I have a code in OpenBUGS but the error is "variable CR is not defined"

Question

model
{
for( i in 1 : N ) {   
dgf[i] ~ dbin(p[i],n[i])    
logit(p[i]) <- a[subject[i]] + beta[1] * CR[i] 
}

for (j in 1:94)
{
a[j]~dnorm(beta0,prec.tau)
}
beta[1] ~ dnorm(0.0,.000001)
beta0 ~ dnorm(0.0,.000001)
prec.tau ~ dgamma(0.001,.001)
tau<-sqrt(1/prec.tau)
}

list( n=c(1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1), dgf=c(0,0,0,0,0,0,0,0,0,0,1,1,0,0,0,0,0,0,0,1,1,1,0,0,1,1,0,0,0,1,1,1,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,1,0,1,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,1,1,0,0,1,1,0,0,1,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,0,0,0,0,0,1,0,0,0,0,0,1,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,1,1,0,0,1,1,0,0,0,0,1,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0), subject=c(1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63,64,64,65,65,66,66,67,67,68,68,69,69,70,70,71,71,72,72,73,73,74,74,75,75,76,76,77,77,78,78,79,79,80,80,81,81,82,82,83,83,84,84,85,85,86,86,87,87,88,88,89,89,90,90,91,91,92,92,93,93,94,94), CR=c(NA,NA,1.41,0.85,1.13,0.65,NA,NA,2.13,1.61,7.31,3.8,1.65,2.32,1.13,2.3,0.99,1.5,1.32,3.95,7.2,2.97,0.83,1.55,NA,6.5,0.89,1.2,1.52,7,8.68,7.41,NA,0.86,NA,1.92,NA,1.31,7.8,1.78,NA,1.67,NA,NA,NA,NA,NA,2.25,0.98,0.82,3.94,1.14,12,2.58,2.42,2.59,NA,NA,NA,NA,6.6,3.22,NA,2.02,2.43,1.96,0.82,1.64,1.81,1.53,1.01,5.21,8.33,1.14,1.49,6,5.6,2,3.33,4.08,NA,NA,1.14,1.25,0.85,5.42,0.85,0.65,1.02,1.33,1.1,1.12,NA,NA,1.53,1.76,2,0.85,2.9,5,4.09,2.68,0.98,1.48,0.66,0.57,5.72,2.34,0.93,2.39,1.39,1.44,4.77,2.39,1.79,1.2,0.81,1.25,4.69,1.22,1.92,1.48,2.46,NA,NA,2.53,1.12,1.74,3.45,1.22,1.27,2.61,1.75,0.82,NA,1.4,NA,5.1,1.24,1.5,1.94,1.24,1.04,1.24,NA,2.39,NA,2.07,2.19,1.6,6,6.38,1.17,1.2,5.62,6.39,1.82,1.31,NA,1.18,3.71,2.03,5.4,2.17,NA,1.94,1.57,1.44,1.35,1.63,1.24,1.54,1.5,NA,NA,NA,NA,1.44,NA,2.19,7.98,2.15,1.71,1.45,NA,0.98,2.37,1.58), N = 188)

I know that the error is because of "NA" in variable "CR" but i don't know how to solve it. i'll appreciate any help.

score 2 · Answer 1 · answered Mar 26 '17 at 11:21

You have 2 options:

1) Remove the missing CR and the corresponding elements of n, dgf, and subject (and reduce N accordingly).

2) Define a stochastic relation for CR within your model, so that the model estimates the missing CR and uses these estimates in the logistic regression. Something like:

for(i in 1:N){
   CR[i] ~ dnorm(CR_mu, CR_tau)
}
CR_mu ~ dnorm(0, 10^-6)
CR_tau ~ drama(0.001, 0.001)

CR_mu and CR_tau are probably not of interest but can be monitored if you want.

Note that both approaches assume that CR are missing at random (and not e.g. censored) - if CR are missing not at random this will give you biased results.

Matt

I have a code in OpenBUGS but the error is "variable CR is not defined"

1 Answers1