Unbiased Estimator


By Hal 2001 (P3046) on Monday, December 18, 2000 - 05:29 pm :

What is the formula for the Unbiased estimator (for mean and variance)? And how and why does it work?

Hal


By anon (P2831) on Monday, December 18, 2000 - 07:35 pm :

unbiased mean = sample mean
unbiased variance = n/(n-1) x sample variance

be back for an explanation

dimitri


By Dave Sheridan (Dms22) on Wednesday, December 20, 2000 - 01:02 am :

Y is an unbiased estimator for a quantity c if we have
E(Y)=c
that is, the expected value of Y is c. So you can take any quantity whose expected value is c, and this is an unbiased estimator. The ones Dimitri quotes are the standard ones - it's easy to show they have the required property. In specific situations, there are other unbiased estimators and in this case you have to use other methods to decide which is the "best" one.

For example, sample mean is
sum(Xi )/n
and the expectation of a sum is the sum of the expectations so expected value of sample mean is sum(mean)/n which is of course simply the mean. The variance is slightly more tricky, but follows a similar pattern.

-Dave


By Anonymous on Wednesday, December 20, 2000 - 01:48 am :
Is s2 a consistent estimator of s2?
By Dave Sheridan (Dms22) on Wednesday, December 20, 2000 - 10:25 am :

By S2 do you mean the variance? We normally use sigma instead of S. The answer is yes, because n/(n-1) tends to 1 as n tends to infinity.

But do you know what "consistent" means in this context?

-Dave