James Campbell

Excellent -- thank you for improving the test, and very excited to have this merged!

@mdhaber thanks for looking into it more -- I'm with you in that I don't know whether there are deeper reasons that I'm not seeing, and your implementation is certainly...

@mdhaber Thanks for continuing to push on this. I'd be happy to reach out to Art Owen and/or participate in that discussion. It's definitely a subtle issue. A couple synthesis...

I don't think simulations would provide a justification, but in this case I think the net result is probably that it's best to close the PR, at least based on...

I was broadly observing that in the `min` and `max` examples, the intervals are smaller (hence less overlap and so lower peaks in the histogram), and the patched version lands...

Thank you for confirming that no other change had been merged. I think I actually made a mistake in my environment and wasn't correctly comparing the patch. I looked again...

No -- I wasn't fixing the seed. Thanks for pointing that out -- that does make the result much more clear, and in favor of the change.  (re: the...

I'd love to send over the code for you to review. I'm traveling this week, so it won't be until next week!

@mdhaber -- here's the replication code, as well as a patch for _resampling.py corresponding to allowing the `bca_patch` method. ``` # %% import numpy as np from scipy import stats...

I'm going to merge this as-is, but also mark this action as explicitly no longer maintained. With changes to the underlying GX that have happened in the past couple of...