|
|
Hi everyone!! I am trying to analyze if there are significant differences between two sites in the concentration of a number of nutrients in soils. I am doing this individually for both for two types of soil: organic and mineral. So I am running 2 Manova tests (one for organic and one for mineral) to test the effect of one IV (with two levels: Site 1 and Site 2) on more than two DV's (the mean concentration of nutrients). I am getting the problem that in one of the tests i am getting a non significant Wilks lambda, however when you inspect the between-subject effects there is a significant difference between both sites in terms of one of the nutrients. Why is this??? I thought that if Wilks was non-significant differences between sites were not supposed to be detected for any of the DV. Then, is it still valid to report these significant differences even though the multivariate test (wilks) was not significant? I have been thinking of running individual independent t-tests (considering my IV has only two levels) for each of my DV's, but I fear of increasing my chance of Type I error. Your advice is very welcome!
|
Locked
