0

I have one independent categorical variable "Period" (day or night) and 4 dependent continuous variables "Behaviours" (walking, lying, standing, grazing) which are compositional in nature (add to 100% for a given period). Original data for behaviours recorded in decimal hours however as each total varied around 12hrs (some cattle removed for milking etc) I decided to change it to % so all come to 100%.

I am struggling to find a way of analysing this compositional data. Can anyone suggest a suitable test to test how period affects behaviour?

Edit: I do have another independent categorical variable ("Milk.Status: Dry or Milk) if this helps to add it in also. Maybe a MANOVA would work?

Kyle
  • 21
  • Could you explain what "behavior" is? Would it be some function of the four components; or would it be all four components together; or is it intended to be something you discover through exploring your data? – whuber Aug 20 '21 at 14:02
  • I measured the time cattle spent carrying out each behaviour over day and night periods. Initially behaviour IV was recorded in hours and minutes (expressed as a decimal), I then converted it to percentage, so for example, each cows 4 behaviours totalled 100% for each period. The behaviours are in separate columns (4 variables), however data can be rearranged so that a single variable (behaviour) is created so all 4 would be together (if that makes it easier), and the time (%) spent doing each would then make a second variable. – Kyle Aug 21 '21 at 10:10

0 Answers0