Most Different Systems Design

They are mirror images. The most similar systems design selects cases alike on background factors but differing on the outcome, isolating the differing factor as the cause (Mill's method of difference). MDSD selects cases differing on most background factors but sharing the outcome, so the factor they nonetheless have in common is the candidate cause (Mill's method of agreement). MSSD controls by similarity to explain divergence; MDSD uses diversity to find a robust common cause.

Equifinality is the possibility that the same outcome arises through different causal pathways in different cases. MDSD assumes that a shared outcome among diverse cases points to a single common cause. If, instead, each diverse case reaches the outcome by its own distinct mechanism, there may be no genuine common cause, and the design's core inference collapses. This is why analysts use process tracing to check that the identified commonality actually operates the same way across cases.

MDSD typically chooses diverse cases that all display the outcome, which lets it find what they share but gives no information about cases lacking the outcome. Without negative cases, the design cannot show that the common factor is also absent where the outcome is absent, leaving open whether the factor is truly necessary or sufficient. Incorporating contrasting cases or combining MDSD with QCA addresses this asymmetry.