Fitting cenexponential family in MCMCglmm
Dear sirs, I am estimating the heritability of survival of Malpura sheep from birth to weaning (90 days), three to six months, six to twelve months and birth to twelve months for a twenty year data set. The data has a right censored exponential data as a response variable. The data have following common structure: $ itag : int 2890 2891 2892 2893 2894 2899 2901 2902 2903 2905 ... $ stag : int 2602 2258 2248 2602 2602 2599 2667 2639 2667 2599 ... $ dtag : int 2422 2720 2199 2483 2659 2638 1807 2488 1823 2082 ... $ season : chr "Winter" "Winter" "Winter" "Winter" ... $ sex : chr "Female" "Female" "Male" "Male" ... $ tyb : chr "Single" "Single" "Single" "Single" ... $ ewt : num 26 33.4 31.4 34 29.8 29 28 28.4 34 28.5 ... $ bwt : num 2.6 3.2 3 3.8 2.8 3.4 2.9 2.4 3.2 3 ... $ dtd : chr "2670" "1966" "444" "1442" ... $ cens : int 0 0 0 0 0 0 0 0 0 0 ... ?itag?, ?stag? and ?dtag? are the identities of individuals, sire and dam, respectively. ?Season?, ?sex?, ?tyb? (type of birth), ?ewt? (ewe weight at the time of lambing), and ?bwt? (birthweight) are the covariates for the model. ?dtd? (response variable) is the days to survival for the specific duration of the study, ranging from 1 to 90 days for birth to weaning data. ?cens? is the censoring code. For birth to weaning data, the lambs alive at the end of weaning (90 days) are recorded as ?right censored?, and time of failure of such lambs is taken as 90 days. Moreover, the lambs which are removed from the flock before weaning due to reasons other than death (e.g., culling) are also recorded as ?right censored? (in Cox regression). In mailing lists on MCMCglmm, I have found the following information: m1<-MCMCglmm (cbind (ymin, ymax) ~1, family="cenexponential", data=dat) (https://stat.ethz.ch/pipermail/r-sig-mixed-models/2010q4/004933.html) I am taking example of birth to weaning data. I have doubts on the following aspects: (1) Will the ?ymax? be equal to 90 days (days to weaning) or equal to the maximum number of days an animal survives even after weaning? What about the information about animals which are still in the flock at the time of data collection? The data on such animals are right censored in Cox regression. What to do with data on such animals while estimating heritability by using MCMCglmm? Will ?dtd? contain data (days to survival) about every animal including the animals which are still in the flock at the time of data collection or for only the animals which died before the end of weaning? (2) What about the data about the lambs which are removed from the flock before weaning due to reasons other than death (e.g., culling)? (3) How to set up data (?dtd?) for left censored and left truncated information? (4) What to do about censoring code. I could not find any information about this anywhere for MCMCglmm. Anticipating response at the earliest. With kind regards.
Dr. Indrasen Chauhan Scientist, ICAR-IVRI,Mukteshwar Nainital,Uttarakhand-263138 (INDIA) Mobile:9461883473,8910856302 [[alternative HTML version deleted]]