On 18 Apr 2002 14:09:05 -0700, [EMAIL PROTECTED] (Voltolini) wrote:

> Dear friends, I would like to know how to discuss the Gamma correlation
> results !
> 
> For example, I am using the software STATISTICA and I have a result like:
> Gamma=0,37; Z=5,47; p=0,00.
> 
> But.... what is the meaning of 0,37 ???
> 
> I found in some books that this is a probability and not a simple
> correlation index like Pearson or Spearman correlation.

Do you mean, not one of the books gives you the formula?
So you wonder what it is that the "probability"  measures?

 - I believe that gamma is included in the SPSS output;  and 
some documentation will be on the SPSS web site (crosstabs).
The old paper documentation says, P-Q  over P+Q, where 
P and Q have these really awkward and confusing definitions
(under the write-up for Kendall's tau)  as the number of
"concordant" and "discordant"  pairs of ranks.  If ties are not
possible, then P over P+Q  would be the fraction "concordant".
P-Q  seems to stretch or adapt  that scaling to the 
correlation-type range of minus 1 to plus 1.  


A:  I could ask:  Do you really want to use a stat-pack that 
doesn't document what they are giving you?
B:  On the other hand, if you want a discussion to relay
to your students, instead of a formula, I haven't given you 
much to work with, either.

-- 
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to