Closed
Description
Summary
One of the diagnostic methods for coordinate-based meta-analyses, as described in Muller et al. (2018), is to perform a jackknife analysis on a CBMA, to see how much each experiment contributes to each resulting cluster.
Additional details
This stems from #34 (comment).
Approach ideas
This should be a Transformer(?) that takes in the MetaResult, loops through experiments in the Dataset, and calculates the summary statistic (e.g., ALE) map using all but one of the experiments, and relates the difference between the full Dataset's statistic and the statistic from the Dataset minus that one experiment.
I'm not entirely sure what the class should look like, but it will have access to the Estimator and Dataset from the MetaResult input.