Re: ANOVA QUESTION - Terminology

On 10 Sep 2006 20:23:52 -0700, "Reef Fish"
<Large_Nassau_Gr0uper@xxxxxxxxx> wrote:


Richard Ulrich wrote:
On 10 Sep 2006 14:20:32 -0700, "Reef Fish"
<Large_Nassau_Gr0uper@xxxxxxxxx> wrote:
[snip, stuff]

What is Group? Richard Ulrich had argue the quackery concept of
an categorical-ordinal variable and CODE the NOMINAL groups
(such as Asian, Black, and American) as 1, 2, and 3 in the variable
called "group" and do a regression on it. That is called a BLUNDER
of the worst kind by undergraduates.
[snip, rest]
RU > >
Rich Ulrich never argued that.
RF >
Let's just let the readers who understand the use of Indicator
variables
for NOMINAL categories go to sci.stat.math and READ the thread
about Vater Sundh's example to find out that that was exactly what
you did.


To be clear -- Reef Fish Bob is the only person posting who
ever thought that the categories nominal and not ordinal,
since Valter declared them as ordinal as a GIVEN.



Reef Fish Bob recently read 4 or 5 posts that defined and described
and gave examples of ordinal categorical variables, and he never
figured it out. I guess he never figured out that there *can* be
"groups" with ranges of scores.
RF>
No need to coach anyone. The ordinal-categorical idea for the use
of a numerical value is AT BEST questionable even if the groups have
ordinal information. But when they are categorical WITHOUT any
ordinal info, as in Vater Sundh's example, then it is a clear case of
error.
RU >
He implicitly called the original poster a liar.
RF >
I did no such. I just SHOWED how HIS data did not have any
ordinal information in them, and showed him how he erred. He

Here is from Bob's post on Aug 14, under Unsolved Problems -
===== to Valter -
The were NOMINAL with NO ordinal information in them. Merely
6 different categories.

You clearly MIS-treated them as if they were numbers.

Now you came back and gave some definition for "ordinal categorical"
as if it would make your error disappear.

Fairly clever misdirection by DISREGARDING the ENTIRE EXAMPLE
in which you erred, and erred badly.
====== end extract

I call that, "clever misdirection," a case of "implicitly calling
the OP a liar."

If I list for you six categories in a specific order,
and say that they are ordered,
I don't have to give any other "ordinal information"
in order to make them ordinal.

Do I? Well, evidently I do, for Reef Fish Bob.

If I happen to label them A-F to make the ordering
evident in a way, Bob is going to assume that letters
make them Nominal. At least, he did for Valter.

[snip]
RU >
Reef Fish Bob has always been poor with definitions and
with verbal reasoning, but that thread seems to exemplify
another low point.
RF >
Yes, Richard. I've heard that line every time I corrected your
technical errors in Statistics. Isn't it intresting that NO
statistician (a real one who is worth his salt) has EVER even
hinted that my "verbal deficiency" is my problem.

No, Bob. You have heard that line when I correct your own
errors in reading. And poor use of dictionaries, etc. Very
seldom is there any technical error in statistics involved, on
either side. Your problem is in communications, not (usually)
in basic statistics.

At least, I am participating in this thread, so there is a potential
for "correcting my errors". That is a welcome contrast to Bob's
recent habit of casting personal aspersions in every other thread.


As I always said. It's for the READERS to decide.

Yes, you say that. Please listen to when the readers
tell you to change your style.


But I think it has become abundantly clear that Richard Ulrich,
over the cumulative weight of evidence, has lost ALL credibility
and a few of his former cheerleaders who had been led astray
by his malpractice to think that Richard Ulrich really knew some
statistics.

Richard Ulrich's CROWNING SUCCESS was to be the only
person in any statistics group who had EVER said, publicly,
that
Y = bX

is an example of a NONLINEAR model, and NOT a Linear model.

Bob has brought this up three or four times lately. AND, here,
he mis-states the position, gravely, by truncation. The simplest
form of a linear model is not present if "there is a constraint
on b, e.g., b > 0."

Here is some reply.

My point, for those who were not there, is that the teaching
of linear models can start from the simplest, and make it plain
what is lost by each degree of complication. Using X and X^2
screws up the likelihood surface and introduces complications
to the theory of inference. So does introducing a constraint on b.

And so on, for other sorts of changes.

Bob does not, by the evidence, see any need for drawing these
distinctions and investigating or teaching what they imply, alone
or in combination.

Bob calls me to be uneducated or ignorant for raising the suggestion.
Bob, I suggest, again exhibits poor reasoning.



He was also the only person who missed ALL SIX examples on
whether they were linear or nonlinear models, that even students
who FLUNKED the undergrad course using the Neter et al book
on Applied Linear Models (which I recomkmended to jp) would
have answered most of those questions correct.

THAT's all the more reason why Richard Ulrich has no place
advising jp or anyone else on the subject which has a strong
Linear Models background in theory and practice, when he
is completely oblivious to those concepts and had NEVER been
educated in them. But worse of all, having received the FREE
lessons in the sci.stat.math group about Linear Models, he is
STILL missing most of what had been taught there.

The records are all in the Google archives.

Indeed. Why does Reef Fish Bob hate to have his real name
identified with his posts?


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

Relevant Pages

  • Re: ANOVA QUESTION - Terminology
    ... Richard Ulrich wrote: ... and gave examples of ordinal categorical variables, ... technical errors in Statistics. ... Linear Models background in theory and practice, ...
    (sci.stat.consult)
  • Re: ANOVA QUESTION - Terminology
    ... Richard Ulrich wrote: ... Depending on design, ... measured as the raw change score or difference; ... Reef Fish Bob seems to have missed a course. ...
    (sci.stat.consult)
  • Re: collinearity and LR
    ... Keep in mind, when you read, that Reef Fish Bob ... Strict definition of multicollinearity? ... Paige Miller, for whatever misunderstanding he might have had, ... certainly would be a FOOL to listen to Richard Ulrich who is ...
    (sci.stat.edu)
  • Re: Is sum of two normally distributed variables normally distributed?
    ... I am sure including a few posted by Richard Ulrich. ... Don't worry, ... Those are called flames. ... -- Reef Fish Bob. ...
    (sci.stat.math)
  • Re: Comparing two variation coefficients
    ... Richard Ulrich wrote: ... And that is how Reef Fish Bob Ling says, ... sneering way that must repel Replies." ...
    (sci.stat.math)