Replies: 1 comment 2 replies
-
Thanks @wlandau for the question! Good one, so far we did not play around with custom contrasts too much as far as I am aware. @clarkliming any thoughts? |
Beta Was this translation helpful? Give feedback.
2 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
When
AVISIT
is an ordered factor and I manually set contrasts, my custom contrasts seem to be ignored:Created on 2024-07-09 with reprex v2.1.0
Session info
I would like to handle time as an ordered factor because the ordering of the levels is important for ARMA correlation structures. Unfortunately, R uses
contr.poly()
on ordered factors by default, which is usually counterintuitive and often statistically inappropriate. Would it be reasonable formmrm
to respect user-defined contrasts so users like me can work around the strange defaults in base R?Beta Was this translation helpful? Give feedback.
All reactions