In article <916fjl$d63$[EMAIL PROTECTED]>,
Gene Gallagher <[EMAIL PROTECTED]> wrote:
>The textbook I'm using this semester presents a 2-factor ANOVA problem
>(3 levels of each factor) in which two of the 9 groups have zero
>variance (identical observations for two replicates). Levene's test
>indicates significant departure from homoscedasticity (this may not be
>known to the authors of the text who provide the solution as if there
>were no problems with homogeneity of variance). Is there ever a case
>when you can trust the ANOVA results despite violations of
>homoscedasticity like this? Obviously, no transformation is appropriate
>and the non-parametric ANOVAs aren't good at handling interaction
>effects (at least not Friedman).
If one gets identical results, either the variance is
extremely small and this is an accident of rounding, or
something else is happening and the model is inappropriate.
Statistics cannot do your thinking for you. There is
far too much trust of cookbook procedures in any case.
--
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, Dept. of Statistics, Purdue Univ., West Lafayette IN47907-1399
[EMAIL PROTECTED] Phone: (765)494-6054 FAX: (765)494-0558
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
