Where mediation can hide disagreement
Genetic liability, symptoms, and executive function need not point in the same direction for every individual or subgroup. The publication title frames symptoms as mediating the relationship between depression genetic liability and executive function, but title-only metadata does not show how discordant patterns were handled. A useful distinction is whether symptom measures add an independently informative layer or mainly absorb variation that conflicts with the genetic association. That could be examined by reporting the direct genetic liability association before and after mediation, alongside cases or strata where symptoms and executive function diverge. Missing symptom measurements also matter: complete-case analysis, imputation, and an explicit missingness model can produce different apparent levels of agreement. The unresolved question is whether the mediation analysis preserves those disagreements as information or compresses them into a cleaner pathway.
