Dear readers,
See image: The length of the first axis is obviously >4, suggesting I should use CCA. But it's all because of the outlier 5_2.
Is that sufficient justification to use CCA?
For some reason it feels like I should apply some bias to the data set and go for an RDA instead (just by wondering the question "what if location 5_2 (which is the second of 6 habitats in location 5) wasn't in the data set.")
What do you guys think? How should I go about this?
Thanks!
Hello, I have the same problem. I have I calculated Variance Inflation Factors (VIF) and all predictor variables had >10 VIF. I wanted to perform CCA to relate environmental variables to small mammals abundance. Thinking of using PCA to rectify for multi collinearity. Advise please
You must keep all your data and not apply bias. Transparency is important. Also, don't be the 'what if' guy lol
Andrew Paul McKenzie Pegman Ok, thanks.
I get what you mean with the "what if" guy, but I don't see how it hurts to wonder.
- Badges
- Science topic