[R] post hoc tests two way repeated measures anova

[Jean-Pierre Bresciani]

Hi,

is there a specific/appropriate function/package to perform post hoc tests
when running a two way repeated measures anova? I'm looking for something
that will be equivalent to the 'TukeyHSD()' for between subjects anova (with
'aov()'). For one way repeated measures anova, the 'pairwise.t.test()'
function seems to work correctly but the results are questionable for two+
way anovas, as if the adjustment of the p values for multiple comparisons
was not correct.

any suggestion?

jp
--
View this message in context: http://www.nabble.com/post-hoc-tests-two-way-repeated-measures-anova-tp18999089p18999089.html
Sent from the R help mailing list archive at Nabble.com.

______________________________________________
R-h...@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.

[Dieter Menne]

Jean-Pierre Bresciani <bresciani <at> tuebingen.mpg.de> writes:

> is there a specific/appropriate function/package to perform post hoc tests
> when running a two way repeated measures anova? I'm looking for something
> that will be equivalent to the 'TukeyHSD()' for between subjects anova (with
> 'aov()'). For one way repeated measures anova, the 'pairwise.t.test()'
> function seems to work correctly but the results are questionable for two+
> way anovas, as if the adjustment of the p values for multiple comparisons
> was not correct.

Classical repeated measurement has fallen a bit out of favor, because mixed
models as in lme (package nlme) and lmer(package lme4) are MUCH more flexible
and in general more stable when data are missing.

So I recommend to switch to lme (better documented, currently), and use

http://finzi.psych.upenn.edu/R/Rhelp02a/archive/20725.html

as a starter when trying to do multiple comparisons.

Dieter Menne (Tübingen)

[Frank E Harrell Jr]

Dieter Menne wrote:
> Jean-Pierre Bresciani <bresciani <at> tuebingen.mpg.de> writes:
>
>> is there a specific/appropriate function/package to perform post hoc tests
>> when running a two way repeated measures anova? I'm looking for something
>> that will be equivalent to the 'TukeyHSD()' for between subjects anova (with
>> 'aov()'). For one way repeated measures anova, the 'pairwise.t.test()'
>> function seems to work correctly but the results are questionable for two+
>> way anovas, as if the adjustment of the p values for multiple comparisons
>> was not correct.
>
> Classical repeated measurement has fallen a bit out of favor, because mixed
> models as in lme (package nlme) and lmer(package lme4) are MUCH more flexible
> and in general more stable when data are missing.

And don't forget generalized least squares (gls in nlme) which has the
same advantages using only fixed effects.

Frank

>
> So I recommend to switch to lme (better documented, currently), and use
>
> http://finzi.psych.upenn.edu/R/Rhelp02a/archive/20725.html
>
> as a starter when trying to do multiple comparisons.
>
> Dieter Menne (Tübingen)
>

--
Frank E Harrell Jr Professor and Chair School of Medicine
Department of Biostatistics Vanderbilt University

[Jean-Pierre Bresciani]

Hi,

and thanks for the advices. I actually tried using 'lme', but haven't
figured out the appropriate syntax to get the random error correct with
repeated measures.
let's say i have 3 independent variables, with 'aov', i would write
something like: aov(dep_var~(indep_var1*indep_var2*indep_var3) +
Error(subject/(indep_var1*indep_var2*indep_var3)).

With 'lme' however, i can't find the right formula. i tried things like:
lme(dep_var~(indep_var1*indep_var2*indep_var3), random = ~1|subject) or
nesting my independent variables in 'subject', but those are obviously wrong
with my design.

i'm quite clueless (and i haven't found any convincing piece of information
about how to correctly use 'lme' or 'lmer'). So, any advice along that line
is more than welcome.

JP

--
View this message in context: http://www.nabble.com/post-hoc-tests-two-way-repeated-measures-anova-tp18999089p19031169.html

Sent from the R help mailing list archive at Nabble.com.

______________________________________________