Date: Wed, 22 Oct 1997 16:46:47 -0500
Reply-To: "Nichols, David" <nichols@SPSS.COM>
Sender: "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From: "Nichols, David" <nichols@SPSS.COM>
Subject: Re: Coefficient of Variation
I don't recall seeing such a test anywhere. We certainly don't
have anything built into SPSS designed for such a purpose. My
guess at how one would approach it would be to transform one
of the variables or groups in a linear manner so that the mean
now equals the other one, but maintaining the CV from the
original scale. The problem has then been restated in terms of
a question of equal variances/standard deviations. You could
then use standard methods for such situations. Only the grouped
situation (comparing the CVs of two groups of cases on the same
variable) is explicitly treated in SPSS; we don't have a test
for equality of variance of correlated variables. You'd want to
make sure to use a robust test like Levene's, as traditional
standards like the Box-Bartlett, Cochran's, or referring the
variance ratio to an F-distribution are not at all robust to
violations of the normality assumption.
Approaching the problem more straigtforwardly involves dealing
with the fact that you're working with ratios, and this makes
things complicated; we certainly don't offer anything along
these lines.
David Nichols
Senior Support Statistician
SPSS Inc.
nichols@spss.com
>----------
>From: Albert Beaulne[SMTP:a.beaulne@V-WAVE.COM]
>Sent: Tuesday, October 21, 1997 9:23 PM
>To: SPSSX-L@UGA.CC.UGA.EDU
>Subject: Coefficient of Variation
>
>How does one know when you have a good CV
>
>If you had two CVs one at .7 and one at .65 how would you know if they are
>different (statistically)
>
>Is there a way to do this analysis in SPSS
>
| 
