I have proportion data for 10 behaviours for morning, day, and evening and the initial beta regression shows that there is a difference in the proportion of behaviours depending on time of day. I'm now trying to find a test that will let me do a pairwise comparison of these three times to see which behavoiurs are different and driving this result. Ie. morning rest is 0.8, day is 0.5, and night is 0.9, feed is 0.7 morning, 0.69 day, and 0.72 night (these aren't the exact numbers but just to give an idea of what my day looks like).
thanks,
So far I have tried an anova and a manova, but having to link each proportion to the respective behaviours seems to cause these tests to fail as it doesn't link these variables properly.
