[R-sig-ME] glmer() Gamma distribution - constant coefficient of variation
hedyeh@h @end|ng |rom u@c@edu
Mon Apr 5 17:25:59 CEST 2021
This email is from an old chain which was giving me issues due to large email size, so I am starting over here to answer some of the questions + add some explanations.
I am running a GLMM with Gamma distribution and identity link which has a constant coefficient of variation (i.e. CV) assumption.
Answer to the previous question on why continuous outcome and not a negative binomial for example: We have eight different outcomes that we are interested at, they are skewed composite scores of different medical diagnostics, and they can go anywhere between 0 to 26+ so we decided to use these variables as continuous outcome.
My question: Previously it was suggested to look at the plot on the right to check the constant CV assumption. So, would you trust the following plots as passing the constant CV assumption? I am leaning toward giving it a pass since although the line on the plot looks like it has a positive slope, the magnitude of y-axis is small.
Here is the link to the plot: https://drive.google.com/file/d/1Sta8yyq8dhnewMSLtvnH7XXzR8i1-cSk/view?usp=sharing
Hedyeh Ahmadi, Ph.D.
Keck School of Medicine
Department of Preventive Medicine
University of Southern California
Institute for Interdisciplinary Salivary Bioscience Research (IISBR)
University of California, Irvine
[[alternative HTML version deleted]]
More information about the R-sig-mixed-models