C'mon, guys, ease up a bit....

I'd thought it obvious that "Albert" (and/or "James Lo", and/or whoever
else is using the account <[EMAIL PROTECTED]>) is a student in an
introductory stats class, and that they've been introduced at least to
the normal (aka Gaussian) distribution (and a table thereof) and to the
Central Limit Theorem, but not yet to "Student"'s t distribution.  And
they have been asked to construct a confidence interval as an exercise.

Since the machinery now available to them only works if the population
variance is known, the _deus_ex_machina_ (alias the textbook author, or
the course instructor) has kindly provided the population mean and
variance as separate information:  known, as it were, full-blown from
the mind (or forehead?) of Zeus.

This is of course not very realistic, in that one would not in a "real"
situation be likely to know these values (as several correspondents have
been at pains to point out);  but this is an <exercise>, and the point
of an exercise is NOT to be realistic, but to provide an artificial
environment in which to practice one aspect (or a few aspects) of the
trade one is in the process of learning about, without having to deal
with all the nasty little complications that are always lurking in the
background (or even the foreground!) in "real" life.

Discussion after the fact would doubtless deal with whether in fact the
CI did embrace the true value of mu (that value falls rather near one
edge of the CI), and therefore whether the present sample mean might be
construed as having arisen from the central 95% or the extreme two-sided
5% of possible samples.

On Fri, 11 Jul 2003, Rich Ulrich wrote:

> On 11 Jul 2003 07:15:38 -0700, [EMAIL PROTECTED] (Jason Owen) wrote:
>
> > Let me get this straight -- you want to create a confidence interval
> > for a mean that you *know* is equal to 53,000?

Well, yes and no.  I doubt that the _querent_ (Albert/James/etc.) wants
to, but their _instructor_ wants them to do so.  And of course for a
known mean:  how else can the instructor point out, afterwards, that the
true mean is (or isn't) within the CI?

   < snip, the rest >

One may note, en passant, that Albert (James, ...) made at least two
errors in the calculations reported in their original post.  One of
those errors was subsequently acknowledged by one of them, who included
a corrected computation in the later post.  In my view it suffices to
remark that an error occurred, and maybe give a hint as to where it
occurred;  to tell them what the error is would be to deprive them of
the pleasure of trouble-shooting it themselves (and thereby beginning to
develop some skill in trouble-shooting).

 -----------------------------------------------------------------------
 Donald F. Burrill                                         [EMAIL PROTECTED]
 56 Sebbins Pond Drive, Bedford, NH 03110                 (603) 626-0816

.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to