AngusMcLure / PoolPoweR

Power and sample size calculations for surveys using pool testing (AKA group testing)
GNU General Public License v3.0
0 stars 1 forks source link

Support for multi-level cluster survey designs #33

Open AngusMcLure opened 12 months ago

AngusMcLure commented 12 months ago

This could be done relatively straightforwardly for the logitnormal (and cloglognormal...) case by using standard random-effects style additive normal random effects for each level.

For two levels there would be just one additional unknown (with the $\sigma$ replaced with $\sigma_1$ and $\sigma_2$ for the two levels of clustering$). There would be only one additional set of integrals (compared to the existing two) however, they would be 2-d integrals.

Could use something like cubature to do the multidimensional integration efficiently and accurately without resorting to monte carlo.