🚀 go-pugleaf

In article <C1BpHJ.21yK@austin.ibm.com> me@vikings.austin.ibm.com (Michael E) writes:
>Consider this: a public school bus picks up tens of kids. For every kid 
>that opts for a private school with no bus, one additional vehicle uses up 
>equipment, fuel, roads, and time (after all the school bus will run for the 
>remaining children). Quite a number of enviromental and parental 
>finance problems the voucher system creates...

Its obious that you've not thought of any of this.

For example, with all of those children clamouring for a private
education, private schools will grow.  Just like computer companies
grow when people decide they want new computeres.

With all of these children attending private schools, mass transit
will become more popular.  It might even take the form of little
yellow busses with more than one final destination.

And last but not least, now that children are allowed to attend
schools based on the desires of the parents, and not those of
politicians, children might be allowed to attend LOCAL schools
instead of being bussed.  And yes, MOST parents want their kids
in the neighborhood school system, not the one ten miles and
several thousand gallons of gasoline away.

>Unanswered!

Unanswered!  Have you stopped beating your wife?

Let me explain the concepts of mean and standard deviation.

Mean describes a number which represents the sum of all samples
divided by the number of total samples.  But see, there is a
small problem when a sample has limited upper or lower bound.
The mean winds up with the same boundaries.  That is, if the
range of acceptance rates is 0 to 100%, the range of the mean
will also be, 0 to 100%.  Thus there is an interestin problem.
If the mean is 98%, there =must= be a very large number of
samples at 100% just to make up for one sample of 50%.

Now, you might argue, surely there can be =many= schools which
only accept 50% of their applicants.  Sure.  But for each and
=every= 50% school, there must be 48 schools which accept 99%
of their applicants.  Of course, you can make it less absurd
by saying that the other schools take 100% of their applicants,
but you then need 24 of these super-accepting schools.

OK, I've now shown that it is simply absurd to suggest that
any significant number of marginally accepting schools exist
in an environment where 98% of the applicants are accepted,
I'll now show that the standard deviation must be rather low.
I'm going to hand-wave this one, so watch real closely.

The standard deviation is



               /---------------
              / n
             / ---       _ 2
        s = /  >   (xi - x)
           /   ---
          /   i = 1
      \  /  ---------------
       \/      ( n - 1 )

(Whew)
                                                         _
By the earlier paragraph, I've shown that the term (xi - x)
is more often than not, a small number.  Squaring small numbers
leaves numbers that aren't much bigger - that is, if you square
n / 2, you wind up with a number much smaller than n**2.  So,
the sum of those terms is about n * (some small number).

But, you suddenly interject, there are going to be a few terms
where (xi - mean(x)) is big, I mean, really big.  Once again
I point you back to that earlier paragraph.  To have ONE school
with an acceptance rate of 10%, well, every other single school
in that entire city would be rather gracious in taking all of
the other students.  So, shut up and sit back down.

OK.  You convince me that there is this evil school which only
takes 10% of the students who apply.  I do the quick math, find
there the other 44 schools (this town only has 45 schools ...)
all accept 100% of their applicants.  Being the nice guy I am,
I then compute the standard deviation.

	s = sqrt ((44 * 2**2 + 88**2) / (45 - 1))

Doing the math, the standard deviation is 13.4.  Ooooo.  That
sounds REALLY BAD.  Think of all those schools with less
than 98% acceptance!  Gee, a standard deviation of 13.4% means
that A LOT of schools have as low as 85% acceptance.  NOT!
There is exactly one school below 98%, the rest are exactly
100%.  That's just the way the fucking math works out.  Dream
up all the evil, racist, upper-classist, elitist schools you
can dream up.  The average, for that one study, must STILL be
98%.

See, in asking your question you demonstrate that you don't
know statistics, or even basic algebra.  What you want is the
skewness.  Actually, what you want in this sample is the
median.  And the median is ... 100% acceptance.  And this is
for your evil 10% acceptance rate "School of Upper Class White
Supremecists".

>In a class on managing a political campaign braodcasted on C-SPAN, a speaker
>noted that one of their polls gave a woman candidate 3% name recognition
>with a 5% margin of error. Just a real life example. 

Big deal.  In my example above, I gave the standard deviation as 13%
with a mean of 98%.  YOU DON'T KNOW SHIT ABOUT STATISTICS.
Standard deviation doesn't mean what you suspect it does.

>I AM trying to get a clue: Give a source, give data!
>Better yet: multiple, verifiable sources.

I don't have to give sources.  You can call any private school
you want.  If they have seats, you have money, and you child
is academically qualified, you get in.  I know Jews who
graduated from Catholic high schools.  It doesn't get much more
obvious than that.

>"There are three kinds of lies: lies, damned lies, and statistics" 
>- Disraeli

Proof by repeated assertion doesn't work in this group.
-- 
John F. Haugh II                  [ PGP 2.1 ] !'s: ...!cs.utexas.edu!rpp386!jfh
Ma Bell: (512) 251-2151           [ DoF #17 ]        @'s: jfh@rpp386.cactus.org

The DDT ban has killed more Africans than the AIDS epidemic.  Legalize DDT now.