Re: Licas and Probability-Statistics

Hazard, a Licas thread.

Comments mostly on the French Proverb introduced by Licas
a couple of weeks ago (thread on Poisson). Translation and
commentary.

On Tue, 25 Oct 2005 10:21:47 EDT, "\"Luis A. Afonso\""
<licas_@xxxxxxxxxxx> wrote:

> Pekka Jarvela (Sep 8, 2005, 5:38) asked:
>
> *** QUESTION: Is there a way to determine which group or combination of groups causes p to be less than 0.05 so that null hypothesis H0: mean(A) = mean(B) = mean(C) = mean(E) = mean(F) will be rejected?***
>
> Richard Ulrich answered (same day 7:00):
>
> ***It might mean that three of the groups tend to differ***
>
> My correction (at the same day 8:12 PM)
>
> ***All text books are careful to point out that the complementary condition of ______
> meanA = meanB = meanC = ... = meanZ
> is that at *least one of these means is different* from the others. The pekka´s concern makes sense.***
>

Licas, you failed before to cite a textbook to support you.
You tried this, and I replied -
===== from Sept 9
Licas>
> Page 460 - (Hogg and Tanis) PROBABILITY and STATISTICAL INFERENCE, 4th edition
>
> _Let X1, X2, X3, X4, be independent r.v. that are N (mu, sigma^2). We shall test
> ___________H0: mu1, mu2, mu3, mu4
> Against ALL (my capitalization) alternative based (etc). (end of quotation)

RU>
Did you cite this too hastily? I don't see any statement with
a set of equal signs, or saying that one group must test as
different from the others.
====== end


Licas, I keep thinking of that French proverb you quoted.
*faîtes comme moi, quand j´ignore un sujet je ne parle point*.

Licas gave a poorer translation than Google's. I reported,
*ridges like me, when I am unaware of a subject I do not speak*. ...

After I corrected the spelling (no accent over the i), Google gave,
"Make like me, when I am unaware of a subject I do not speak."

Licas offered, phrase by phrase, "Do as I do / when I ignore a
matter / I do not say (absolutely) nothing."

I don't know whether that first phrase is supposed to be
condescending, or just a mild introduction. "Make like me"
is fairly literal, I think, and is intelligible as informal or
slang for "Do as I do." Either version of the first phrase is fine.
And the final phrase seems to mean "I say nothing" or "I do
not speak", which Google renders as good English, where
Licas has a double negative. "I say (absolutely) nothing"
would be what Licas intended, I think.

Licas's "when I ignore a matter" is not, I think, acceptable for
the second phrase. "unaware" seems to carry a better meaning.

Licas is misled by preserving 'ignore', which is a 'false friend'
as linguists call it; the meaning does not carry across to English,
since the translation that results is simply a definition. Ignoring
can be snubbing, or keeping silent. One version of ignoring, is
saying nothing. "When I ignore something, I don't speak."
That tautology would never be a proverb or adage.

-- "ignoring a matter" is not the same as "being ignorant
about a matter" and *ignorance* seems to be the gist.

Loosely reworded, the French means "If you don't know what
you are talking about, shut up."

I wonder if this proverb is possibly an obscure one, even in
Europe. I suspect that Licas has heard this proverb all his life,
because his acquaintances have tried to give him a message.
Unfortunately, he always assumed it was a generality aimed at
other people and "never took it personally".

Another, more pointed version, might be, "Since you don't
know what *we* are talking about, please shut up." -- That is
a bit broader, since it objects to the well-thought-out,
accurate statement that happens to be totally irrelevant.



[snip, a couple of lines.]
Licas >
> My comment: the example *cooked* by Richard´s irrelevant to the discussion: WHAT is the complementary Hypotheses of H0, i.e., what is H1?

To Licas: "Since you don't know what *we* are talking about,
you might shut up... or be more polite in introducing your own
topic." It is okay to introduce new considerations, but it is rather
unusual for a non-statistician to presume that he can control the
statistical discussion.

We, the rest, were talking about p-values of follow-up tests.
A lot has been written on the subject, which I'm pretty sure
that Licas has failed to read. Even now.

By the way -
The complement to "all means are equal" is "The set of means
varies more than would be expected by chance."
Can that be written as a slash through one of the equal-signs? -- No,
not always.

The Original Post was concerned with *odd* outcomes, and not
with the ones that are easy to explain. None of us had any
interest in discussing the outcomes that are easy to explain.

[snip, Licas writing about outcomes that are easy to explain.]

--
Rich Ulrich, wpilib@xxxxxxxx
http://www.pitt.edu/~wpilib/index.html
.