I want to test for relative global impact of absolute latitude and altitude on a variable that ranges from 0 to 1 (six possible values). I have thousands of observations sampled across the world. I harbour some doubts regarding doing that test:
Do I need to account for spatial dependence given that the data span the whole world? For example, should I add non-absolute latitude and longitude as covariates?
Also, the variable is ordinal and calculated numerically for 6 levels, so would it be ok to treat it as a numerical variable? any reference to back up a decision on this?
Thanks in advance
