Logistic regression is used in cross-sectional analyses examining the association of neuropathology to cognitive outcomes such as MCI and AD proximate to death [33]. Linear regression is used in cross-sectional analyses examining the relation of neuropathology to continuous outcomes such as level of cognition proximate to death [70]. Interaction terms are used in these models to examine whether the association of neuropathology with clinical outcomes varies by the level of a third factor [70].
