R help - Mar 2020 - Help! Can a nested fixed effects model with autocorrelation be possible?

Hi everyone!
I'm a little bit lost as to which statistical model I should use.

I'm evaluating the decomposition of two plant species (factor:
"species")
under two treatments: "temperature" (with 2 levels) and "water
level" (3
levels). These two treatments are completely crossed. I have 5 replicates
per plant species for each time period (6 dates).
So I have 6 dates x 2 species x 2 temperatures x 3 water levels x 5
replicates: 360 obs. (180 obs. per species).

The comparison between plant species isn't relevant in my study, rather I
want to test the significance of the treatments within the plant species.
So I think that a nested model could fit my data. Could this be possible?

On the other hand, I've been reading that there could be temporal
correlation problems in my data, and that this can be addressed with nlme
package. But as I understand you can't fit a nested fixed effects model
using lme. My design doesn't have random effects, they are all fixed
factors, and, from what I understand, time can't be used as a random effect
because it is not a categorical variable. Furthermore, "species"
can't be a
random factor because it only has two levels. In summary, I'm completely
lost.

My questions are:
Can I use a nested model for the structure of my data?
Is there such a thing as a nested fixed effects model with autocorrelation
approach?
Can I use lme or lme4 if that model exists?
Could I transform time variable to a categorical factor and use it as a
random factor nested in species?
Is there a simple way to analyse this data set?

Thank you!!

(I'm very sorry for my english, it is not my native language)
-- 
Candela Madaschi

	[[alternative HTML version deleted]]

Your English is excellent, but your statistics is bad.

This list is for questions about R programming, not statistics, although
sometimes the two topics do intersect. If you wish to seek online help of
this sort, the  special R Help list r-sig-mixed-models may be where you
should post, as it appears that mixed effects models might be useful for
you. Another specific statistical help site is stats.stackexchange.com .

However, I think you would do much better to consult with a local
statistical expert, as your statistical background seems minimal. But
that's just my opinion.

Bert Gunter

"The trouble with having an open mind is that people keep coming along and
sticking things into it."
-- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )


On Wed, Mar 4, 2020 at 10:30 AM Candela Madaschi <candelamadaschi at
gmail.com>
wrote:
> Hi everyone!
> I'm a little bit lost as to which statistical model I should use.
>
> I'm evaluating the decomposition of two plant species (factor:
"species")
> under two treatments: "temperature" (with 2 levels) and
"water level" (3
> levels). These two treatments are completely crossed. I have 5 replicates
> per plant species for each time period (6 dates).
> So I have 6 dates x 2 species x 2 temperatures x 3 water levels x 5
> replicates: 360 obs. (180 obs. per species).
>
> The comparison between plant species isn't relevant in my study, rather
I
> want to test the significance of the treatments within the plant species.
> So I think that a nested model could fit my data. Could this be possible?
>
> On the other hand, I've been reading that there could be temporal
> correlation problems in my data, and that this can be addressed with nlme
> package. But as I understand you can't fit a nested fixed effects model
> using lme. My design doesn't have random effects, they are all fixed
> factors, and, from what I understand, time can't be used as a random
effect
> because it is not a categorical variable. Furthermore, "species"
can't be a
> random factor because it only has two levels. In summary, I'm
completely
> lost.
>
> My questions are:
> Can I use a nested model for the structure of my data?
> Is there such a thing as a nested fixed effects model with autocorrelation
> approach?
> Can I use lme or lme4 if that model exists?
> Could I transform time variable to a categorical factor and use it as a
> random factor nested in species?
> Is there a simple way to analyse this data set?
>
> Thank you!!
>
> (I'm very sorry for my english, it is not my native language)
> --
> Candela Madaschi
>
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> 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.
>
	[[alternative HTML version deleted]]

Hi Candela,
Try posting your question to the R-SIG-mixed-models list. You will
probably get an expert answer there.

Jiim

On Thu, Mar 5, 2020 at 5:30 AM Candela Madaschi
<candelamadaschi at gmail.com> wrote:
>
> Hi everyone!
> I'm a little bit lost as to which statistical model I should use.
>
> I'm evaluating the decomposition of two plant species (factor:
"species")
> under two treatments: "temperature" (with 2 levels) and
"water level" (3
> levels). These two treatments are completely crossed. I have 5 replicates
> per plant species for each time period (6 dates).
> So I have 6 dates x 2 species x 2 temperatures x 3 water levels x 5
> replicates: 360 obs. (180 obs. per species).
>
> The comparison between plant species isn't relevant in my study, rather
I
> want to test the significance of the treatments within the plant species.
> So I think that a nested model could fit my data. Could this be possible?
>
> On the other hand, I've been reading that there could be temporal
> correlation problems in my data, and that this can be addressed with nlme
> package. But as I understand you can't fit a nested fixed effects model
> using lme. My design doesn't have random effects, they are all fixed
> factors, and, from what I understand, time can't be used as a random
effect
> because it is not a categorical variable. Furthermore, "species"
can't be a
> random factor because it only has two levels. In summary, I'm
completely
> lost.
>
> My questions are:
> Can I use a nested model for the structure of my data?
> Is there such a thing as a nested fixed effects model with autocorrelation
> approach?
> Can I use lme or lme4 if that model exists?
> Could I transform time variable to a categorical factor and use it as a
> random factor nested in species?
> Is there a simple way to analyse this data set?
>
> Thank you!!
>
> (I'm very sorry for my english, it is not my native language)
> --
> Candela Madaschi
>
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> 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.