based on number of patients in each of the TS (Epilepsy Topology+ Stimulation) and SS nested datasets
based on inverse variance weighting: inverse of SS data variance (wilson CI or bootstrapped) and the credible interval for the posterior
I suspect if we use 3, this will heavily weigh it towards the posterior calculations, i.e. TS data. Given TS data is heavily mesial temporal biased, this could reintroduce the bias.
Given TS has more data than SS, option 2 could similarly reintroduce the bias heavily.
So then why even bother? because, e..g Somatosensory, is insular for TS but parietal for SS. We want to capture both.
So I suggest we take a simple mean of the two so that the TS-posterior and SS data are given equal weights.
could be:
I suspect if we use 3, this will heavily weigh it towards the posterior calculations, i.e. TS data. Given TS data is heavily mesial temporal biased, this could reintroduce the bias. Given TS has more data than SS, option 2 could similarly reintroduce the bias heavily.
So then why even bother? because, e..g Somatosensory, is insular for TS but parietal for SS. We want to capture both. So I suggest we take a simple mean of the two so that the TS-posterior and SS data are given equal weights.