Dear Tali,

There is a simple explanation for this.  Family-level BMS uses a sampling method (MCMC), therefore running the procedure several times may result in slightly different results.  The more samples you use, the less likely it will be that there are differences across analyses.  You could change this manually by editing the function spm_compare_families which uses a variable called Nsamp.  The higher you set this value, the more consistent the results will be, but it will also take longer to run. 

Will and Maria: I noted that the default value for Nsamp is currently 1e4.  It may be worth considering choosing a larger default value? 

All the best,
Klaas