Struggling mightily with some analyses for my thesis. I am using SAS for this analysis and trying to do MANOVA.

I have some time-activity budget observations (n=1109) of waterfowl in multiple habitats.

My question is do behaviors differ among the 6 habitat types among 3 time periods throughout winter.

DV: (proportions) of time spent in 7 behaviors
IV: Habitat (6), Month (3)
Random: Year (2)

I know the data is non-normal and transformation hasn't worked. I'm willing to live with this violation bc MANOVAs (wilks lambda) are robust to violations of normality.

Determining outliers has become confusing. I'm trying to use mahalanobis distance and getting all kinds of mixed up. Trying to regress MD on the centroid of the variables and then select which ones are outliers and whether or not they make biological sense. I'm getting there with this portion, but still a little foggy.

Reduction of variables is another question. Do I use PROC candisc? and what does it all mean?

I need help badly.... it has taken me weeks to get this far and I need something within the next few days to present to committee members (not for actual defense tho).

Any help with coding/understanding the process is greatly appreciated.