paired duos data with exclosure cages

[Anthonin]

Hi,
( I'm French sorry if english bad)
I need help on how treat my data.

For 12 weeks i have followed grass hight for 4 grassland (ABCD) . With 1 mesurement per week per grassland.
In each grassland, i have placed exclosure cage so animals cannot eat grass.
for each grassland :
- A : 9 exlcosure cages
- B : 6 exlcosure cages
- C : 6 exlcosure cages
- D : 8 exlcosure cages

mesurement where done like this :
Once a week on each grassland
5 mesurements per exclosure (each corner and middle) AND 5 mesurements 10 meter in west (each corner and middle of an imaginary square)
SO each exclosore is paired with an imaginary corner where grass can be eaten.

so in total for exemple on grassland A :
6*5 mesurement on exclosure AND 6*5 on correspondant imaginary square
this 12 times for each


Here is the stat question : does the grass in exclosure is highter than in the imaginary square ? ( i want to see if red deer uses my grassland, then determine wich one are the best nor with places are most used in the grassland)


thanks

tell me if i need to do something to be more precise

 

[Karabiner]

So your total sample size is n=9+6+6+8=29.

I'd say that you could use a "mixed" analysis of variance, i.e. a repeated-measures ANOVA with the within-subjects
factor "week" (12 levels), a within-subjects factor "location of measurement" (2 levels: inside, outside), and a within-
subjects factor "measurement" (5 levels), plus a between-subjects factor "type of grassland" (4 levels). You seem
to be interested in the main effect of "location of measurement", and in the interaction between "type of grassland"
and "location of measurement" (i.e. is the difference between inside and outside different between types of
grassland).


Just my 2pence

Karabiner

 

[Anthonin]

Hi
Thanks for your answer,
in fact the total sample is 58, because each enclosure come with an imaginary square

Im ok with you for the anova, i have turned into a mean the 5 mesurement of each plot,
i would be a 3 way repeated anova ? with factor A : grass land type(A/B/C/D), factor B : location of mesurement : ins/out, Factor time ( 12 levels)

is that correct ?

 

[Karabiner]

in fact the total sample is 58, because each enclosure come with an imaginary square

These are paired data. Therefore IMO you should rather consider it as n=29 sites, and at each site you
have 2 locations (inside and outside).

i have turned into a mean the 5 measurement of each plot,

This is not really necessary, but maybe it will do no harm.

i would be a 3 way repeated anova ? with factor A : grass land type(A/B/C/D),

I wrote: mixed ANOVA. Grass land type is not a repeated-measures (witin-subjects) factor,
but a between-subjects factor (grouping factor). Since you aggregate the 5 measurements by
averaging, you have left 2 within-subject factors: "location" (inside/outside) and "week".

With kind regards

Karabiner

 

[Anthonin]

oh sorry for malentendu, i'm new to this type of analysis
i'm going to do that, and keeping you updated of the result

Thanks a lot
Anthonin

 

[Anthonin]

I have just finished
i have only week and location who are significative

I'm kinda shocked by this because CD grassland where very less used than AB but it seem that they are all the same, with grass shorter outside, as planed
i dont realy know how to interpret time other than saying that grass is taller with time, inside and oustide, showing that deer eat less in grassland over time



EDIT : I have run another test, indeed for grassland D i only had 9 week of survey
So i decided to only analysis 9 week of each grassland
comes out everything is significant now, and to me, this method is less biaised and more accurate