I try to compare the splines for my 6 categories. To this end, first I tried the ordered factor approach and posted the result here. This comparison is not sufficient for me since I would like to have pairwise comparisons.

  1. To this end, first, by keeping the ordered factor approach, I thought that by changing the reference level, I can have the pairwise comparisons. I did this by changing the category names since I cannot relevel an ordered factor. Normally, when F is the reference level, if there is a significant difference between F and G5 (G5 as an example), I expect to have this significant difference also between G5 and F when G5 is the reference level. But it was not the case. I don't understand why.

  2. Then I follow this blog post and create pairwise comparisons. When I did the ordered factor approach, the differences were significant for F-G1P1 and F-G5P1 (See here). Using the same model, I create the pairwise comparisons and my output is the following: enter image description here

The method says "Where the confidence interval excludes zero, we might infer significant differences between a pari of estimated smooths." This is not the case for my smooths. Why I don't get the same significant results as in my previous analysis where I use the ordered factor approach, F being the reference level?

Thank you in advance.



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.