# What's wrong with my BANOVA (Bayesian ANOVA)?

#### Tokyo

##### New Member
Hi everyone... I'm new to Bayesian statistics and I'm trying to perform a BANOVA. I'm using the following book to help me: "Doing Bayesian Data Analysis A Tutorial Introduction with R and BUGS", written by John K. Kruschke.

I'm working on a simulated data set, trying to find a difference in plant size (Tamanho, in Portuguese) in different regions in Brazil (Região, in Portuguese).

Here is my prior:

modelo="
model{
for(i in 1:N){
y~dnorm(mu,tau[x])
mu<-a0+a[x]
}

sigma~dunif(0,20)
a0~dnorm(0,0.001)

for(j in 1:Lvs){
a[j]~dnorm(0.0,atau)
tau[j]~dgamma(sG,rG)
}
atau<-1/pow(aSD,2)
sG<-pow(m,2)/pow(d,2)
rG<-m/pow(d,2)
m~dgamma(1,0.222)
d~dgamma(1,0.222)
aSD<-abs(aSDunabs)+0.1
aSDunabs~dt(0,0.001,2)
}
"
...

here is my chains initialization command:

if(F){
modelGenInits()
}else{
a0=mean(Tamanho)
a=aggregate(Tamanho,list(Região),mean)[,2]-a0
tau=1/(aggregate(Tamanho,list(Região),sd)[,2])^2

genInitList<-function(){
return(
list(
a0=a0,
a=a,
tau=tau,
m=mean(tau),
d=sd(tau),
aSDunabs=sd(a)
)
)
}
}

However, when I try to initialize my model, I keep on receiving this message:
modelInits(bugsInits(genInitList))
Initializing chain 1:
initial values loaded but chain contain uninitialized variables

I am aware that I should provide initial values for my model, but I believe they are there in my initialization command! I'm sure I'm following all the recommendations in chapter 18 of Kruschke's book!

Could someone tell me where I went wrong?

Thanks a lot