Felix Ungman wrote,
> Sent: Monday, August 13, 2001 9:17 am
> To: extropians@extropy.org
> Subject: Re: Definition of Racism (without rent-a-riot)
>
>
> On fredag 10 augusti 2001 21.40, Harvey Newstrom wrote:
> >Mike Lorrey wrote,
> >> Wrong. If the statistics say that, say, 10% of blacks commit
> crime while
> >> only 1% of whites do, then it is a reasonable assumption to
> say that any
> >> individual black person is ten times more likely to commit crime than
> >> any white person, all other factors being equal.
> >
> >This is just plain wrong. Any freshman course on statistics
> will teach that
> >this is wrong. I am white. Do I have 1% chance of committing a crime?
> >Charles Manson was white. So was Hitler. Do they all have the
> same chance
> >of committing a crime? We cannot deduce anything about any
> individual based
> >on their race. All whites added together might add up to a 1%
> crime rate,
> >but you have no information about the percent chance of any individual
> >committing a crime.
>
> Mike is using subjective probabilites. That is, given a
> completely random black man, without further information, Mike
> *belives* that that man has 10% probaility of commiting crime.
> This would be subjectively correct. But extremely lazy. Further
> information might reveal that the man's wearing a suit and tie
> and titles himself as an accountant. Computing the fraction of
> criminal accountants compared to the total number of accountants
> might give the probability that this particular man being
> criminal is 0.2% (In Mikes opinion of course, the man himself,
> having a sawn-off shotgun in his suitcase, about to rob the
> nearest bank knows otherwise).
No offense, but it is clear that you and Mike have never taken a college
level statistics course, especially one specializing in biological or
psychological statistics. I have, and the real world gets a lot more
complicated than these arm-chair statistics would imply.
You are simply taking an average here. You have not calculated standard
deviation, variation, distribution, precision, accuracy or error rate. You
have not normalized the data, isolated a single variable, or validated the
sample. The kind of statistics you are applying only work for
random-distribution, normal bell-curve, single-variable, regularly occurring
events like cards or dice.
If you made a card game out of this, your calculations would help you count
cards and predict the game. But if you tried to predict social behavior,
biological distribution, or psychological trends using these methods, you
would fail.
-- Harvey Newstrom <http://HarveyNewstrom.com> <http://Newstaff.com>
This archive was generated by hypermail 2b30 : Fri Oct 12 2001 - 14:40:08 MDT