ARMA(1,1) as part of hierarchical model term

Modeling
15

The time series thing looks good.

If that sig2v is a variance add a sqrt:

v[i] ~ normal(0, sqrt(sig2v));

If you parameterize it in terms of standard deviation (sigv or something), then you can easily switch between centered and non-centered parameterization using:

vector[N] v; // centered

And

vector<multiplier = sigv>[N] v; // non-centered

(Updated non-centered parametrization with offset and multiplier - #3 by daniel_h)