[geeks] What desk toy or "tchotchke" says "geek" to you?

[Mike Meredith]

On Thu, 30 Mar 2006 12:35:44 -0600, Brian Dunbar wrote:
> One problem is that you're talking about soft numbers that are easily
> fudged.  Not cooking the books by the bean counters but ... people
> lie.

You don't say! Perhaps the people who come up with the crime statistics
may also have thought of this problem ?

The serious UK crime statistics are based on the collation of recorded
crimes. I'm not familiar with the behind the scenes processes involved
but I do know that to become a recorded crime you have to 'report' it
twice ... the initial call, followed by a signed statement. 

As far as I am aware the police have to be satisfied that a crime did
take place before it can be recorded. There is some concern in the UK
about the accuracy of recorded crime, but it is a belief that crime
statistics are under reported to some extent ... probably varying
according to the type of crime.

>  People lie to pollsters and they lie to the Man and they'll lie even

Most people are pretty bad at lying (and the police not totally
hopeless at figuring out when they're lying), and are aware that the
police can easily change the recorded crime from what was claimed to
'wasting police time'. I don't think false reports are a huge problem,
and some crimes are considerably less prone to them (after all the
murder victim isn't going to lie about it).

> Social issue numbers are harder to verify and harder to tally and, in
> general, squishy.

Indeed which is why the people designing the process to collate them are
very careful about how the process works. They're also very pedantic
about what exactly is being measured ... 'recorded crimes' != 'total
crimes'. And any decent statistician should be able to estimate a
'squishyness' value.

Recorded crime stats are probably relatively accurate.