@_jcken all of my recent examples have been deliberately for the case of partially overlapping samples (some independent / some paired) - a classic example here is performance in two optional modules.
Question: do students get the same average mark; but you have some that do both modules, some that only do module A, some that only do module B.