# Picking the correct test, need help urgently!

#### Annielk

##### New Member
Hello,
I feel like the answer for this is probably really obvious but I am being missing it altogether. This is my data:

Habitat Associating Not Associating Total Observations
Flood 51 0 51
Levee 524 72 596
Liana 162 98 260
Total 737 170 907

[/IMG] <-- as the table formatting has gone wrong when I typed it up.

Basically, my null hypothesis is: The frequency of time monkeys spend associating with other species as opposed to not associating occurs significantly differently in different habitats.
The 3 habitats were the squirrel monkeys were observed are flooded forest (flood), levee and liana. The data is obviously discrete so I would have usually thought a chi2 would be appropriate. However, my issue is that I cannot assume independence between the "observed associating" or "observed not associating" (an association counts as when a squirrel monkey was observed as sharing the habitat with a different species). Therefore, I am not sure what statistical test to use to see if there is a significant different between the proportion of time spent associating dependent on habitat.
Any help would be really appreciated as I have been stuck on this for a couple of days now! Thanks!!

#### ajestudillo

##### New Member
I think you can do Cramer's correlation to see if the time spent associating with other species change as a function of the habitat. Alternatively, you can use a multinomial regression. But not very sure of how it works...

#### Karabiner

##### TS Contributor
Just to be sure, it is not possible to determine individual behaviours
(e.g. "monkey #A, from the in flooded forest, was observed
associating 3 times and not associating 2 times")?

With kind regards

K.

#### Annielk

##### New Member
Just to be sure, it is not possible to determine individual behaviours
(e.g. "monkey #A, from the in flooded forest, was observed
associating 3 times and not associating 2 times")?

With kind regards

K.
The data wasn't recorded on an individual basis. We observed a group and have a lot of data for their behaviors and habitat etc. A line of data for example would include squirrel monkey, group size etc. Associations were recorded on the side where it was noted If capuchins were present per data entry. Each entry was taken every 3 minutes. An association wasn't a direct interaction just the presence of the capuchin species. I hope this answers your question?