"AIDS Treatment News" online * New Issue #302 (searchable/indexed)

F. Frank LeFever flefever at ix.netcom.com
Mon Oct 26 22:23:37 EST 1998

Sometimes less is more.  Better to pin this fellow down to one point at
a time.

Par example,

In <3634fdab.238354 at netnews.worldnet.att.net>
johnburgin at worldnet.att.net writes: 

- - - - - - - - -(snip)- - - - - - - - - - - - - - - - -

>>> When you can tell me why
>>>HIV+ individuals without the use of your drugs are living well past
>>>the ever extending latent phase of the HIV to AIDS timetable,
>>Distribution? Standard deviation? ANY DATA AT ALL? I thought not.
>Would a thousand people be considered signifigant to you?  Two
>thousand?   http://www.heal-la.org/

- - - - - - - - -(snip) - - - - - - - - - - -

I think this exchange in itself suffices to show that this "johnburgin"
(pseudonym?) is not equipped to deal with scientific writing, howsoever
much he reads (if he does).  This is something we have to go over
endlessly with undergraduates (and often fail to get the point across
before the term ends):  "significant" in scientific usage is a
shorthand term for the fuller expression: "significant of a non-chance
finding"  (e.g. a difference from a comparison value too large to be
reasonably accepted as reliable, a difference one would RARELY get by
chance when sampling from a population in which there really is no
difference, etc.)

This is DIFFERENT from usage in other forms of discourse.  The life or
death of ANY individual is "significant" in the sense of being
important to that individual, to his friends and family, etc., etc.;
but (to answer his question), NO, a thousand or two thousand is NOT
necessarily significant in the scientific (inferential statistical)

As the other fellow (Carlton?) said, let us know the populations being
compared,  the method of sampling AND THE STANDARD DEVIATIONS INVOLVED
so we can calculate the standard error and determine whether this is a
significant departure from expectation.

F. Frank LeFever, Ph.D.
New York Neuropsychology Group

More information about the Immuno mailing list