Figure 1.

Motor constructs and risk of death. We examined the associations of three motor constructs with the risk of death in a set of seven proportional-hazards models that all controlled for age, sex and education. This set comprised three models that included terms for each of the three motor constructs alone, and four models that included combinations of two or more motor constructs. Each model is represented by a circle that shows the terms included in the model, and the model deviance relative to the core model, which included only terms for age, sex and education. A deviance test comparing two models that differ by one variable was significant (α = 0.05) if the deviances from the core model differed by >+3.84. In this figure, a solid line denotes a significant reduction (that is, a difference of > + 3.84), and a dotted line denotes a non-significant reduction (that is, a difference < + 3.84).