this is for holding javascript data
- Methods: In formula (6), and with the assumption of independence, of scaling factor and unobserved difference between groups (high and low disease group), we lose more information. It is understood that this assumption is necessary to calculate the unconditional variance of the disease effect estimate at each site, but, if we remove the independence assumption (we don't drop the covariance) and we use an approximate instead (eg. numeric approximation), we lose less information although the calculation will be more complex.
