06-07-2017 05:26 AM
I am running a nested model with four levels: juveniles, dad, mom, and year.
I have juveniles that are nested under dad, dad is nested under mom, and mom is nested under year. The variable “ontsurv” is the binary response variable, referring to whether a juvenile survived or not (I am using lizards in the analysis).
I wrote the following syntax for my model:
proc glimmix method = quad(fastquad qpoints=6) data=final; parms (3, 1, 19);
class mom dad year;
model ontsurv (event = '1')= /S dist = binary link = logit ddfm =bw;
random int/ subject = year;
random int/ subject = mom(year);
random int/ subject = dad(mom year);
covtest 'var (year)= 0' 0 / est parms ;
covtest 'var (mom(year)) = 0' . 0 / est parms;
covtest 'var (dad(mom year)) = 0' . . 0/ est parms;
I am not sure if this model implies that var (year) represents the random intercept variance of the year level , var mom (year) the variance of the categorical variable mom, and dad (mom year) the variance of dad.
I would really appreciate some help with this syntax.