高手们,有个问题想请教:对于无信息先验分布(non - informative prior)是怎么定义的?我看了一些材料,多处都是按照以下方式定义的。但是对于无信息先验分布,方差不是应该足够大吗(In case of non-informative priors, the variances should have been specified as very large or instead uniform distributions covering the entire range of plausible values should have been used. )?能否帮我解释下?Thanks.model
{
for (i in 1:N)
{
pred<-a*pow(d,b)
h~dnorm(pred,prec)
}
# Priors
a ~ dnorm(0,1.0E-6) # Is non-informative prior for a?
b ~ dnorm(0,1.0E-6)
prec~dgamma(0.001,0.001)
}