Mixed outcomes

Abstract

Measurements of both continuous and discrete outcomes are encountered in many statistical problems. A common example is a developmental toxicity study, which typically involves exposing timepregnant animals (rats, mice) to an environmental agent during major organogenesis. The dams are sacrificed prior to normal delivery, at which time the uterus is removed and thoroughly examined. Among viable fetuses, the incidence of any malformation (binary data) and reductions in fetal weight (continuous data) are typically of primary concern, since both have been found to be sensitive indicators of a toxic effect. This motivates the formulation of a joint distribution with mixed continuous and discrete outcomes.