R help - Dec 2013 - Repeated me

[ I've also sent this message to other lists. Sorry for multiple messaging ]

Dear colleagues,

I want to perform a repeated measures two-way ANOVA (two fixed crossed
factors). I've found the way to do it (in SPSS, and also in R), but anyway
I think my data don't meet the requirements for that analysis (that is,
normality, sphericity and so on).

Anyway, I was told (here in this list and also in person) to consider the
subjects IDs as a third factor, and to perform a classical ANOVA (because
if I do that I will have one single observation for each combination of the
three factors). That way, I wouldn't need to check for
"sphericity" but for
the usual ANOVA assumptions.

My first question is related to that:

(1) Just for checking if I understood this right: are both (parametric)
approaches equivalent? I mean: is "pure" repeated measures ANOVA (the
one
available in SPSS, for instance) equivalent to by-passing it by making
subjects work as a factor and then applying "classical" ANOVA?

Another different approach is using non-parametric alternatives. I've found
ONE-way non-parametric tests both for repeated measures comparisons and for
independent sets of observations. And also I've been told about the
existence of multiple-factor non-parametric ANOVA (based on ranks, and also
based on permutation exact tests).

My research is not about ANOVA; I just want to use it as a way to compare
results in a more sophisticated and scientific way than just saying which
combination is better. So I am looking for the easiest approach.

And because of that, I am thinking of performing an exact
(permutation-based, non-parametric) ANOVA, using subjects as a factor.

(2) I know that I didn't tell you anything about my data or my context, but
do you think that approach can be appropriate?

Looking forward to your answers,


--
vicent
@vginer_upv
about.me/vginer_upv

	[[alternative HTML version deleted]]

a) Cross posting is not something to apologize for... it is something to not do.

b) This message is off topic here. Please read the Posting Guide mentioned in
the footer of this or any other any R-help message before posting here again.
---------------------------------------------------------------------------
Jeff Newmiller                        The     .....       .....  Go Live...
DCN:<jdnewmil at dcn.davis.ca.us>        Basics: ##.#.       ##.#.  Live
Go...
                                      Live:   OO#.. Dead: OO#..  Playing
Research Engineer (Solar/Batteries            O.O#.       #.O#.  with
/Software/Embedded Controllers)               .OO#.       .OO#.  rocks...1k
--------------------------------------------------------------------------- 
Sent from my phone. Please excuse my brevity.

Vicent Giner-Bosch <vigibos at eio.upv.es> wrote:
>[ I've also sent this message to other lists. Sorry for multiple
>messaging ]
>
>Dear colleagues,
>
>I want to perform a repeated measures two-way ANOVA (two fixed crossed
>factors). I've found the way to do it (in SPSS, and also in R), but
>anyway
>I think my data don't meet the requirements for that analysis (that is,
>normality, sphericity and so on).
>
>Anyway, I was told (here in this list and also in person) to consider
>the
>subjects IDs as a third factor, and to perform a classical ANOVA
>(because
>if I do that I will have one single observation for each combination of
>the
>three factors). That way, I wouldn't need to check for
"sphericity" but
>for
>the usual ANOVA assumptions.
>
>My first question is related to that:
>
>(1) Just for checking if I understood this right: are both (parametric)
>approaches equivalent? I mean: is "pure" repeated measures ANOVA
(the
>one
>available in SPSS, for instance) equivalent to by-passing it by making
>subjects work as a factor and then applying "classical" ANOVA?
>
>Another different approach is using non-parametric alternatives. I've
>found
>ONE-way non-parametric tests both for repeated measures comparisons and
>for
>independent sets of observations. And also I've been told about the
>existence of multiple-factor non-parametric ANOVA (based on ranks, and
>also
>based on permutation exact tests).
>
>My research is not about ANOVA; I just want to use it as a way to
>compare
>results in a more sophisticated and scientific way than just saying
>which
>combination is better. So I am looking for the easiest approach.
>
>And because of that, I am thinking of performing an exact
>(permutation-based, non-parametric) ANOVA, using subjects as a factor.
>
>(2) I know that I didn't tell you anything about my data or my context,
>but
>do you think that approach can be appropriate?
>
>Looking forward to your answers,
>
>
>--
>vicent
>@vginer_upv
>about.me/vginer_upv
>
>	[[alternative HTML version deleted]]
>
>______________________________________________
>R-help at r-project.org mailing list
>https://stat.ethz.ch/mailman/listinfo/r-help
>PLEASE do read the posting guide
>http://www.R-project.org/posting-guide.html
>and provide commented, minimal, self-contained, reproducible code.