Finalise the MvNormal model

[drbenvincent]

(m,c) should be drawn from a multivariate normal distribution

• [x] finish JAGS model (for separate models)
• [ ] think about reasonable priors

updated goals:

• [ ] resolve error in mixed model first, before...
• [ ] resolving error in fully hierarchical model. This may in fact automatically fixed if the mixed models get fixed.

[drbenvincent]

At this point we have...

Separate model: This works, and I'm happy with the setting up of the variance/covariance matrix. The resulting samples seem to be good.

Hierarchical model: This is NOT working, and is NOT a full hierarchical model. We are doing hierarchical inference on alpha and epsilon, but not on (m,c). We are in the middle of attempting to do hierarchical inference on r (correlation coefficient between m and c) but this is currently not working. Maybe we don't need to do this.

[drbenvincent]

Hierarchical model Attempting to put hyper priors on m and c and just have one single r for all participants. But the model crashes. Not sure why

[drbenvincent]

So the MixedModel doesn't work either, so there is some other error than needs to be resolved here before addressing the fully hierarchical model.