I am stuck again.  I should probably have waited to reach this point
before sending my previous response... Here's where I am at:

On 6/27/07, John Randall <[EMAIL PROTECTED]> wrote:
Then $E(S^2)=E((1/n-1)\sum (X_i-\bar X)^2
=(1/n-1)( \sum E((X_i-\mu)^2)-nE((\bar X-\mu)^2 )
=(1/n-1)( \sum \sigma^2(X_i)- n\sigma^2(\bar X) )$

But this seems trivially wrong.  For the instance where our population is
number of heads from a coin toss, and a sample size of 2, the expected
value calculations are all 0.25.  However, that last expression
 $(1/n-1)( \sum \sigma^2(X_i)- n\sigma^2(\bar X) )$
would have four possibilities in this instance:

X_0  X_1  \bar X
  0    0     0
  0    1     0.5
  1    0     0.5
  1    1     1

If I interpret the \sum in that expression as applying to everything to its
right, my corresponding answers are never greater than 0, because
X_i can never be greater than n*\bar X.  Alternatively, if I interpret the
\sum as applying only to \sigma^2(X_i) then my corresponding answers
are always 0, because \sum X_i and n*\bar X must be equal.

Am I missing something?

Thanks,

--
Raul
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to