Hello,

I am in trouble with relationships between various
methods used to describe dispersion patterns of
individuals among samples.
In benthic ecology, the fisher index (I=variance to
mean ratio) is in use to test for non-randomness in
spatial dispersion. Another way to describe dispersion
is the Taylor's power law where var=a*mean^b, with b=1
indicates randomness and b>2 aggregation.
A combination of the two gives I=(a*mean^b)/mean and
for b=1, I=a. 
A chi-square test is used to test the statistical
significance of I*(n-1), with n the number of samples.
Then for b=1, the test for I is a*(n-1). Surprisingly
(to me) the test for dispersion  becomes a function of
the sample size. Furthermore, for a=2 and n>9 for
example, there is a significant departure from
randomness, despite b=1. So that the "power" of the
test decreases when sample size increases. 
Taylor (1961) states that the term "a" in the power
law is a sampling factor. Could we consider from the
relation between fisher index and power law that "a"
actually is a function of sample size? This is of
interest for me because the power law is sometime used
to deduce the sample size needed to achieve a given
precision. 
Many thanks for any comment about this.  

Lenaick


___________________________________________________________
Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en fran�ais !
Yahoo! Mail : http://fr.mail.yahoo.com
.
.
=================================================================
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