Lately, M-estimation has been my go-to tool for analyses. It's just such a flexible tool and makes variance estimation super easy
I did a talk a few weeks back, the recording is available at
https://lshtm.cloud.panopto.eu/Panopto/Pages/Viewer.aspx?id=9068252c-3aa1-4a26-a9a0-af2d00d802b1
with slides and code here
