Re: Do you want it?

Reef Fish wrote:
Greg Heath wrote:
Reef Fish wrote:
Greg Heath wrote:
Reef Fish wrote:

I think Greg's attention span is good only for a few lines. So, I'll
snip all except those few line:

I think the emphasis in a "sample variance" is that it is obtained from
a SAMPLE of data values. The criterion of estimation does not
alter the fact that they are all called "sample variance", as a short
for of "sample estimate of the population sigma under criterion #".

What modifications to above discussion result when the population
mean, M, is known and the unbiased estimate for the covariance
matrix is

S = Sum(Xi-M)(Xi-M)'/N ?

Hope this helps.

You should snip that line from your sig file. What you post seldom
helps anyone.

That's debatable.

In this case, you're asking a question that is already
answered.

I inferred that it had been answered. However, I can't find where it
had been directly asked. Therefore I made it a point to get a direct
answer to a direct question. In particular,

"Does any part of the sample variance discussion change when
the population mean is used instead of the sample mean?"

Simple direct question asked.

Simple direct answer expected, i.e., either "No" or "Yes, because ...".

A bit too advanced for you to figure out the logic, isn't it, that the
answer is "NO" ?

I figured the answer had to be NO. However, I was not absolutely
100% sure. Therefore, for my benefit as well as others, I wanted
a direct answer.

You S came from a SAMPLE didn't it?

Of course. However, that's not the point. See the word SAMPLE?

But that WAS the whole point.

That was YOUR point...which was well made. However, I
was trying to clarify a condition upon which previous replies
"appeared" to be based (if not from you, then someone else).
Namely, the repeated phrase "when the population mean is
unknown". If that phrase deserved to be repeated, then
certainly the clarification that nothing changes when the
population mean is used is worthwhile.

Now that you've said NO, let me ask you a potentially
redundant question:

Does the test statistic N*s^2/sigma^2 (using M) have
the same probability distribution as (N-1)*s^2/sigma^2
(using XBAR)?

I even had the whole post trimmed down
to 4 lines so that even with your short attention span
you wouldn't miss the point:

I got your point a long time ago in another thread.
However, you still haven't gotten my point.

I think the emphasis in a "sample variance" is that it is obtained from
a SAMPLE of data values. The criterion of estimation does not
alter the fact that they are all called "sample variance", as a short
for of "sample estimate of the population sigma under criterion #".

So, what's the relevance of the unbiased estimate of the
covariance in your question?

An equation with M instead of Xbar for clarification.

Makes no difference in calling it a SAMPLE variance as long as it is
computed from a SAMPLE. The M or unbiasedness are irrelevant.

I agree. Again, however, I was addressing another issue. Namely,
what differences are there in using a sample variance based on
the population mean vs the sample mean? The most important of
these is probably the answer to the question "Are the same
distributions assumed for the sample statistics?"

RF says to Greg the slow learner

It's hard to learn when the so-called teacher prefers to repeatedly
answer his own question instead of yours.

You and Luis A. Afonso should make a good team.

I wouldn't think of breaking up your tag-team tirades.

Hope this helps.

Greg

.

Relevant Pages

  • Re: Do you want it?
    ... The criterion of estimation does not ... alter the fact that they are all called "sample variance", ... for of "sample estimate of the population sigma under criterion #". ... Simple direct question asked. ...
    (sci.stat.math)
  • Re: Do you want it?
    ... alter the fact that they are all called "sample variance", ... You should snip that line from your sig file. ... Twice used by Jack the OP. ... by Afonso and others, you'll be treated equally, in order for the ...
    (sci.stat.math)
  • Re: Do you want it?
    ... alter the fact that they are all called "sample variance", ... You should snip that line from your sig file. ... Simple direct question asked. ... was trying to clarify a condition upon which previous replies ...
    (sci.stat.math)
  • Re: Do you want it?
    ... Jack Tomsky wrote: ... "Now consider the sample variance, ... RF> some estimation criterion. ... RF> For (N-1), # = Unbiased Estimate. ...
    (sci.stat.math)
  • Re: Do you want it?
    ... "Now consider the sample variance, ... RF> some estimation criterion. ... RF> For (N-1), # = Unbiased Estimate. ... mean, M, is known and the unbiased estimate for the covariance ...
    (sci.stat.math)