The name is absent

college attendance duration can end either because the individual drops out of college or
because they graduate from college. Thus, for the second duration we take a competing
risks approach (see Katz, 1986, Hausman and Han, 1990; Anderson, 1992; Sueyoshi,
1992; and McCall, 1996, for example).

Let T^f be the waiting time between high school graduation and college enrollment.
Let T^g be the duration of college attendance until graduation (graduation duration) and let
T^d be the duration of college attendance until dropout (dropout duration). Further, let T^s
= min(T^g, T^d).

To incorporate the waiting duration into the statistical model, the survival
function conditional on the unobserved variable θ^f and the vector of observed predictor
variables z is assumed to have the following form

S^f (k^f∖z,θ^f ) = P(T^f > k^f∖z,θ^f )

= exp[-θ^f ∑ exp(γ_r^f + (β_r^f )′z)]

r=1

where the parameters γ^f_r are the baseline hazard parameters and the vector β^f_r measures the

(possibly time-varying) effects of the regressors on waiting time until college enrollment,

<<   9   10   11   12   13   14   15   >>

More intriguing information