[R-sig-ME] glmmTMB dispformula models

Djenifer Kappel K@ppe|D @end|ng |rom c@rd|||@@c@uk
Mon Jun 20 11:54:53 CEST 2022


Hi all,

I am using glmmTMB to model the effects of genetic factors on longitudinal data of blood tests.
We are interested in the effect of those factors on the mean blood test outcome (using a gamma distribution), but also if they affect the variance of the outcome (dispersion).
This is what I am testing at the moment:
  model <- glmmTMB(outcome ~ gene_sets + covar1+ covar2+covar3+ (1|Individual),
                    dispformula = ~gene_sets,
                    data = mydata,
                    family = Gamma(link=log),
                    na.action = na.exclude,
                    control = glmmTMBControl(profile = TRUE))

After reading the glmmTMB vignette it is not clear to me what variance is being modelled when we specify a dispformula.
Is this the residual variance of the model on the full dataset or the intra-individual variability (on the repeated measures).
I supposed it is the latter given that we cannot specify a random effect on dispformula, but I would appreciate if anyone could help me understand this.

Many thanks!
Djenifer

--

Djenifer Kappel, Ph.D
Research Associate

MRC Centre for Neuropsychiatric Genetics and Genomics
Room 2.01 Desk 36
Cardiff University School of Medicine
Hadyn Ellis Building
Maindy Road
Cardiff
CF24 4HQ

Email: KappelD using cardiff.ac.uk<mailto:KappelD using cardiff.ac.uk>

Please note, replies are not expected outside of your working hours.
The University welcomes correspondence in Welsh or English. Corresponding in Welsh will not lead to any delay.
Djenifer Kappel, Ph.D
Cydymaith Ymchwil

Canolfan y Cyngor Ymchwil Feddygol ar gyfer Geneteg Niwroseiciatrig a Genomeg
Ystafell 2.01 - 36
Ysgol Meddygaeth Prifysgol Caerdydd
Adeliad Hadyn Ellis
Heol Maindy
Caerdydd
CF24 4HQ

Ebost: KappelD using cardiff.ac.uk<mailto:KappelD using cardiff.ac.uk>

Noder, ni ddisgwylir atebion y tu allan i'ch oriau gwaith arferol.
Mae'r Brifysgol yn croesawu gohebiaeth yn Gymraeg neu yn Saesneg.
Ni fydd gohebu yn Gymraeg yn creu unrhyw oedi.



	[[alternative HTML version deleted]]



More information about the R-sig-mixed-models mailing list