model
{
for (i in 1:Nobservations)
	{
	#Outcome model (repeated observations of FFQ)
	logFFQ1[i]~dnorm(meanlogFFQ[i],taueps)
	logFFQ2[i]~dnorm(meanlogFFQ[i],taueps)

	#Model for mean of log FFQ
	meanlogFFQ[i]~dnorm(meanmeanlogFFQ[i],taur)
	
	#Define the fixed effects part of the mean FFQ
	meanmeanlogFFQ[i]<-beta[1]+beta[2]*X[i]

	
	#Biomarker model for log protein	
	logprotein1[i]~dnorm(X[i],tauu)
	logprotein2[i]~dnorm(X[i],tauu)
	
	X[i]~dnorm(meanX[i],taux)
	meanX[i]<-alpha[1]+alpha[2]*AGE[i]+alpha[3]*BMI[i]
	}

#Define lambda (a noninformative prior is assigned)	
	tauu<-lambda*taux/(1-lambda)
	
#Noninformative priors on the model parameters
lambda~dunif(0,1)
taueps~dgamma(3,0.1)
taux~dgamma(3,0.1)
taur~dgamma(3,0.1)

#Define the signal attenuation 
attenuation<-beta[2]/(pow(beta[2],2)+taux/taur+taux/taueps)

#Priors for the fixed effects
for (i in 1:nalphas){alpha[i]~dnorm(0,1.0E-6)}
for (i in 1:nbetas){beta[i]~dnorm(0,1.0E-6)}

#Deterministic transformations (obtain variances)
sigma2eps<-1/taueps
sigma2x<-1/taux
sigma2u<-1/tauu
sigma2r<-1/taur
	
}