Currently, we accumulate normalised values of each statistic per block, and then use the the block sizes for all stats to perform weighted averages across blocks in the Jackknife calculation. We might get slightly more accurate results by accumulating unnormalised stats values for each block, and then correctly normalising for each drop-one jackknife estimate.
stschiff
commented
2 years ago
Currently, we accumulate normalised values of each statistic per block, and then use the the block sizes for all stats to perform weighted averages across blocks in the Jackknife calculation. We might get slightly more accurate results by accumulating unnormalised stats values for each block, and then correctly normalising for each drop-one jackknife estimate.