[R] inconsistencies between HLM and clmm results for the signs for the coefficients

Debbie Hahs-Vaughn debb|e @end|ng |rom uc|@edu
Thu Oct 20 03:51:57 CEST 2022

I've generated a very simple multilevel ordinal model using HLM software and using the clmm function from the ordinal package in R. The threshold values and coefficients are comparable, however the signs for the coefficients (and only the coefficients; not the thresholds) are opposite (e.g., positive in HLM, negative in clmm).  This then impacts the odds ratios for the predictors.  HLM reports odds greater than one and clmm reports odds less than one.  I anticipate there is a very easy reason for what's happening, and it's just escaping me.  Can someone please advise on what I've overlooked?

	[[alternative HTML version deleted]]

More information about the R-help mailing list