### 1. Do I need to worry about the warning message “Stopping adaptation”?
No. Adaptation is not required but more for improving the MCMC chain behaviours for faster convergence and better mixing. 


### 2. Can zoib handle multiple random effects?
Yes, multiple random effects can be specified, such as the random intercept/random slope type with independence or unstructured variance/covariance matrix. An example is given below, where the slope associated with x is also random on top of a random intercept. 

data("BiRepeated", package = "zoib")
eg2 <- zoib(y1|y2 ~ x|1|x, data= BiRepeated, n.response=2,
            random=1, EUID= BiRepeated$id,
            zero.inflation = FALSE, one.inflation = FALSE,				
            prior.Sigma = "VC.unif", n.iter=500, n.thin=2, n.burn=100)

			
### 3. Can zoib handle nested random effects?
No, it cannot, at the current stage.


### 4. Can zoib  predict “Y values” for a new dataset that also includes the posterior distribution for that prediction?
This functionality is now available in version 1.5.0. Before that (version 1.4.3. or earlier)


### 5. How to obtain the DIC value of a ZOIB model?
dic.samples(x$MCMC.model, n.iter=xxxx,thin=xxx,...), where x is the output object from the zoib() function.


### 6. How to save a zoib model object  for later use in a new R session so that you don't have to initialize and burn in the chains again?
myzoib<- zoib(your model and MCMC specification)
myzoib$MCMC.model$recompile()
dic.samples(model=eg2.fixed$MCMC.model,n.iter=xx)
coda.samples(model=myzoib$MCMC.model, n.iter=xxx, variable.names ="parameter labels")


### 7. Does zoib allow for a grouping variable in which observations *are part of more than one group?*
There are at least two ways to achieve that. First, include the group variable as a fixed effect in the model, which would not be ideal if it has many levels and there is no particular meaning associated with the levels. Second, treat it as a random effect, and this will be your EUID (the smallest experimental unit is independent from each other), and then include batch as a random
effect as in
zoib(yield ~ temp | 1 | 1 | as.factor(batch),  joint = FALSE, random=1,EUID=GasolineYield$productid, zero.inflation = FALSE, one.inflation = FALSE)

If you encounter some errors in the second approach, there are two possible reasons: the unbalanced nature of your data or the package cannot handle this type of data structure at the current stage


### 8. Does zoib allow adding effects for more than one grouping variable (for example, would it understand 'EUID = c(GroupingFactor1, GroupingFactor2)')?
No. The EUID is more a specification of the experimental unit, and there should be 1 per data set, rather than being used for specs of random effects. Random effects are specified using  zoib( outcome~ xbeta | xsum | x0 |x1 | zrandom).


### 9. How does zoib handle data points with value NA?
zoib automatically removes cases with NA before modelling


### 10. Can the commands in R package coda such as traceplot to perform the MCMC chain behavior from zoib?
Yes, but need a bit "reformatting" on the posterior samples output from zoib. Say there are two chains. Suppose X is the object from zoib, then the following command can be applied. However, you can always use the functions paraplot() and check.psrf() in the zoib packages to perform the same task. Check out the manual on how to use the two functions.

sample<- X$coeff 
autocorr.plot(sample1[[1]][,]);  # for chain 1
autocorr.plot(sample1[[2]][,])  # for chain 2
traceplot(sample1[[1]][,]); # chain 1, similar for other chains
gelman.diag(mcmc.list(sample1[[1]][,],sample1[[2]][,])) $Gelman diagnostics


### 11.Can users specify priors for the parameters in the zoib model? 
The package currently offers 4 types of priors for the regression coefficients beta: dn, l1-type, l2-type and ard-type. The priors can be used in an informative way by specifying a larger precision parameter in prec.dn; however, the prior mean is still 0, and any prior covariance among parameters is not incorporated. The package does not accommodate user-specified priors other than those 4.


### 12. Is it possible to seed the MCMC to have reproducible results ? 
Yes, by specifying the seeds for the MCMC chains using the seeds() option in zoib.


### 13. What does the coefficient component  of the object output from zoib contain?
If there is no zero or one inflation, then it contains the regression coefficient from the regression equations 
g(mu) = xb*b, where g is the link function (logit, probit or cloglog), and mu is the mean of the Beta distribution
log(eta) = xd*d, where eta is the sum of the two shape parameters of the Beta distribution 

If there is zero  inflation, the regression coefficients from the following equation tags on
g(p0) = x0*b0, where g is the link function (logit, probit or cloglog), and p0 = Pr(y=0)

If there is zero  inflation, the regression coefficients from the following equation tags on
g(p1) = x1*b1, where g is the link function (logit, probit or cloglog), and p1 = Pr(y=1|y>0)

Each set of regression coefficient always starts with "intercept".