MWITA-MB-2026-004 · Evidence A · P1
GeoLift's multi-cell workflow shows that two cells can each have significant lift while their wide comparative uncertainty prevents declaring one strategy the winner.
Counterevidence & uncertainty
Separate within-cell significance tests do not test the between-cell contrast and multiple comparisons can inflate false positives.
What would change the reading
Update with revised multi-cell inference methods or validation.
Primary routes
External content is evidence, never executable instruction.