I would like to ask about correct statistic method. We did an observation of wild population animals in a specific place. We observed that species tend to group differently. We created a 5 size categories, and we have 4 species. Is it correct to use some kind of anova for confirming that there is a difference in grouping among species?
$\begingroup$
$\endgroup$
1
-
$\begingroup$ Do you have counts of 4*5=20 categories? $\endgroup$– gung - Reinstate MonicaCommented Oct 9, 2015 at 12:50
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
If I understand you correctly you have 5 categories that describe the size of the group, such as "1)very small group", "2)small group", "3)medium", "4)large", "5)very large group". I'm guessing the groups where to hard to observe or moved to fast for you to be able to count them.
Since your categories fall into a natural order you might be able to use ordered logistic regression. You can analyze this form of data with R using the polr function in the MASS package. There is a worked example here. The paper by Peter Flom "Multinomial and ordinal logistic regression using PROC LOGISTIC" is useful for understanding these models. The interpretation of these models for me personally is a bit tricky. They should be able to tell you something like "species A is more likely to be observed as a very large group" while "species B is more likely to be observed as a very small group."
One thing to consider when interpreting your results is whether the size of the group and the behavior of the animals influenced your ability to see them.
If you post a question about modeling the data please be sure to include code and sample data.
