On Sat, 29 Jan 2000 [EMAIL PROTECTED] wrote:
> In a message dated 1/29/2000 1:54:04 PM Eastern Standard Time,
> [EMAIL PROTECTED] writes:
>
> > Of course not. Significance level cannot be computed: it is an
> > arbitrary choice of the investigator (or the analyst).
>
> I think there is a semantical misunderstanding here.
Precisely. That was the point of my response.
> Significance level may be computed for a given numerator degrees of
> freedom, denominator degrees of freedom, and observed value of F.
This is not true. What you can COMPUTE is the probability that the
observed value of F (or greater) would occur if the null hypothesis
(of no differences among any of the population means, which is consistent
with an F value of 1) were true.
> Generally, investigators set a significance level which their, say,
> regression coefficients must meet. That's different.
No, it is not. Same principle, applied to different parameters.
> I found several web pages which compute significance level. For
> example, see site http://www.cytel.com/statable/
You can find all sorts of damfool things in the world if you look hard
enough. That you found one (or even several) doesn't mean it's right.
Even respected statistical packages err on this point: SPSS, for
example, routinely uses the label "Sig." when what is reported is "p".
What these routines compute is not in fact significance level, it is the
observed probability value corresponding to the value of F (or t, or z,
or whatever) one supplies. Useful only for determining whether a
particular result meets the significance level chosen by the investigator
or analyst.
I am reminded of the conundrum attributed to A. Lincoln:
"If you call a tail a leg, how many legs does a horse have?" His answer:
"Four. Calling a tail a leg doesn't make it one."
Just so.
-- DFB.
------------------------------------------------------------------------
Donald F. Burrill [EMAIL PROTECTED]
348 Hyde Hall, Plymouth State College, [EMAIL PROTECTED]
MSC #29, Plymouth, NH 03264 603-535-2597
184 Nashua Road, Bedford, NH 03110 603-471-7128
===========================================================================
This list is open to everyone. Occasionally, people lacking respect
for other members of the list send messages that are inappropriate
or unrelated to the list's discussion topics. Please just delete the
offensive email.
For information concerning the list, please see the following web page:
http://jse.stat.ncsu.edu/
===========================================================================
