[CLUE-Talk] Bowling for Columbine

[Timothy C. Klein]

* David Anselmi (anselmi at americanisp.net) wrote:
> Timothy C. Klein wrote:
> [...]

> >I hate this saying.  As a budding mathematician, I must say:  statistics
> >*never* lie.  People do, and sometimes people intentionally use
> >misleading statistics.
> 
> The saying isn't that statistics lie, it's that if someone is telling 
> you statistics you'd be better off if they were (merely) telling you 
> damn lies.  It's easier to detect a lie (even for mathematicians, I'd 
> guess) than a misleading use of statistics.
> 

Yeah, I realize how it is usually used:  if someone gives you a
statistic, don't trust them, more or less.  The problem with statistics
is this, as I see it:  people follow them blindly, or consider them all
lies.

The actual fact is that is isn't really that hard to determine if a
given statistic is valid:  one needs only to ask a few simple questions.
If those questions are answered, you can determine if the given stat is
at least reasonable.  If those questions are answered and shed bad light
on the data collection/analysis, *or*, and here is the catch, those
questions aren't answered (or worse, even asked), then the statistic
should not be bandied about by anyone.   The media and politicians are
the worst offenders in this camp.

But, alas, I am an idealist.  

Tim

PS -> The kind of question you need are:  Who gathered the data?  What
was the sample size?  What measures were taken to make sure the sample
was as random as possible?

Thats it if you don't know/care about the math.  The last category is
generally the one that invalidates stats
--
==============================================
== Timothy Klein || teece at silverklein.net   ==
== ---------------------------------------- ==
== "Hello, World" 17 Errors, 31 Warnings... ==
==============================================