Re: History of French

From: Herman Rubin (hrubin_at_odds.stat.purdue.edu)
Date: 09/23/04 

Date: 23 Sep 2004 15:45:21 -0500

In article <nsanders.DIE.SPAM-26D04F.22591421092004@news.verizon.net>,
Nathan Sanders <nsanders.DIE.SPAM@wso.williams.edu> wrote:
>In article <cipga7$4dfi@odds.stat.purdue.edu>,
> hrubin@odds.stat.purdue.edu (Herman Rubin) wrote:

>> In article <nsanders.DIE.SPAM-F347F2.19335919092004@news.verizon.net>,
>> Nathan Sanders <nsanders.DIE.SPAM@wso.williams.edu> wrote:
>> >In article <s6fnk05046446df31n29i7bvc9f45uspf4@4ax.com>,
>> > Mxsmanic <mxsmanic@hotmail.com> wrote:

>> >> > Just because think you've seen or heard it, doesn't mean it's true.

>> >> Why not?

>> >* because you have a tin ear for phonetics

>> Possible.

>Keep this in mind for a response to a later question of yours...

>> >* because your examples are not statistically significant

>> This is utter nonsense.

>You think it's scientifically tenable to make a claim about an
>arbitrarily large population based soley on evidence from a couple of
>people? How many such claims do you think would (or should) be
>accepted by the scientific community?

How many people do you think you need?

Drug testing typically has about 3000. Public opinion
polls are on the order of a few thousand. The size of
sample needed essentially does not depend on that of the
population if random sampling is used.

And a negative statement can be made on ONE case.

>> >* because your examples are not randomly sampled

>> I doubt that anyone, other than someone who deliberately
>> starts out to produce a random sample, has such.

>Which is why accepted scientific conclusions are based on planned
>experiments, not on after-the-fact analysis of memories of a few
>chance encounters.

They are? Not in astronomy or ecology. How much planning
went into Darwin's observations, or those of Wallace, which
had essentially the same conclusions?

How much planning went into the key conclusion of Rutherford
that cosmic ray particles induced nuclear reactions? Or
even those of Lavoisier, who started chemistry by demolishing
the phlogiston theory of heat?

How much planning went into Galileo's conclusion that the
period of a pendulum did not depend on the amplitude, or
Newton's arriving at universal gravity from the apple?

>> Considering the problems in producing a random sample,
>> I doubt that this has ever been done.

>I used "random sample" here in the sense used in the experimental
>sciences. Is it "truly random" in a purely mathematical sense?
>Usually not, due to issues of practicality. But it's an accepted term
>in research.

The people who toss it around like this use statistical
procedures based on the mathematical sense. Not only
that, they misuse them.

>> Stop throwing around statistical terms when you do not
>> know what they mean.

>Or perhaps you should learn what the terms mean in the fields that use
>them, especially the ones relevant to the current discussion.

Mathematics and statistics are what are called "context free".
They can be used in ANY situation, but those using them need
to know the language. The essence of mathematics is that
statements are PRECISE, which is also what is needed for good
communication. The language has a rigid grammar and a SMALL
vocabulary, and one cannot use sloppiness and expect to get
reasonable answers.

>Complaining to me that the ubiquitious term "random sample" as used by
>experimental researchers describes something that isn't truly random
>is silly pedantry. Take it up with academia, not with me.

You are tossing it around.

>> means. There is NO direct relationship between statistical
>> significance and importance, and the null hypothesis is
>> rarely even tenable.

>I never said anything about "importance", only confidence of
>correctness.

The same holds there. The null hypothesis is rarely even
tenable, and what is tested is not that. The question is
not correctness, but appropriateness, and the user has to
be the one to quantify it.

>> >measured with a tool that does not accurately measure the data.

>> How do you know this?

>I refer you back to the top of this post, regarding mxsmanic's tin ear
>for language, which is well-known and well-documented on this group.

I am quite capable of making the distinctions which you accuse
him of not making, and I also am cognizant of the problems from
those who do not use language precisely enough. 

-- 
This address is for information only.  I do not claim that these views
are those of the Statistics Department or of Purdue University.
Herman Rubin, Department of Statistics, Purdue University
hrubin@stat.purdue.edu         Phone: (765)494-6054   FAX: (765)494-0558

Relevant Pages

  • Re: _Verum Et Factum Convertuntur_ (or: Surprised By Syntax)
    ... >I still don't know what you mean by 'mathematical language'. ... to understand the syntax of mathematics is quite small. ... >>>I would not agree that the major part of mathematics is communication, ... "learned" probability and statistics. ...
    (sci.lang)
  • Re: _Verum Et Factum Convertuntur_ (or: Surprised By Syntax)
    ... >> Notice that my quotation stated that Ebonics IS a language, ... >Mathematics, statistics and science do indeed deal with precise concepts ... successor, and having the property that if a set has an element ...
    (sci.lang)
  • Re: Riemann integrability of a composite function
    ... >Department of Mathematics and Statistics, ... On a compact set, g is uniformly continuous and bounded. ... Herman Rubin, Department of Statistics, Purdue University ...
    (sci.math)
  • Re: English is a mess.
    ... claiming that mathematics is a language, because if this is his idea ... English that, like any field of specialization, includes some extra ... language (and indeed you don't acquire them like other features, ... Herman Rubin, Department of Statistics, Purdue University ...
    (sci.math)
  • Re: Lisp Math Engine
    ... >>descendant of S) for statistics, I don't know much about it, but the ... > R is good for statistics and supports the associated mathematics ... > As far as the Lispiness of the language, ... > the Lisp analogy. ...
    (comp.lang.lisp)
Loading