Re: null means null

[Phillip Good]

Ah, poor reading habits defeat so many.  I wrote that one can only apply permutation methods if one has a null hypothesis.  And was wrong in the case of a single sample where exchangeability follows on assuming symmetry (under the primary hypothesis) about 0.75.

Dennis Roberts <[EMAIL PROTECTED]> wrote:

At 12:05 PM 2/24/2004, Phillip Good wrote:
>Null means null. A terifying habit is to state in error that the primary
>hypothesis is null when it is not. For example:
>
>
>Dennis Roberts <[EMAIL PROTECTED]>wrote:
>
>"I would say the null is .75 ..."

Geez Phillip ... so sorry if I terrified you or anyone else.

And I am willing to admit I was in error and that we should say that H(1)
is .75 in this case. However, as you say below ... one has to convert the
hypothesis to a null form ... pray tell what will that null be in this
case? Will we just toss out 50/50 ... ??? Where's the logic in that? This
is precisely why ... formulated nulls in most cases are meaningless.

One "could" argue that the null is that there is no difference between
drive thru and inside sales ... ie, p and q are .5 ... if that is t! he case,
then the alternative is that it is not 50/50 ... it's not that it is .75.
Thus, when it is stated as .75 ... I don't really see a null that goes
along with that.

What I would do in this case is to build a 95% CI based on the given data
over the 50 days ... and see if .75 is inside or outside of that CI.

Thus, I will test the .75 against the CI ... but I don't (now that you have
corrected me) have a clue as to what the null would be. So, what I am
asking you is ... what IS the null in this case? What hypothesis (just one)
are we trying to nullify?


>
>Do you Yahoo!?
>Yahoo!
>Mail SpamGuard - Read only the mail you want.

Dennis Roberts
[EMAIL PROTECTED]
http://www.personal.psu.edu/users/d/m/dmr/droberts.htm


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

Phillip Good
http.ms//www.statistician.usa
"Never trust anything that can think for itself if you can't see where it keeps its brain."  JKR

---

Do you Yahoo!?
Yahoo! Mail SpamGuard - Read only the mail you want.