of UCR crime figures
in the US led him to discount Lott's findings on these grounds alone.
See
http://timlambert.org/guns/files/maltz.html
for a summary
--
Tim
