[R-sig-ME] Conceptual doubts on how to define my random effects

Mario Garrido g@@dio @ending from po@t@bgu@@c@il
Tue Oct 9 11:45:16 CEST 2018


Dear lme4users,
I have some conceptual doubts on understanding the meaning of some random
factors.

Briefly, I am checking the dynamics of infection by bacteria on individuals
from 3 different species of hosts. I am trying to fit my data to a
generalized linear model with repeated measures. My data consists on the amount
of bacterial colonies  (Myc.qPCR) at different time points (day) fin the
blood of the 33 individual hosts (exp.ID), belonging to 3 different
species  (sp). At time zero all the values are 0 cause the indivuals were
still not infected.
As I want to check whether the dynamics in the three species are different
I included in my first model day*sp as fixed factor

I am trying to fit the random effect but I have some problems with the
interpretation. Any reading recommendations for Ecologists that we are not
experts in statistics to understand the ecological meaning of random
effects?

I know for use that I want to control  for variance at individual level so,
|exp.ID is needed, I also want to 'force the intercept to be zero (since at
day zero amount of bacteria is zero) , and I also wonder whether I have to
nest the individuals within day.
Thus, my model would be



lme(lg(Myc.qPCR)~day*sp,random=?,method="ML")

And I have doubts on how to define my random effect, cause, indeed, I do
not know what is the exact meaning of them under lme4. I understand that as
I have a Mixed-effects model with temporal pseudoreplication I hould use
day in the random factor, but how to include that intercept is zero?
(1|exp.ID)
(day|exp.ID)
(0+day|exp.ID)
+(1|exp.ID)+(0+day|exp.ID)

Thanks!

-- 
Mario Garrido Escudero, PhD
Dr. Hadas Hawlena Lab
Mitrani Department of Desert Ecology
Jacob Blaustein Institutes for Desert Research
Ben-Gurion University of the Negev
Midreshet Ben-Gurion 84990 ISRAEL

gaiarrido using gmail.com; gaadio using post.bgu.ac.il
phone: (+972) 08-659-6854

	[[alternative HTML version deleted]]



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